su.class.cs244a

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Welcome the CS244a Newsgroup!
Date: Tue, 08 Jan 2002 17:06:49 -0800
Organization: Stanford University
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: dnab4046e1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010538455 22131 171.64.70.225 (9 Jan 2002 01:07:35 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.73 [en]C-{C-UDP; EBM-SONY1}  (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2365


You are welcome to post questions related to CS244a
to this newsgroup. In fact, we prefer that you post
questions here instead of sending email.

Each day, a TA is on duty and will answer questions posted
to this group. You can find out who is on duty each day
by looking at the class web page.

- NM
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine36.Stanford.EDU!anhoe
From: An-hoe Shih 
Newsgroups: su.class.cs244a
Subject: regarding echodemo in programming assignment #1
Date: Tue, 8 Jan 2002 17:11:15 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine36.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2366

Hi Professor and TAs,

   On the first page of the programming assignment #1 sheet, it asks us to
debug echodemo.c. However, I am not clear on how and what to turn in for
this part of the assignment. Can you clarify?

Thank you,

An-Hoe

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jing Jiang" 
Newsgroups: su.class.cs244a
Subject: alternative final?
Date: Wed, 9 Jan 2002 03:06:21 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: jjiang.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2314.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
Xref: nntp.stanford.edu su.class.cs244a:2367

Will there be an alternative final exam for this class? Thanks!

Jing


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: regarding echodemo in programming assignment #1
Date: Wed, 9 Jan 2002 11:43:57 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2368

Hi,

You need to write a brief report on the types of bugs you found in the
program and how to fix the bugs. Your deliverables are only the brief
report as part of your write-up (README) on the actual homework problem.

Best,

Xin

On Tue, 8 Jan 2002, An-hoe Shih wrote:

> Hi Professor and TAs,
> 
>    On the first page of the programming assignment #1 sheet, it asks us to
> debug echodemo.c. However, I am not clear on how and what to turn in for
> this part of the assignment. Can you clarify?
> 
> Thank you,
> 
> An-Hoe
> 
> 


-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: alternative final?
Date: Wed, 09 Jan 2002 14:25:06 -0800
Lines: 1
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010615865 2892 171.66.211.98 (9 Jan 2002 22:37:45 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2369

There are no plans for an alternative final.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Using purify
Date: Wed, 9 Jan 2002 20:33:27 -0800
Lines: 25
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine40.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2370


Hi all,

I am having some trouble using purify. Could you please help.

> I have copied the Makefile and the echodemo.c file into a directory and
> once I compile the code and run the following command ..i get an error:
>
>  purify -cache-dir=/tmp gcc -o  echodemo.purify e
> chodemo.o
> In order for Rational PurifyPlus to work properly, you must source
> the appropriate shell resource file in /usr/pubsw/etc/rational. See
> the release announcement at
>     http://ls.stanford.edu/services/pubsw/package/licensed/rational.html
> for more details.
>
> Could you please tell me if there is a problem with the license or if I
am
> not using the tool correctly.
>
> Would appreciate your help.
>
> Thanks,
> Piryanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Using purify
Date: Wed, 9 Jan 2002 20:39:29 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Priyanka Gupta 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2371

Piryanka,

  You need to source the csh script in /usr/pubsw/etc/rational

  >source /usr/pubsw/etc/rational/purifyplus_setup.csh

  should do the trick if you are using csh/tcsh. I've added the
line to my .cshrc so I don't have to bother sourcing it by hand each time
I log on.

                HTH,
                ~m


>
> Hi all,
>
> I am having some trouble using purify. Could you please help.
>
> > I have copied the Makefile and the echodemo.c file into a directory and
> > once I compile the code and run the following command ..i get an error:
> >
> >  purify -cache-dir=/tmp gcc -o  echodemo.purify e
> > chodemo.o
> > In order for Rational PurifyPlus to work properly, you must source
> > the appropriate shell resource file in /usr/pubsw/etc/rational. See
> > the release announcement at
> >     http://ls.stanford.edu/services/pubsw/package/licensed/rational.html
> > for more details.
> >
> > Could you please tell me if there is a problem with the license or if I
> am
> > not using the tool correctly.
> >
> > Would appreciate your help.
> >
> > Thanks,
> > Piryanka
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!spevack.Stanford.EDU!spevack
From: Max Spevack 
Newsgroups: su.class.cs244a
Subject: Trouble signing up for section
Date: Thu, 10 Jan 2002 05:07:56 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: spevack.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2372

Hi,

I tried to sign up for the class, but I received an error message:

You don't have permission to access /cgi-bin/cs244a/assign_TA_main.pl on
this server.

Has anyone else had difficulty, or is my build of Mozilla just really bad?
:-)

--Max

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Romain Thibaux 
Newsgroups: su.class.cs244a
Subject: Re: Trouble signing up for section
Date: Thu, 10 Jan 2002 07:28:50 -0800
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: thibaux.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010676798 9978 128.12.191.77 (10 Jan 2002 15:33:18 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.4.2-2 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2373

Same for me

    Romain

Max Spevack wrote:

> Hi,
>
> I tried to sign up for the class, but I received an error message:
>
> You don't have permission to access /cgi-bin/cs244a/assign_TA_main.pl on
> this server.
>
> Has anyone else had difficulty, or is my build of Mozilla just really bad?
> :-)
>
> --Max

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine27.Stanford.EDU!anhoe
From: An-hoe Shih 
Newsgroups: su.class.cs244a
Subject: turn in the problem set via SCPD courier service
Date: Thu, 10 Jan 2002 16:33:42 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2374

Dear Professor and TAs,

   Fop the SCPD students, it is easier for us to turn in the problem sets
via SCPD courier service. I am wondering whether it is OK for us just
to turn in the problem sets to the courier service on the same day the
problem set is due?

   Judging from the courier schedule, it seems that the courier finishes
picking up the assignments from each company by 2:30pm. But, I don't
know if the problem sets can actually get to Mina Madrigal Torres's desk
by 5pm on the same day. However, the route form that is attached to the
problem set should indicate that we indeed turn in on the day and probably
before 2:30pm.


Thank you,


An-Hoe

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: Trouble signing up for section
Date: Thu, 10 Jan 2002 16:38:26 -0800
Organization: Stanford University
Lines: 3
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: dnab404691.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010709553 14231 171.64.70.145 (11 Jan 2002 00:39:13 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.73 [en]C-{C-UDP; EBM-SONY1}  (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2375

Try again, and see if you find the same problem.
I think the machine that hosts the cgi script was 
inaccessible for a little while this morning.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Darren Lewis" 
Newsgroups: su.class.cs244a
Subject: PS1, question 2
Date: Thu, 10 Jan 2002 21:34:16 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: darren.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2376

For 2b and c, what size packet and bandwidth should we use for the
calculations? Thanks.


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Would there be any exceptions for qual-takers?
Date: Thu, 10 Jan 2002 23:59:49 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2377

Dear professor and TAs,

I wonder whether there would be some leniency about the deadlines for
assignments,
for qual-takers...

Thank you..



.

Path: shelby.stanford.edu!nntp.stanford.edu!appenz
From:  (Guido Appenzeller)
Newsgroups: su.class.cs244a
Subject: Re: turn in the problem set via SCPD courier service
Date: 11 Jan 2002 15:56:23 GMT
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
X-Newsreader: TIN [version 1.2 PL2]
Xref: nntp.stanford.edu su.class.cs244a:2378

An-Hoe,

yes, if you are a SITN student and you hand your problem set to the courier
before the deadline that qualifies as being on time.

  Guido


An-hoe Shih  wrote:
: Dear Professor and TAs,

:    Fop the SCPD students, it is easier for us to turn in the problem sets
: via SCPD courier service. I am wondering whether it is OK for us just
: to turn in the problem sets to the courier service on the same day the
: problem set is due?

:    Judging from the courier schedule, it seems that the courier finishes
: picking up the assignments from each company by 2:30pm. But, I don't
: know if the problem sets can actually get to Mina Madrigal Torres's desk
: by 5pm on the same day. However, the route form that is attached to the
: problem set should indicate that we indeed turn in on the day and probably
: before 2:30pm.


: Thank you,


: An-Hoe


--

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine4.Stanford.EDU!sandeept
From: Sandeep Vilas Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Trouble signing up for section
Date: Fri, 11 Jan 2002 09:44:24 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Max Spevack 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2379

I used Mozilla 0.9.5 on Linux a few days ago (Tuesday I think) and had no
problem.  Sounds like a server problem anyway.

Good luck!

-Sandeep
P.S. Don't forget to sign up on eeclass.stanford.edu as well.

On Thu, 10 Jan 2002, Max Spevack wrote:

> Hi,
>
> I tried to sign up for the class, but I received an error message:
>
> You don't have permission to access /cgi-bin/cs244a/assign_TA_main.pl on
> this server.
>
> Has anyone else had difficulty, or is my build of Mozilla just really bad?
> :-)
>
> --Max
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Stevens Code
Date: Fri, 11 Jan 2002 10:48:49 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2380

Is the code available in the Stevens book (such as the written and readn
functions) available for our use?  Thanks.

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Permission Denied
Date: Fri, 11 Jan 2002 11:20:24 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2381

In the specification of assignment 1 it says that there is a sample ftpd to
test with located at
/usr/class/cs244a/WWW/homeworks/hw1/ftpd_src/wu-ftpd-2.0/bin/ftpd.  However
I get a Permission Denied error when trying to enter the WWW directory from
the leland machines.  Anyone else having a problem or am I doing something
wrong?  Thanks

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine17.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS1, question 2
Date: 11 Jan 2002 20:32:53 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine17.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2382

"Darren Lewis"  writes:

>For 2b and c, what size packet and bandwidth should we use for the
>calculations? Thanks.


You don't need these parameters for this question; i.e. just compute the time
for a single bit to make its way to the destination and back.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Pete Belknap" 
Newsgroups: su.class.cs244a
Subject: PS1, Question 7
Date: Fri, 11 Jan 2002 13:20:40 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2383

How does the cumulative Service Process, S(t), work for question 7?

Specifically, how can S(t) be greater than A(t) ?  To me, this would seem to
say that we have serviced more bits than have arrived...  I guess my
question is what exactly is the service process and what exactly it means to
'service' a bit...  Anybody know?

thanks,
Pete




.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine31.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS1, Question 7
Date: 11 Jan 2002 21:48:02 GMT
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine31.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2384

"Pete Belknap"  writes:

>How does the cumulative Service Process, S(t), work for question 7?

>Specifically, how can S(t) be greater than A(t) ?  To me, this would seem to
>say that we have serviced more bits than have arrived...  I guess my
>question is what exactly is the service process and what exactly it means to
>'service' a bit...  Anybody know?

>thanks,
>Pete


Part (b) of the question should clarify the meaning of the service process.
(S(t) reflects the service rate of the queue, i.e. it's the total number
of bits that could have been sent out by time t if they'd been available).

To give an example, you could imagine a router capable of filling a 100 b/s
output link, for which only a 10 b/s data stream is arriving:  S(t) = 100t,
A(t) = 10t.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: Stevens Code
Date: Fri, 11 Jan 2002 14:04:49 -0800
Organization: Stanford University
Lines: 6
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010786620 23818 171.66.211.98 (11 Jan 2002 22:03:40 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2386

Russ,
You'll find a detailed description of 
acceptable re-use in Handout #1, under
the section on the Honor Code.

- NM
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine36.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Would there be any exceptions for qual-takers?
Date: 11 Jan 2002 22:15:02 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine36.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2387

"BAEHOPIL"  writes:

>Dear professor and TAs,

>I wonder whether there would be some leniency about the deadlines for
>assignments,
>for qual-takers...

>Thank you..

The deadlines for hw#1 and ps#1 were picked with EE Quals in mind.
PS#1 won't take you long. HW#1 is not due until the week after Quals.
So no extensions...

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!anhoe
From: An-hoe Shih 
Newsgroups: su.class.cs244a
Subject: Are the slides from Review Session (today) posted online?
Date: Fri, 11 Jan 2002 15:43:26 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2388

Dear Professor and TAs,

   Are the slides from Review Session today available online? I cannot
find them anywhere. If not, can you post them?

Thanks,


An-Hoe

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine23.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Permission Denied
Date: Fri, 11 Jan 2002 16:54:34 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine23.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2389

that's the same error i got too.
~Ashmi

On Fri, 11 Jan 2002, Russell Greene wrote:

> In the specification of assignment 1 it says that there is a sample ftpd to
> test with located at
> /usr/class/cs244a/WWW/homeworks/hw1/ftpd_src/wu-ftpd-2.0/bin/ftpd.  However
> I get a Permission Denied error when trying to enter the WWW directory from
> the leland machines.  Anyone else having a problem or am I doing something
> wrong?  Thanks
>
> --Russ
>
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine23.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Discussing Problem Set
Date: Fri, 11 Jan 2002 17:01:48 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine23.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2390

hi Professor McKeown and TAs,
are we allowed to discuss problem set questions with fellow students? of
course we would cite who we discuss with.
thanks,
ashmi


###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Permission Denied
Date: 12 Jan 2002 02:28:05 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine33.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2391

"Russell Greene"  writes:

>In the specification of assignment 1 it says that there is a sample ftpd to
>test with located at
>/usr/class/cs244a/WWW/homeworks/hw1/ftpd_src/wu-ftpd-2.0/bin/ftpd.  However
>I get a Permission Denied error when trying to enter the WWW directory from
>the leland machines.  Anyone else having a problem or am I doing something
>wrong?  Thanks

>--Russ


There's now a copy of the executable (ftpd) in /usr/class/cs244a/bin,
which you can use for testing at the moment.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Discussing Problem Set
Date: 12 Jan 2002 02:31:55 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine33.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2392

Ashmi  writes:

>hi Professor McKeown and TAs,
>are we allowed to discuss problem set questions with fellow students? of
>course we would cite who we discuss with.
>thanks,
>ashmi


Please refer to the discussion of the honour code in handout 1.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Darren Lewis" 
Newsgroups: su.class.cs244a
Subject: Re: PS1, Question 7
Date: Fri, 11 Jan 2002 21:29:24 -0800
Lines: 36
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: darren.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2393

How about d(t), the delay of a single bit through the queue? Does d(1) refer
to the delay of the first bit through the queue, or the average delay of
bits through the queue at time 1, or something else entirely? Thanks.

-Darren

"Matthew Jonathan Holliman"  wrote in message

> "Pete Belknap"  writes:
>
> >How does the cumulative Service Process, S(t), work for question 7?
>
> >Specifically, how can S(t) be greater than A(t) ?  To me, this would seem
to
> >say that we have serviced more bits than have arrived...  I guess my
> >question is what exactly is the service process and what exactly it means
to
> >'service' a bit...  Anybody know?
>
> >thanks,
> >Pete
>
>
> Part (b) of the question should clarify the meaning of the service
process.
> (S(t) reflects the service rate of the queue, i.e. it's the total number
> of bits that could have been sent out by time t if they'd been available).
>
> To give an example, you could imagine a router capable of filling a 100
b/s
> output link, for which only a 10 b/s data stream is arriving:  S(t) =
100t,
> A(t) = 10t.
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: PS 1, Question 5b
Date: Fri, 11 Jan 2002 23:14:05 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2394

I'd appreciate a clarification:

How far along is the transmission of the first-in-line packet when our
packet arrives? Just beginning? Just done?

Also, what should we take the "arrival" of our packet, referred to above, to
mean? When the first bit hits the router? When the last bit has come in?

Thanks,
Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Re: PS 1, Question 5b
Date: Sat, 12 Jan 2002 11:14:03 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Steven Siuhong Ngai 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2395


Steven,

This is what my interpretation is:

1) The first in-line packet is just beginning to be transmitted.

2) A packet would arrive once the last bit comes thru.

If Profs or TAs disagree with this interpretation, then please reply back.

Thnaks,
Priyanka

On Fri, 11 Jan 2002, Steven Siuhong Ngai wrote:

> I'd appreciate a clarification:
>
> How far along is the transmission of the first-in-line packet when our
> packet arrives? Just beginning? Just done?
>
> Also, what should we take the "arrival" of our packet, referred to above, to
> mean? When the first bit hits the router? When the last bit has come in?
>
> Thanks,
> Steve
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Fluid Model Example (Lecture 2)
Date: Sat, 12 Jan 2002 11:39:42 -0800
Lines: 42
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2396


Hello Professor and TAs,

I am not clear on the Fluid Model Example(last slide for lecture 2).

It mentions the following:
A(t) = 1000 b/s
Every seconds a train of 100 bits arrive.
D(t) = 500 b/s

Here are the questions I have:
1) What really is the difference between departure rate and service rate.
Is Service Rate the rate at which a router could operate and departure
rate the rate at which it is currently operating depending on the number
of bits/packets in the queue? In which case is it true that S(t) >= D(t)

2) Does Arrival Rate refers to the maximum data rate? Or the actual data
rate. Looking at all the questions it seems that although the arrival rate
could be 1000b/s, bits might actually be arriving only at 100b/s

3) Arrival rate and Service rate are completely independent of each other?

4) In this particular example, I do not see why there would be a queue
occupancy. The packets are coming at  100b/s whereas the service rate
could be as much as 500b/s, then why would there ever be anything in the
queue. Would the departing process would start only once the
entire 100 bit packet for example has arrived?

5) Also, I do not understand the solution properly.
The solution to the problem according to me should have been:

During each cycle the queue fills at a rate of 1000b/s for 01.s, then
drains at rate of 500b/s for 0.2 s. Not sure how I would get queue
occupancy after these calculations though?

I would really appreciate if you could try and clear these for me.

Thanks,
Priyanka



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS1, Question 7
Date: 12 Jan 2002 20:26:21 GMT
Lines: 10
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine16.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2397

"Darren Lewis"  writes:

>How about d(t), the delay of a single bit through the queue? Does d(1) refer
>to the delay of the first bit through the queue, or the average delay of
>bits through the queue at time 1, or something else entirely? Thanks.

>-Darren


d(1) is the time spent in the queue by a bit that arrived at time t=1.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Fluid Model Example (Lecture 2)
Date: 12 Jan 2002 21:15:06 GMT
Lines: 69
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine16.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2398

Priyanka Gupta  writes:


>I am not clear on the Fluid Model Example(last slide for lecture 2).

>It mentions the following:
>A(t) = 1000 b/s
>Every seconds a train of 100 bits arrive.
>D(t) = 500 b/s

The slide states that the *maximum* departure rate (i.e. service rate)
is 500 b/s; as the figure shows, D(t) is not equal to 500t.

Also, keep in mind that A(t) and D(t) represent the *cumulative* number of
bits that have arrived and departed, respectively; so A(t) = 1000 would
state that 1000 bits were in the buffer at time t=0, and that no more
ever arrived again.

For this fluid model example, A(t) = 100*floor(t) + 1000*(t - floor(t))
for t between [0,0.1], [1,1.1], [2,2.1], etc. (bits are arriving at rate
1000 b/s during those times); and for other t, A(t) remains constant
(100*ceil(t), as no more bits are arriving).  (I hope I got this right;
anyway, the picture is easier to look at to see how A(t) behaves).


>Here are the questions I have:
>1) What really is the difference between departure rate and service rate.
>Is Service Rate the rate at which a router could operate and departure
>rate the rate at which it is currently operating depending on the number
>of bits/packets in the queue? In which case is it true that S(t) >= D(t)

Yes.

>2) Does Arrival Rate refers to the maximum data rate? Or the actual data
>rate. Looking at all the questions it seems that although the arrival rate
>could be 1000b/s, bits might actually be arriving only at 100b/s

The arrival rate during the first tenth of each second is 1000 b/s in this
example; the average arrival rate is 100 b/s.

>3) Arrival rate and Service rate are completely independent of each other?

Yes.

>4) In this particular example, I do not see why there would be a queue
>occupancy. The packets are coming at  100b/s whereas the service rate
>could be as much as 500b/s, then why would there ever be anything in the
>queue. Would the departing process would start only once the
>entire 100 bit packet for example has arrived?

On the average, bits are arriving at 100 b/s.  However, the instantaneous
rate is bigger than this (viz. 1000 b/s), hence the queue occupancy.

For this model, you needn't think in terms of packets (unless the problem
asks you to do so).  So in this case, the departure process begins
immediately, as the graph of D(t) on the slide indicates.

>5) Also, I do not understand the solution properly.
>The solution to the problem according to me should have been:

>During each cycle the queue fills at a rate of 1000b/s for 01.s, then
>drains at rate of 500b/s for 0.2 s. Not sure how I would get queue
>occupancy after these calculations though?

Since the example assumes just a flow of bits, the queue begins draining at
time t=0 rather than t=0.1 s.

Hope this helps.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: PS1, Problem 6
Date: Sat, 12 Jan 2002 14:02:41 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2399

1. Does the output link have to have at least I seconds between consecutive
packets on a link as well?

2. Since there is only one output link, we can start sending bits before the
entire packet arrives, correct?  (like the fluid model in class)

Thanks for your time,
Bret



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine4.Stanford.EDU!sandeept
From: Sandeep Vilas Tamhankar 
Newsgroups: su.class.cs244a
Subject: Fluid Model Slide
Date: Sat, 12 Jan 2002 14:34:30 -0800
Lines: 71
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2400

I stared at this slide for several minutes trying to figure out how the
example leads to the solution.  Finally, I decided to ignore the question
and just draw the Q(t) described in the solution.  Here's what I came up
with:

"During each cycle, the queue fills at rate 500b/s for 0.1s..." means that
you're receiving 50 bit packets in the "train", not 100 bits.  Secondly,
it means that the incoming rate is 500b/s, not 1000 as the example states.

"then drains at rate 500b/s for 0.1s", which makes sense since you
wouldn't start dumping a packet (of 50 bits) until you'd received those 50
bits, and your departure rate is 500b/s, as the example states.

So, if you draw Q(t), you get something like this:

   /\
  /  \
 /    \
/      \______________________
0  .1  .2                    1.0

The peak occurs at t=0.1 and has value 50 bits.  The average value of
Q(t) is the area under the curve from t=0 to t=1.  Since we've just got a
triangle, that's 0.5 * 0.2 * 50 = 5 bits, the final answer in the
solution.

For about a minute, I looked at the example graph, which shows data
arriving and departing simultaneously, and my interpretation of it was
that some earlier packet is departing while this packet is arriving.  But
if that were the case, it would mean that in the cycle, the router's
receiving more packets than are departing, which means that over time, its
buffers will grow to unbounded size (i.e. overflow).  Here's what I think
A(t) and D(t) should look like:

A(t):
                       -------- (etc.)
                      /
                     /
                    /
   -----------------   A(t)=50
  /
 /
/
0  0.1            1.0 1.1


D(t):
                          ------- (etc.)
                         /
                        /
                       /
      -----------------  D(t)=50
     /
    /
   /
---
0 0.1 0.2             1.1 1.2

In other words, D(t) is just a 0.1 shifted A(t).  And this makes sense
since the departure rate is the arrival rate.  And the Q(t) drawing I
started with is equal to A(t)-D(t).  If the departure rate was different
from the arrival rate, it would still have a similar offset, but the
slope would be different...and would make Q(t) not be an isosceles
triangle, but it would still be a triangle.

So, uh, any comments, critiques, general booing at my ineptitude, shouts
of "You are my networking hero; I'm naming my first child after you!" ?
Basically, am I totally off here?

-Sandeep

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: PS1, Problem 7(b)
Date: Sat, 12 Jan 2002 14:56:06 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2401

Can we sketch the fluid models as answers for the discreet processes that
are stable rather than write down X(t), d(t), D(t) as equations, or at least
just include those equations for the first second?  They are not
complicated, just annoying to write down as equations because the time
intervals are discreet.

Thanks,
Bret



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine4.Stanford.EDU!sandeept
From: Sandeep Vilas Tamhankar 
Newsgroups: su.class.cs244a
Subject: Fluid Model Slide Part II
Date: Sat, 12 Jan 2002 16:21:42 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2402

After reading question 7 a few times, I now realize the critical
mistaken assumption I made about that slide: I assumed it referred to a
store-and-forward router.  If you look at it as "continuous arrival and
service" processes, then the solution does make sense (although IMO it
does skip a few statements that would make it clearer).

Basically, you look at t=0 to t=0.1 as having a net arrival rate of
500b/s (1000 arriving - 500 departing), then t=0.1 to t=0.2 has a net
departure rate of 500b/s (500 departing - 0 arriving).  You end up with
the same Q(t) I graphed earlier.

Well, at least this exploration taught me more about the
store-and-forward case -- of course I could've learned that in due time
with question 7...I think...haven't actually done question 7 yet.

Sorry if I confused/bored you.  I'm going to go crawl in my cave now.

-Mr. Foot-in-mouth

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine41.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS1, Problem 6
Date: 13 Jan 2002 02:38:22 GMT
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine41.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2403

"Bret Taylor"  writes:

>1. Does the output link have to have at least I seconds between consecutive
>packets on a link as well?

No, you can assume continuous transmission over the output link.

>2. Since there is only one output link, we can start sending bits before the
>entire packet arrives, correct?  (like the fluid model in class)

No, the router is store-and-forward.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine4.Stanford.EDU!sandeept
From: Sandeep Vilas Tamhankar 
Newsgroups: su.class.cs244a
Subject: PS1, Question 9b
Date: Sat, 12 Jan 2002 18:53:07 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2404

1. Is this a store-and-forward switch, or continuous?
2. If store-and-forward, what's the packet size and inter-packet delay?

Thanks.

-Sandeep

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: A problem with echodemo Makefile
Date: Sat, 12 Jan 2002 19:36:11 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine39.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2405



  Hi everybody,

  I have a problem using purfiy with echodemo, when I issue "make
purify.all" command, I get the following message from purfiy :

"In order for Rational PurifyPlus to work properly, you must source
the appropriate shell resource file in /usr/pubsw/etc/rational. See
the release announcement at
    http://ls.stanford.edu/services/pubsw/package/licensed/rational.html
for more details."

What shall I do? I coulndn't find any useful info about this appropriate
shell in the given link.

  Yousr,
  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: HW Q. #6
Date: Sat, 12 Jan 2002 19:51:11 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2406

Hi,

  The problem states that "Only one of the input links is active".
Does that mean only one of the links is active at any given time
or does it mean we can completely ignore the other two links
treating them as meaningles, red-herrings thrown in at the whim of
the problem creator?  Any insight would be greatly appreciated.
Thanks

                        ~~Martin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: A problem with echodemo Makefile
Date: Sat, 12 Jan 2002 20:46:39 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2407

This is excerpted from an earlier post

>You need to source the csh script in /usr/pubsw/etc/rational
>
> >source /usr/pubsw/etc/rational/purifyplus_setup.csh
>
>  should do the trick if you are using csh/tcsh. I've added the
> line to my .cshrc so I don't have to bother sourcing it by hand each time
> I log on.

--Russ

"Amin Firoozshahian"  wrote in message

>
>
>   Hi everybody,
>
>   I have a problem using purfiy with echodemo, when I issue "make
> purify.all" command, I get the following message from purfiy :
>
> "In order for Rational PurifyPlus to work properly, you must source
> the appropriate shell resource file in /usr/pubsw/etc/rational. See
> the release announcement at
>     http://ls.stanford.edu/services/pubsw/package/licensed/rational.html
> for more details."
>
> What shall I do? I coulndn't find any useful info about this appropriate
> shell in the given link.
>
>   Yousr,
>   Amin
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: Are the slides from Review Session (today) posted online?
Date: Sat, 12 Jan 2002 23:32:13 -0800
Organization: Stanford University
Lines: 7
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010907060 7412 171.66.211.98 (13 Jan 2002 07:31:00 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2408


Until they are posted (they'll be on the "timetable" web page),
you can always look at them on the video:
   
http://scpd.stanford.edu/scpd/students/Video/embed.asp?Stream=http://cobb.stanford.edu/winter02/cs244a/020111ps/020111-cs244aps-100.wmv&ChatURL=http://chat.stanford.edu/jchat/cs244a.asp&HomePage=http://www.stanford.edu/class/cs244a/&CourseName=CS244A&videoXML=http://stanford-online.stanford.edu/winter2002/cs244avideo.xml

- Nick
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: Fluid Model Slide
Date: Sat, 12 Jan 2002 23:41:29 -0800
Organization: Stanford University
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010907617 7473 171.66.211.98 (13 Jan 2002 07:40:17 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2409

Sandeep,

You're on the right tracks, but I think you might be confused
about the meaning of A(t), D(t) on slide 33 of Handout #3.

A good way to review this (and any material) is from the
online video. You'll find this portion at:
 
http://scpd.stanford.edu/scpd/students/Video/embed.asp?Stream=http://cobb.stanford.edu/winter02/cs244a/020110/020110-cs244a-100.wmv&ChatURL=http://chat.stanford.edu/jchat/cs244a.asp&HomePage=http://www.stanford.edu/class/cs244a/&CourseName=CS244A&videoXML=http://stanford-online.stanford.edu/winter2002/cs244avideo.xml
Then go to "Fluid Model".


The problem takes some thinking about to find the solution,
but the question (and solution) are correct. 

You were right to draw Q(t). It makes the answer much
easier if you start there. From the question, you
correctly deduced Q(t). 
During the first 0.1s, bits are arriving at 1000b/s
and departing at 500b/s. So the net effect is that they
accumulate in the queue at 1000-500=500b/s. By the
end of 0.1s the queue has 50 bits in it. Because
no more bits arrive in the period [0.1,0.2], the
queue empties at 500b/s. So the average queue
occupancy over the period [0,0.2] is the same
as the average occupancy over the period [0,0.1]
which is 0.5 x (0.1 x 500) = 25 bits. The rest
follows from the normal equation for an expected
value.

Hope this helps.

- Nick
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: PS1, Problem 7(b)
Date: Sat, 12 Jan 2002 23:44:48 -0800
Organization: Stanford University
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1010907815 7532 171.66.211.98 (13 Jan 2002 07:43:35 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2410

You need to write down all the equations.
It's good practice and will make you a better person ;-)

More importantly, it will prepare you for some nice
concepts later in the class when we'll be talking about
how to provide delay guarantees from end to end in a
packet switched network. The idea builds on these fluid
models.

- Nick
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Yichen Xie 
Newsgroups: su.class.cs244a
Subject: testhw.pl & submit_test.pl
Date: Sun, 13 Jan 2002 00:41:14 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: dn800cb149.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.5) Gecko/20011014
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2411

Has anyone had any success with the test scripts yet? I can't find 
testhw.pl mentioned in the specification, and submit_test.pl fails on 
some permission errors...

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: More trouble with purify
Date: Sun, 13 Jan 2002 00:56:32 -0800
Lines: 42
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2412

Hi,

I cannot make purify work for the echodemo. This is what I did:
After making the echodemo, I tried to run purify manually by writting

purify gcc -o echodemo.purify echodemo.o -lsocket -lnsl

This what I get:

Purify 2002.05.00 Solaris 2 (32-bit) Copyright (C) 1992-2001 Rational
Software Corp.  All rights reserved.
Instrumenting: echodemo.o  Linking
gcc: file path prefix
`/usr/pubsw/apps/rational-2002.05.00/releases/purify.sol.2002.05.00/nld32/'
never used

I already sourced the environment script for running purify.

Please, can you help me know my problem? and what is the correct
syntax to run purify. I tried for a long time and till now I cannot figure
it :(

I also tried to run pruify by typing:
 purify gcc -g -O0 -ansi -Wall -DSOLARIS  -o echodemo.purify  echodemo.o
-lsocket -lnsl

But I got the same previous error.

Another thing, in the provided Makefile, there is the following line:
${BINS.PURIFY}: ${OBJS}
	${PURIFY} ${CC} ${CCFLAGS} -o 

Shouldn't this run Purify automatically?

By the way, I am connected to Solaris using xwin. I tried to run purify
from elaine and saga machine. In both cases, I am having the same above
problem.

Thanks,

Moh'd

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga19.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: PS1,problem 9c
Date: Sun, 13 Jan 2002 01:34:30 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga19.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2413

hi

In part (ii), it is not clear if the idle periods "I" for the two
processes can overlap. If so, then how do we decide on the overlapping
period? It seems like that the solution needs to depend on this
overlapping period.

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: Re: More trouble with purify
Date: Sun, 13 Jan 2002 02:22:28 -0800
Lines: 59
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Moh'd Saleem Saleem Alrawashdeh 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2414


Purify only instruments your executable, but you then you need to run it.
You've done everything correctly and you can disregard the "file path
prefix" warning. You just need one more step - running "echodemo.purify"
which purify builds for you.

Hope this helps,
Alex

On Sun, 13 Jan 2002, Moh'd Saleem Saleem Alrawashdeh wrote:

> Hi,
>
> I cannot make purify work for the echodemo. This is what I did:
> After making the echodemo, I tried to run purify manually by writting
>
> purify gcc -o echodemo.purify echodemo.o -lsocket -lnsl
>
> This what I get:
>
> Purify 2002.05.00 Solaris 2 (32-bit) Copyright (C) 1992-2001 Rational
> Software Corp.  All rights reserved.
> Instrumenting: echodemo.o  Linking
> gcc: file path prefix
> `/usr/pubsw/apps/rational-2002.05.00/releases/purify.sol.2002.05.00/nld32/'
> never used
>
> I already sourced the environment script for running purify.
>
> Please, can you help me know my problem? and what is the correct
> syntax to run purify. I tried for a long time and till now I cannot figure
> it :(
>
> I also tried to run pruify by typing:
>  purify gcc -g -O0 -ansi -Wall -DSOLARIS  -o echodemo.purify  echodemo.o
> -lsocket -lnsl
>
> But I got the same previous error.
>
> Another thing, in the provided Makefile, there is the following line:
> ${BINS.PURIFY}: ${OBJS}
> 	${PURIFY} ${CC} ${CCFLAGS} -o 
>
> Shouldn't this run Purify automatically?
>
> By the way, I am connected to Solaris using xwin. I tried to run purify
> from elaine and saga machine. In both cases, I am having the same above
> problem.
>
> Thanks,
>
> Moh'd
>
>

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: More trouble with purify
Date: Sun, 13 Jan 2002 10:03:29 -0800
Lines: 79
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine5.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2415

Also, you don't have to type the compile line.  The 'all.purify' 
Makefile target does it for you.  So just type:

     make all.purify

and it'll build echodemo.purify.

-Sandeep
P.S. Make sure you have the Makefile, that was located in the same 
directory where you got echodemo.c.

Alex Khomenko wrote:

> Purify only instruments your executable, but you then you need to run it.
> You've done everything correctly and you can disregard the "file path
> prefix" warning. You just need one more step - running "echodemo.purify"
> which purify builds for you.
> 
> Hope this helps,
> Alex
> 
> On Sun, 13 Jan 2002, Moh'd Saleem Saleem Alrawashdeh wrote:
> 
> 
>>Hi,
>>
>>I cannot make purify work for the echodemo. This is what I did:
>>After making the echodemo, I tried to run purify manually by writting
>>
>>purify gcc -o echodemo.purify echodemo.o -lsocket -lnsl
>>
>>This what I get:
>>
>>Purify 2002.05.00 Solaris 2 (32-bit) Copyright (C) 1992-2001 Rational
>>Software Corp.  All rights reserved.
>>Instrumenting: echodemo.o  Linking
>>gcc: file path prefix
>>`/usr/pubsw/apps/rational-2002.05.00/releases/purify.sol.2002.05.00/nld32/'
>>never used
>>
>>I already sourced the environment script for running purify.
>>
>>Please, can you help me know my problem? and what is the correct
>>syntax to run purify. I tried for a long time and till now I cannot figure
>>it :(
>>
>>I also tried to run pruify by typing:
>> purify gcc -g -O0 -ansi -Wall -DSOLARIS  -o echodemo.purify  echodemo.o
>>-lsocket -lnsl
>>
>>But I got the same previous error.
>>
>>Another thing, in the provided Makefile, there is the following line:
>>${BINS.PURIFY}: ${OBJS}
>>	${PURIFY} ${CC} ${CCFLAGS} -o 
>>
>>Shouldn't this run Purify automatically?
>>
>>By the way, I am connected to Solaris using xwin. I tried to run purify
>>from elaine and saga machine. In both cases, I am having the same above
>>problem.
>>
>>Thanks,
>>
>>Moh'd
>>
>>
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga17.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: getservbyname
Date: Sun, 13 Jan 2002 13:02:26 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga17.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2416



   Hi everybody,

  I have a question abot getservbyname, as far as I've understood, this
function returns a port number associated with the requried service and
specified protocol. I think that this port number is used in the local
host not in the server.

       First, am I right?

       Second, if I am, then why should we use this same port number to
connect the server? (in echodemo, I mean)

       Third, in the assignment we should use the specified port number
(by the user) or default FTP port (21) to connect the FTP  server,
shouldn't we?

   Yours,
   Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: PS1 Problem 7b
Date: Sun, 13 Jan 2002 16:30:17 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2417


When we are modeling the queue of a store-and-forward router, should we
still compute X(t) (queue occupancy) the same way as in the fluid model?

E.g. if the queue contains a 100-bit packet and it is being transmitted
out on a 100 bps link, does X(t) decrease linearly from 100 to 0?

Thanks,
Alex

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Accessing the ftpd
Date: Sun, 13 Jan 2002 16:44:55 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2418

Hi,

I am trying to get my hands on the ftp dameon to test with. However, for
both given directories in the handout:

/usr/class/cs244a/WWW/homeworks/hw1/ftpd_src/wu-ftpd-2.0/bin/ftpd
/usr/class/cs244a/WWW/homeworks/hw1/ftpd_src/

I have no permission.

Can you help me?

Thanks,

Moh'd



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Accessing the ftpd
Date: Sun, 13 Jan 2002 17:49:07 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2419

Matt placed a copy of the binary in /usr/class/cs244a/bin earlier.

-Sandeep

Moh'd Saleem Saleem Alrawashdeh wrote:

> Hi,
> 
> I am trying to get my hands on the ftp dameon to test with. However, for
> both given directories in the handout:
> 
> /usr/class/cs244a/WWW/homeworks/hw1/ftpd_src/wu-ftpd-2.0/bin/ftpd
> /usr/class/cs244a/WWW/homeworks/hw1/ftpd_src/
> 
> I have no permission.
> 
> Can you help me?
> 
> Thanks,
> 
> Moh'd
> 
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: PASV mode
Date: Sun, 13 Jan 2002 18:13:40 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2420

  Hello,

    Can we use PASV mode for LIST and RETR?

                Thanks.
                ~~Martin

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine22.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS1,problem 9c
Date: 14 Jan 2002 02:20:12 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine22.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2421

>In part (ii), it is not clear if the idle periods "I" for the two
>processes can overlap. If so, then how do we decide on the overlapping
>period? It seems like that the solution needs to depend on this
>overlapping period.


They can, depending on the length of the burst.  You should consider
both cases (overlapping and not) in your solution.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine22.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS1 Problem 7b
Date: 14 Jan 2002 02:23:26 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine22.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2422


>When we are modeling the queue of a store-and-forward router, should we
>still compute X(t) (queue occupancy) the same way as in the fluid model?

>E.g. if the queue contains a 100-bit packet and it is being transmitted
>out on a 100 bps link, does X(t) decrease linearly from 100 to 0?

Yes.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Response size
Date: Sun, 13 Jan 2002 19:42:54 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2423

The project doc says that no NVT ASCII string will be longer than 1024 
bytes.  Does this mean that no response from the ftp server (on the 
control connection) will be longer than 1024 bytes?

Thanks.

-Sandeep
-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: HW Q. #6
Date: 14 Jan 2002 05:45:32 GMT
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga21.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2424

Yes, you can treat this as effectively a single input, single output 
problem.

Arun

Martin Casado  wrote:

: Hi,

:   The problem states that "Only one of the input links is active".
: Does that mean only one of the links is active at any given time
: or does it mean we can completely ignore the other two links
: treating them as meaningles, red-herrings thrown in at the whim of
: the problem creator?  Any insight would be greatly appreciated.
: Thanks

:                         ~~Martin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: testhw.pl & submit_test.pl
Date: 14 Jan 2002 05:50:55 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga21.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2425

We've looked into the submit script and think we've fixed the permissions 
problem. Let us know if you continue to have difficulty.

Yichen Xie  wrote:

: Has anyone had any success with the test scripts yet? I can't find 
: testhw.pl mentioned in the specification, and submit_test.pl fails on 
: some permission errors...

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: getservbyname
Date: 14 Jan 2002 06:01:29 GMT
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga21.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2426


Yes, this is correct. Getservbyname returns the port associated with the 
FTP service. It is assumed that this the default port on which this 
service can be expected to be provided. 

For ftpcopy, if a port is supplied you should use it; oteherwise, you 
shoud prefer to use the default port determined by getservbyname.

Arun

:    Hi everybody,

:   I have a question abot getservbyname, as far as I've understood, this
: function returns a port number associated with the requried service and
: specified protocol. I think that this port number is used in the local
: host not in the server.

:        First, am I right?

:        Second, if I am, then why should we use this same port number to
: connect the server? (in echodemo, I mean)

:        Third, in the assignment we should use the specified port number
: (by the user) or default FTP port (21) to connect the FTP  server,
: shouldn't we?

:    Yours,
:    Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: PASV mode
Date: 14 Jan 2002 06:33:21 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga21.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2427

It is not necessary to use PASV to complete this assignment. However any 
reasonable technique that implements the required functionality, and is 
explained in your documentation, is acceptable.

Arun

Martin Casado  wrote:
:   Hello,

:     Can we use PASV mode for LIST and RETR?

:                 Thanks.
:                 ~~Martin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Response size
Date: 14 Jan 2002 07:00:21 GMT
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga20.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2428

Yes, this is acceptable for NVT strings read over the control connection.  
You do not need to worry about servers which return longer responses (some
servers have unusually long greetings, for example). The strings sent over
the data connection should not be limited, since a LIST string can be of
arbitrary length.

Arun

 Sandeep Tamhankar 
 wrote:

: The project doc says that no NVT ASCII string will be longer than 1024 
: bytes.  Does this mean that no response from the ftp server (on the 
: control connection) will be longer than 1024 bytes?

: Thanks.

: -Sandeep
: -- 
: ---------------------------------------------
: Sandeep V. Tamhankar			
: M.S. Student
: Computer Science
: Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: bonus question
Date: Sun, 13 Jan 2002 23:07:33 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2429

Dear all,

There hasn't been 10 persons who get correct answer. Please get your
answer in ASAP if you want to win the prize.

Good luck,

Xin

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: store-and-forward queues
Date: 14 Jan 2002 07:55:07 GMT
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga19.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2430

When the first packet arrives at a store-and-forward router with an empty
queue, is it stored in the queue before it is forwarded, or is it stored
and handled somewhere else?

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: store-and-forward queues
Date: 14 Jan 2002 10:06:42 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2431

The packet is stored completely in the queue before it can be sent on the
outbound link. Thus immediately before the service process begins to
transmit the buffered packet, the queue will have one packet worth of bits 
in it.

Arun

Derrick Wen-Shiuan Tong  wrote:

: When the first packet arrives at a store-and-forward router with an empty
: queue, is it stored in the queue before it is forwarded, or is it stored
: and handled somewhere else?

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Donghyun Kim" 
Newsgroups: su.class.cs244a
Subject: Q9
Date: Mon, 14 Jan 2002 02:12:03 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: kidhyun.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2432

could you clarify what is circuit switch and what is packet switch?



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Mudassir I Sheikha" 
Newsgroups: su.class.cs244a
Subject: email address for problem sets?
Date: Mon, 14 Jan 2002 03:01:02 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: pakistan.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4807.1700
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Xref: nntp.stanford.edu su.class.cs244a:2433

is there an email address where we can mail our problem sets if we can't
make it to gates on time?

mudassir




.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Q9
Date: 14 Jan 2002 18:42:21 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2434

There is a relatively thorough discussion in the lecture notes for
Thursday (slides 18-22). One way to think about the two is that circuit
switch has some state associated with a connection between two endpoints,
allowing it to make some guarantees about the minimum available data rate.
With packet switch, there is no such state, and the packets are routed
independently of one another. (At the end of the quarter we'll discuss
some ways you can try to make such guarantees using packet switch, though
for the time being, you do not need to worry about this).

 Donghyun Kim  wrote:
: could you clarify what is circuit switch and what is packet switch?



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Q9: Simultaneous arrival of packets
Date: 14 Jan 2002 21:43:30 GMT
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga19.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2435

For the router in question 9, if two packets are arriving on both inputs
simultaneously (one on each input), would they be received concurrently or
sequentially?

In other words, would the arrive duration be 

 (bits in packet / arrival rate)         concurrently received

or 

 (2 * bits in packet / arrival rate)     sequentially received


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Are the slides from Review Session (today) posted online?
Date: Mon, 14 Jan 2002 14:05:39 -0800
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2436

They are online now. Sorry for the delay.

  Guido

> Dear Professor and TAs,
> 
>    Are the slides from Review Session today available online? I cannot
> find them anywhere. If not, can you post them?
> 
> Thanks,
> 
> 
> An-Hoe

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: PS1, Question 9b
Date: Mon, 14 Jan 2002 14:18:38 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2437

Sandeep ,

> 1. Is this a store-and-forward switch, or continuous?
> 2. If store-and-forward, what's the packet size and inter-packet delay?

Does this make a difference for you? To guarantee that there will be no 
overrun the switch has to not drop packets with worst case packet size and 
inter-packet delay. You can assume that if it is store and forward it has 
enough buffer space to store at least one packets.

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshdan Sundar Maddali 
Newsgroups: su.class.cs244a
Subject: Question to SITN Students : IS there is a Web Based Access to NewsGroup
Date: Mon, 14 Jan 2002 14:58:29 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic0.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2438

Hi Fellow SITN students,

  I am Lakshman Maddali a NDO Student taking  CS244A course.

  i am unable to view the postings on the newsgroup from outside
stanford [error message is
  news server  not found] . i believe this is due to the fact that the
news servers are
  internal to stanford only. I am able to access the newsgroup by the
  setting the SSH Port  forwards for X terminals, and opening the
newsgroup
  from  epic.stanford.edu. But doing so is very clumsy. Wanted to know
if there is
  any better way [ may be a http based interface ]

Thanks

Lakshman Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!spevack.Stanford.EDU!spevack
From: Max Spevack 
Newsgroups: su.class.cs244a
Subject: PS1: 9c question
Date: Mon, 14 Jan 2002 15:23:18 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: spevack.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2439

Hello,

Is it proper to assume, when planning for the worst case, that the gap
between bursts, I, is zero?  It seems like we would want our router to be
able to function when it is receiving a continuous stream of bits without
pause.

Is it reasonable to proceed like this?

Thanks,
Max

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Enrollment on the EE website
Date: Mon, 14 Jan 2002 16:13:44 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2440

From lecture number 2 I came to know that I have to enroll on EE website.
Can I get a link to the same?

I tried
https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=ee384
and got an error that my student number is not in the class list.

Do I need to enroll somewhere else?

Thanks,
Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Enrollment on the EE website
Date: Mon, 14 Jan 2002 17:41:45 -0800
Lines: 59
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma10.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
 boundary="------------923E4488853F279527FC453A"
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2441


--------------923E4488853F279527FC453A
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit


Hello everyone,
Please make sure you enroll to the cs244a class, not the ee384.
the correct url is
https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=cs244a

if you receive a message saying 'you're already subscribed' then everything
is fine.

Regards,
Antonios.

Dhawal Kumar wrote:

> From lecture number 2 I came to know that I have to enroll on EE website.
> Can I get a link to the same?
>
> I tried
> https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=ee384
> and got an error that my student number is not in the class list.
>
> Do I need to enroll somewhere else?
>
> Thanks,
> Dhawal Kumar

--------------923E4488853F279527FC453A
Content-Type: text/html; charset=us-ascii
Content-Transfer-Encoding: 7bit

<!doctype html public "-//w3c//dtd html 4.0 transitional//en">
<html>
 
<br>Hello everyone,
<br>Please make sure you enroll to the cs244a class, not the ee384.
<br>the correct url is <a href="https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=ee384">https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=cs244a</a><a href="https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=ee384"></a>
<p>if you receive a message saying 'you're already subscribed' then everything
is fine.
<p>Regards,
<br>Antonios.
<p>Dhawal Kumar wrote:
<blockquote TYPE=CITE>From lecture number 2 I came to know that I have
to enroll on EE website.
<br>Can I get a link to the same?
<p>I tried
<br><a href="https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=ee384">https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=ee384</a>
<br>and got an error that my student number is not in the class list.
<p>Do I need to enroll somewhere else?
<p>Thanks,
<br>Dhawal Kumar</blockquote>
</html>

--------------923E4488853F279527FC453A--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Closing sockets and a random experience
Date: Mon, 14 Jan 2002 18:26:41 -0800
Lines: 29
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2442

I noticed in the example echo client code that on error it closes the 
socket and then exits with code 1.  Is there any reason to do this? 
Since the process is dying, wouldn't the socket automatically get 
cleaned up (by the OS)?  I look at this as analogous to free'ing 
allocated variables before exiting.

Also, on a semi-related subject, has anyone run their ftpcopy program on 
Linux?  I started the project on my home computer, and when I called 
getservbyname to get ftp's default port number (21), I ended up 
receiving 5376.  This seemed rather odd, considering that /etc/services 
did clearly state that ftp over tcp was running on port 21.  When I ran 
the code on an elaine machine, I got the expected value.

Here's a theory: the port returned in the struct is in network-byte 
ordering, which PCs aren't, but Solaris boxes are.  PC's and Solaris 
boxes have different endian-ness, right?

Anyway, I promptly switched to using an elaine machine, but I thought it 
was kinda interesting.

-Sandeep

-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: Closing sockets and a random experience
Date: Mon, 14 Jan 2002 18:48:03 -0800
Lines: 51
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2443

I am pretty sure this is because of byte ordering.  It works on my Linux
box.  The only catch is the Solaris link flags ("-lsocket -lnsl").  My
makefile contanins these lines to make it easier

LDFLAGS = -lsocket -lnsl

$(EXECUTABLE): $(LINKERFILES)
        $(PRELINK) $(CC) $(CFLAGS) -o $(EXECUTABLE) $(LINKERFILES)
$(LDFLAGS)

To make on Linux, I just type

    % make LDFLAGS=

The default make works on Solaris so their grading script will work.

Bret

"Sandeep Tamhankar"  wrote in message

> I noticed in the example echo client code that on error it closes the
> socket and then exits with code 1.  Is there any reason to do this?
> Since the process is dying, wouldn't the socket automatically get
> cleaned up (by the OS)?  I look at this as analogous to free'ing
> allocated variables before exiting.
>
> Also, on a semi-related subject, has anyone run their ftpcopy program on
> Linux?  I started the project on my home computer, and when I called
> getservbyname to get ftp's default port number (21), I ended up
> receiving 5376.  This seemed rather odd, considering that /etc/services
> did clearly state that ftp over tcp was running on port 21.  When I ran
> the code on an elaine machine, I got the expected value.
>
> Here's a theory: the port returned in the struct is in network-byte
> ordering, which PCs aren't, but Solaris boxes are.  PC's and Solaris
> boxes have different endian-ness, right?
>
> Anyway, I promptly switched to using an elaine machine, but I thought it
> was kinda interesting.
>
> -Sandeep
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Closing sockets and a random experience
Date: 15 Jan 2002 03:40:42 GMT
Lines: 46
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine22.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2444

Yes, this is an issue caused by the byte ordering on your system. The
value returned by the call is stored as 0x0015, which on a little endian
machine is interpreted as 0x1500 (5376). Since all x86 PCs are little
endian, this is a problem. On the Sun workstations, the byte ordering is
big endian, which is incidentally the same as network byte ordering. Thus 
the proper port 21 is read.

To avoid this sort problem, you merely need to use the functions ntohs()  
and ntohl() when grabbing data out of a structure containing fields
formatted in network byte order. This will transparently handle endianess
issues by either doing nothing or swapping the bytes around. This will
ensure cross machine portability, which is certainly something we are
looking for.

Arun

Sandeep Tamhankar  wrote:
: I noticed in the example echo client code that on error it closes the 
: socket and then exits with code 1.  Is there any reason to do this? 
: Since the process is dying, wouldn't the socket automatically get 
: cleaned up (by the OS)?  I look at this as analogous to free'ing 
: allocated variables before exiting.

: Also, on a semi-related subject, has anyone run their ftpcopy program on 
: Linux?  I started the project on my home computer, and when I called 
: getservbyname to get ftp's default port number (21), I ended up 
: receiving 5376.  This seemed rather odd, considering that /etc/services 
: did clearly state that ftp over tcp was running on port 21.  When I ran 
: the code on an elaine machine, I got the expected value.

: Here's a theory: the port returned in the struct is in network-byte 
: ordering, which PCs aren't, but Solaris boxes are.  PC's and Solaris 
: boxes have different endian-ness, right?

: Anyway, I promptly switched to using an elaine machine, but I thought it 
: was kinda interesting.

: -Sandeep

: -- 
: ---------------------------------------------
: Sandeep V. Tamhankar			
: M.S. Student
: Computer Science
: Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Antonios Hondroulis 
Newsgroups: su.class.cs244a
Subject: Re: Closing sockets and a random experience
Date: Mon, 14 Jan 2002 21:06:21 -0800
Lines: 56
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga15.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2445

Sandeep Tamhankar wrote:
> 
> I noticed in the example echo client code that on error it closes the
> socket and then exits with code 1.  Is there any reason to do this?
> Since the process is dying, wouldn't the socket automatically get
> cleaned up (by the OS)?  I look at this as analogous to free'ing
> allocated variables before exiting.
> 
> Also, on a semi-related subject, has anyone run their ftpcopy program on
> Linux?  I started the project on my home computer, and when I called
> getservbyname to get ftp's default port number (21), I ended up
> receiving 5376.  This seemed rather odd, considering that /etc/services
> did clearly state that ftp over tcp was running on port 21.  When I ran
> the code on an elaine machine, I got the expected value.
> 
> Here's a theory: the port returned in the struct is in network-byte
> ordering, which PCs aren't, but Solaris boxes are.  PC's and Solaris
> boxes have different endian-ness, right?
> 
> Anyway, I promptly switched to using an elaine machine, but I thought it
> was kinda interesting.
> 
> -Sandeep
> 
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 

Hello Sandeep,

with regard to your first question (explicit close
vs. exit): Even though these are kernel implementation
details, it can be the case that an exit from a program
causes buffered (for transmission) data to be discarded.
Close() on the other hand, depending on the options
of the socket (set when it was opened) may or may not have
to wait until all pending data has been transmitted. For example, the
option SO_LINGER makes a close() block until
all pending data has been transmitted. Considering that
during an exit the system is trying to reclaim resourses
in a reasonable amount of time, I believe that a sensible
implementation would probably discard buffered data, rather
than try to deliver them reliably (in the case of TCP).
In conclusion, use close() to ensure correct semantics!

As for the port number, Please note that this is in
network byte order, which can be different from the
machine's byte order. You can use the ntohs(), ntohl()
functions to make sure you read the right numbers at
any machine running your code.

Regards,
Antonios.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Antonios Hondroulis 
Newsgroups: su.class.cs244a
Subject: Re: PS1: 9c question
Date: Mon, 14 Jan 2002 21:11:28 -0800
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga15.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2446

Max Spevack wrote:
> 
> Hello,
> 
> Is it proper to assume, when planning for the worst case, that the gap
> between bursts, I, is zero?  It seems like we would want our router to be
> able to function when it is receiving a continuous stream of bits without
> pause.
> 
> Is it reasonable to proceed like this?
> 
> Thanks,
> Max

Your answer should apply for the general case where I is
any non-negative number, not just 0. Think of it as
a function of I. Of course this function of I may or maynot
have its maximum value at I=0, but this isn't what the
assignment is about.

Regards,
Antonios.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Antonios Hondroulis 
Newsgroups: su.class.cs244a
Subject: Re: Q9: Simultaneous arrival of packets
Date: Mon, 14 Jan 2002 21:18:20 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga15.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2447

Derrick Wen-Shiuan Tong wrote:
> 
> For the router in question 9, if two packets are arriving on both inputs
> simultaneously (one on each input), would they be received concurrently or
> sequentially?
> 
> In other words, would the arrive duration be
> 
>  (bits in packet / arrival rate)         concurrently received
> 
> or
> 
>  (2 * bits in packet / arrival rate)     sequentially received

Hi Derrick,
Please keep in mind that in circuit switching there's a
circuit set up before data transmission can proceed. Also,
packet switches have buffers to store packets arriving
on the input ports before they can be forwarded on the
output ports.

Regards,
Antonios.
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!znash
From: Zack Al-Ubaidi Nash 
Newsgroups: su.class.cs244a
Subject: getsockname
Date: Mon, 14 Jan 2002 21:35:11 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2448

Hello,
	I am having a hard time determining the local ip address after
connecting a socket to the ftp server.  I need to obtain the local ip
address so that I can bind the data socket to that ip address and then retrieve the port
number by using getsockname.  I am having a hard time retrieving the local
ip address to pass to bind.  My understanding running getsockname after a
socket is connected not binded will return the local ip and port num, but
all I recieve is a 0 or null.

thanks,
Zack

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Yichen Xie 
Newsgroups: su.class.cs244a
Subject: question about "ext" and filename matching
Date: Mon, 14 Jan 2002 21:43:39 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: dn800cb149.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.5) Gecko/20011014
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2449

Does "extension" mean ftpcopy should match suffix ".ext" or just "ext"? 
In other words, should a.tgz match as a gz file? Thanks.

Yichen

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Failure to establish data connection
Date: Mon, 14 Jan 2002 22:49:47 -0800
Lines: 42
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2450

Hi,

I followed the steps mentioned in the handout to establish a data
connection. I have the following questions:

1. Should we use port 0 for the client? Or should we get a random port
that is larger than the 1024? Is there a way to know what are the nonused
ports so that we can use one of them?

2. For listen command, should the second parameter be 1 (since we will
have one outside connection with the server)?

3. About the 8 steps to establish a data connection, which are
mentioned in the handout. Why we need step 4 in the establishing of the
connection? Since the port of the socket is already defined in the bind()
step. And when we read it back using the getsockname(), we will always
get the same port number that we set in the bind step? So I am not finding
a use for it.

4. I did all the steps till 7 ( which is sending a ftp CMD over control
connection). For the PORT command, the server reply was successful.
However, when I sent the command "LIST \r\n" to start reading the
directory, what I got is Transient negative reply, with the following
message from the server:

Can't build data connection: Connection refused

I got this message, although the server returned a Positive Completion
Code for the PORT command that should precede it. I tried with Port 0 as
mentioned in the handout, and with port 1400 as a random port. For both
cases, I got the same wrong behavior.

I spent some time, but with no hope. I believe that I am doing everything
as mentioned in the handout, but I don't know why the server is not
accepting my data connection. I am conecting to ftp.stanford.edu, and
before this steps, I ran USER, PASS, TYPE commands successfully.

I will appreciate any help.
Thanks,

Moh'd

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!byang
From: Beverly Yang 
Newsgroups: su.class.cs244a
Subject: Re: testhw.pl & submit_test.pl
Date: Mon, 14 Jan 2002 23:04:37 -0800
Lines: 37
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2451


Actually, I am still having problems.  The file

/afs/ir/class/cs244a/bin/testhw.pl

doesn't exist.  The submit script in the same directory runs, but it gives
me the following output:

saga1:~/cs244a/hw1/test> /afs/ir/class/cs244a/bin/submit
Format: submit TA Assignment

TA is one of: croyer

Assignment is one of: hw1 hw2
hw1             - Selective Repeat
hw2             - Router Table


My TA is not listed, and the assignment names are incorrect.  It looks
like test_code.cfg contains the correct information, but is not being read
in correctly.  Please let us know when these problems have been fixed.

Thanks!

On 14 Jan 2002, Arun Upadhyaya Kishan wrote:

> We've looked into the submit script and think we've fixed the permissions
> problem. Let us know if you continue to have difficulty.
>
> Yichen Xie  wrote:
>
> : Has anyone had any success with the test scripts yet? I can't find
> : testhw.pl mentioned in the specification, and submit_test.pl fails on
> : some permission errors...
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: bonus question
Date: Mon, 14 Jan 2002 23:24:32 -0800
Organization: Stanford University
Lines: 6
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1011079552 27898 171.66.211.98 (15 Jan 2002 07:25:52 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2452


Just to let everyone know that we now have the 10th
correct answer. The winner will be announced in due course.

Thanks,
- NM
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Alberto Robles 
Newsgroups: su.class.cs244a
Subject: downloading demo & ftp files?
Date: Mon, 14 Jan 2002 23:44:31 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2453

Hey Folks,

Is there a way to copy the demo & ftp files to our home directories
through AFS?

I can see the files and access them independently via http @
stanford.edu/class/cs244a/homeworks, but permissions to the
WWW prevent me from copying the entire the hw1 directory over via
/usr/class/cs244a/WWW/...

Cheers,
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 00:19:52 -0800
Lines: 43
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2454

I struggled with that for a few hours today, too.  I tried using 
inet_ntoa on a part of the hostent structure, but I couldn't get the 
damn thing to compile (and I tried casting all sorts of things all sorts 
of ways, too).

And just when I thought all hope was lost and that I should ask one of 
my friends (it took all my will power to not throw my computer off of my 
balcony at that point), I found an example on p. 243-244 of the Stevens 
book.

It really hurts when you run into something so simple (in the sense of, 
"Damn it, I just want my IP address, is that so much to ask for???), yet 
so seemingly insurmountable.

Good luck!

-Sandeep

Zack Al-Ubaidi Nash wrote:

> Hello,
> 	I am having a hard time determining the local ip address after
> connecting a socket to the ftp server.  I need to obtain the local ip
> address so that I can bind the data socket to that ip address and then retrieve the port
> number by using getsockname.  I am having a hard time retrieving the local
> ip address to pass to bind.  My understanding running getsockname after a
> socket is connected not binded will return the local ip and port num, but
> all I recieve is a 0 or null.
> 
> thanks,
> Zack
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: downloading demo & ftp files?
Date: Tue, 15 Jan 2002 00:25:23 -0800
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2455

Yeah, it's kind of annoying.  I was using a web browser on my home 
machine and wanted to download the files on elaine, so I ended up going 
through the web to find the exact URLs and using wget on elaine* to get 
the files.

If you're using a browser on a machine that has access to your home 
directory, then just click on the files in the browser and save them; 
it's the easiest way (particularly since we're only talking about 2-3 
files).

-Sandeep

Luis Alberto Robles wrote:

> Hey Folks,
> 
> Is there a way to copy the demo & ftp files to our home directories
> through AFS?
> 
> I can see the files and access them independently via http @
> stanford.edu/class/cs244a/homeworks, but permissions to the
> WWW prevent me from copying the entire the hw1 directory over via
> /usr/class/cs244a/WWW/...
> 
> Cheers,
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: downloading demo & ftp files?
Date: Tue, 15 Jan 2002 00:47:03 -0800
Lines: 50
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2456

Thanks Sandeep - 

I used the click and save method for the 3 demo files, but also
wanted to snag the wu_ftp and parse files... 

"wget -r /cs244a/homeworks/hw1" did the trick 
( though it also downloaded every link and image from the cs244a site...
)

Nifty tool though.

Sandeep Tamhankar wrote:
> 
> Yeah, it's kind of annoying.  I was using a web browser on my home
> machine and wanted to download the files on elaine, so I ended up going
> through the web to find the exact URLs and using wget on elaine* to get
> the files.
> 
> If you're using a browser on a machine that has access to your home
> directory, then just click on the files in the browser and save them;
> it's the easiest way (particularly since we're only talking about 2-3
> files).
> 
> -Sandeep
> 
> Luis Alberto Robles wrote:
> 
> > Hey Folks,
> >
> > Is there a way to copy the demo & ftp files to our home directories
> > through AFS?
> >
> > I can see the files and access them independently via http @
> > stanford.edu/class/cs244a/homeworks, but permissions to the
> > WWW prevent me from copying the entire the hw1 directory over via
> > /usr/class/cs244a/WWW/...
> >
> > Cheers,
> >
> 
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 08:31:30 -0800
Lines: 73
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2457

hi,
in the explanation of bind() (top of pg 92, stevens), it is mentioned that
a tcp client should normally not bind an ip address to the socket and
allow the kernel to do it when the socket connects.

does this apply to us while establishing the data connection, or is this
actually talking about what the server will do at its end? (it is the
server that has the outgoing link on the data connection).

any ideas?
thanks!
~Ashmi


On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:

> I struggled with that for a few hours today, too.  I tried using
> inet_ntoa on a part of the hostent structure, but I couldn't get the
> damn thing to compile (and I tried casting all sorts of things all sorts
> of ways, too).
>
> And just when I thought all hope was lost and that I should ask one of
> my friends (it took all my will power to not throw my computer off of my
> balcony at that point), I found an example on p. 243-244 of the Stevens
> book.
>
> It really hurts when you run into something so simple (in the sense of,
> "Damn it, I just want my IP address, is that so much to ask for???), yet
> so seemingly insurmountable.
>
> Good luck!
>
> -Sandeep
>
> Zack Al-Ubaidi Nash wrote:
>
> > Hello,
> > 	I am having a hard time determining the local ip address after
> > connecting a socket to the ftp server.  I need to obtain the local ip
> > address so that I can bind the data socket to that ip address and then retrieve the port
> > number by using getsockname.  I am having a hard time retrieving the local
> > ip address to pass to bind.  My understanding running getsockname after a
> > socket is connected not binded will return the local ip and port num, but
> > all I recieve is a 0 or null.
> >
> > thanks,
> > Zack
> >
> >
>
>
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 08:41:03 -0800
Lines: 48
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Sandeep Tamhankar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2458


> I struggled with that for a few hours today, too.  I tried using
> inet_ntoa on a part of the hostent structure, but I couldn't get the
> damn thing to compile (and I tried casting all sorts of things all sorts
> of ways, too).
>
> And just when I thought all hope was lost and that I should ask one of
> my friends (it took all my will power to not throw my computer off of my
> balcony at that point), I found an example on p. 243-244 of the Stevens
> book.

I don't think the example on page 244 is really adequate.  The current
host may have multiple interfaces and there is no way to deduce which
interface the socket was bound to.  I think a better solution may be
to bind the socket to INADDR_ANY and request an ephemeral port, then
use getsockname().  Here is an excerpt from Steven's sock_bind_wild(..)


       socklen_t       len;

       struct sockaddr_in      sin;

       bzero(&sin, sizeof(sin));
       sin.sin_family = AF_INET;
       sin.sin_addr.s_addr = htonl(INADDR_ANY);
       sin.sin_port = htons(0);        /* bind ephemeral port */

       if (bind(sockfd, (SA *) &sin, sizeof(sin)) < 0)
               return(-1);
       len = sizeof(sin);
       if (getsockname(sockfd, (SA *) &sin, &len) < 0)
               return(-1);
       return(sin.sin_port);

I think you can get the octet/dot notation of the IP address using inet_ntoa(..),
but I haven't tried it.

                    HTH,
                    ~~m

> It really hurts when you run into something so simple (in the sense of,
> "Damn it, I just want my IP address, is that so much to ask for???), yet
> so seemingly insurmountable.
>
> Good luck!
>
> -Sandeep

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 09:47:33 -0800
Lines: 78
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2459

I tried this and got the same result as what I'd gotten yesterday during 
my unsuccessful attempts: an address of 0.0.0.0.  It seems that binding 
and getting the socket name doesn't update sin.sin_addr from the initial 
value that we set (INADDR_ANY, which == 0).

I agree that p. 244 is inadequate since you don't know which interface 
to use; there must be a way.  At the moment, however, I don't know what 
that way is.

If anyone knows the right way to do this, please don't be shy.  In the 
meantime, I'm going to ask a few friends (one of them is developing a 
C/C++ server right now, so the knowledge should be relatively fresh in 
his mind).

-Sandeep

Martin Casado wrote:

>>I struggled with that for a few hours today, too.  I tried using
>>inet_ntoa on a part of the hostent structure, but I couldn't get the
>>damn thing to compile (and I tried casting all sorts of things all sorts
>>of ways, too).
>>
>>And just when I thought all hope was lost and that I should ask one of
>>my friends (it took all my will power to not throw my computer off of my
>>balcony at that point), I found an example on p. 243-244 of the Stevens
>>book.
>>
> 
> I don't think the example on page 244 is really adequate.  The current
> host may have multiple interfaces and there is no way to deduce which
> interface the socket was bound to.  I think a better solution may be
> to bind the socket to INADDR_ANY and request an ephemeral port, then
> use getsockname().  Here is an excerpt from Steven's sock_bind_wild(..)
> 
> 
>        socklen_t       len;
> 
>        struct sockaddr_in      sin;
> 
>        bzero(&sin, sizeof(sin));
>        sin.sin_family = AF_INET;
>        sin.sin_addr.s_addr = htonl(INADDR_ANY);
>        sin.sin_port = htons(0);        /* bind ephemeral port */
> 
>        if (bind(sockfd, (SA *) &sin, sizeof(sin)) < 0)
>                return(-1);
>        len = sizeof(sin);
>        if (getsockname(sockfd, (SA *) &sin, &len) < 0)
>                return(-1);
>        return(sin.sin_port);
> 
> I think you can get the octet/dot notation of the IP address using inet_ntoa(..),
> but I haven't tried it.
> 
>                     HTH,
>                     ~~m
> 
> 
>>It really hurts when you run into something so simple (in the sense of,
>>"Damn it, I just want my IP address, is that so much to ask for???), yet
>>so seemingly insurmountable.
>>
>>Good luck!
>>
>>-Sandeep
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 12:13:17 -0800
Lines: 38
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2460

See page 250 in Stevens.  There is a method that returns the local host's IP
addresses using the uname() method and subsequent networking method calls:

char **my_addrs(int *addrtype)
{
    struct hostent *hptr;
    struct utsname myname;

    if (uname(&myname) < 0)
        return NULL;

    if ((hptr = gethostbyname(myname.nodename)) == NULL)
        return NULL;

    *addrtype = hptr->h_addrtype;
    return hptr->h_addr_list;
}

Hope this works,
Bret

"Zack Al-Ubaidi Nash"  wrote in message

> Hello,
> I am having a hard time determining the local ip address after
> connecting a socket to the ftp server.  I need to obtain the local ip
> address so that I can bind the data socket to that ip address and then
retrieve the port
> number by using getsockname.  I am having a hard time retrieving the local
> ip address to pass to bind.  My understanding running getsockname after a
> socket is connected not binded will return the local ip and port num, but
> all I recieve is a 0 or null.
>
> thanks,
> Zack
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 12:29:48 -0800
Lines: 50
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2461

To get the local ip address, I used the getenv command to get the host
name and resolved that host name using the gethostbyname. It gives me back
a 32 bit address. I think that it should be fine.

Moh'd


 On Tue, 15 Jan 2002, Bret Taylor wrote:

> See page 250 in Stevens.  There is a method that returns the local host's IP
> addresses using the uname() method and subsequent networking method calls:
>
> char **my_addrs(int *addrtype)
> {
>     struct hostent *hptr;
>     struct utsname myname;
>
>     if (uname(&myname) < 0)
>         return NULL;
>
>     if ((hptr = gethostbyname(myname.nodename)) == NULL)
>         return NULL;
>
>     *addrtype = hptr->h_addrtype;
>     return hptr->h_addr_list;
> }
>
> Hope this works,
> Bret
>
> "Zack Al-Ubaidi Nash"  wrote in message
> 
> > Hello,
> > I am having a hard time determining the local ip address after
> > connecting a socket to the ftp server.  I need to obtain the local ip
> > address so that I can bind the data socket to that ip address and then
> retrieve the port
> > number by using getsockname.  I am having a hard time retrieving the local
> > ip address to pass to bind.  My understanding running getsockname after a
> > socket is connected not binded will return the local ip and port num, but
> > all I recieve is a 0 or null.
> >
> > thanks,
> > Zack
> >
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 12:33:46 -0800
Lines: 77
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2462


Here is how ftp in inetutils does it

First it connects to the remote ftpd (for the command connection not the data)

  connect(s, (struct sockaddr *)&hisctladdr, sizeof (hisctladdr)) < 0)

Then it calls getsockname with an uninitialized sockaddr_in

    struct  sockaddr_in myctladdr;
    getsockname(s, (struct sockaddr *)&myctladdr, &len) < 0)

When it gets ready to open a data port, it uses myctladdr for the
ip and requests an ephemeral port from the os which it grabs from
another call to getsockname(..).

 data_addr = myctladdr;
    if (sendport)
        data_addr.sin_port = 0; /* let system pick one */
*** SNIP ****
    data = socket(AF_INET, SOCK_STREAM, 0);
*** SNIP ****
    if (bind(data, (struct sockaddr *)&data_addr, sizeof (data_addr)) < 0) {
*** SNIP ****
    if (getsockname(data, (struct sockaddr *)&data_addr, &len) < 0) {

Perhaps this is a valid approach to using getsockname(..).

Then again, you may wish to avoid this issue by using PASV mode communication
on LIST and RETR, which is imho quite a bit easier.


                ~~m




> See page 250 in Stevens.  There is a method that returns the local host's IP
> addresses using the uname() method and subsequent networking method calls:
>
> char **my_addrs(int *addrtype)
> {
>     struct hostent *hptr;
>     struct utsname myname;
>
>     if (uname(&myname) < 0)
>         return NULL;
>
>     if ((hptr = gethostbyname(myname.nodename)) == NULL)
>         return NULL;
>
>     *addrtype = hptr->h_addrtype;
>     return hptr->h_addr_list;
> }
>
> Hope this works,
> Bret
>
> "Zack Al-Ubaidi Nash"  wrote in message
> 
> > Hello,
> > I am having a hard time determining the local ip address after
> > connecting a socket to the ftp server.  I need to obtain the local ip
> > address so that I can bind the data socket to that ip address and then
> retrieve the port
> > number by using getsockname.  I am having a hard time retrieving the local
> > ip address to pass to bind.  My understanding running getsockname after a
> > socket is connected not binded will return the local ip and port num, but
> > all I recieve is a 0 or null.
> >
> > thanks,
> > Zack
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: getsockname
Date: Tue, 15 Jan 2002 12:54:09 -0800
Lines: 100
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2463

One of my friends just e-mailed me back with basically the same idea as 
what inetutils does.

But I played with passive mode and I agree, it's much easier.  Thanks a 
ton, Martin.

-Sandeep

Martin Casado wrote:

> Here is how ftp in inetutils does it
> 
> First it connects to the remote ftpd (for the command connection not the data)
> 
>   connect(s, (struct sockaddr *)&hisctladdr, sizeof (hisctladdr)) < 0)
> 
> Then it calls getsockname with an uninitialized sockaddr_in
> 
>     struct  sockaddr_in myctladdr;
>     getsockname(s, (struct sockaddr *)&myctladdr, &len) < 0)
> 
> When it gets ready to open a data port, it uses myctladdr for the
> ip and requests an ephemeral port from the os which it grabs from
> another call to getsockname(..).
> 
>  data_addr = myctladdr;
>     if (sendport)
>         data_addr.sin_port = 0; /* let system pick one */
> *** SNIP ****
>     data = socket(AF_INET, SOCK_STREAM, 0);
> *** SNIP ****
>     if (bind(data, (struct sockaddr *)&data_addr, sizeof (data_addr)) < 0) {
> *** SNIP ****
>     if (getsockname(data, (struct sockaddr *)&data_addr, &len) < 0) {
> 
> Perhaps this is a valid approach to using getsockname(..).
> 
> Then again, you may wish to avoid this issue by using PASV mode communication
> on LIST and RETR, which is imho quite a bit easier.
> 
> 
>                 ~~m
> 
> 
> 
> 
> 
>>See page 250 in Stevens.  There is a method that returns the local host's IP
>>addresses using the uname() method and subsequent networking method calls:
>>
>>char **my_addrs(int *addrtype)
>>{
>>    struct hostent *hptr;
>>    struct utsname myname;
>>
>>    if (uname(&myname) < 0)
>>        return NULL;
>>
>>    if ((hptr = gethostbyname(myname.nodename)) == NULL)
>>        return NULL;
>>
>>    *addrtype = hptr->h_addrtype;
>>    return hptr->h_addr_list;
>>}
>>
>>Hope this works,
>>Bret
>>
>>"Zack Al-Ubaidi Nash"  wrote in message

>>
>>>Hello,
>>>I am having a hard time determining the local ip address after
>>>connecting a socket to the ftp server.  I need to obtain the local ip
>>>address so that I can bind the data socket to that ip address and then
>>>
>>retrieve the port
>>
>>>number by using getsockname.  I am having a hard time retrieving the local
>>>ip address to pass to bind.  My understanding running getsockname after a
>>>socket is connected not binded will return the local ip and port num, but
>>>all I recieve is a 0 or null.
>>>
>>>thanks,
>>>Zack
>>>
>>>
>>
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Symbolic Links
Date: Tue, 15 Jan 2002 13:45:06 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2464

How are we supposed to handle symbolic links to directories?  For example if
the local dir is /AAA/BBB/ and the remote dir is /CCC/ and a symbolic link
(localed in the /CCC/ directory) takes us to /ZZZ/ should we create the ZZZ
directory as a subdirectory of /AAA/BBB/ or create a /ZZZ/ directory
seperate from /AAA/BBB/?.  As I write this it seems to me that the second
option is the one that needs to be followed because if /ZZZ/ is in a
different place on the local machine, the symbolic link won't work after
being copied.  Any thoughts?   Thanks.

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: question about "ext" and filename matching
Date: Tue, 15 Jan 2002 14:25:47 -0800
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2465

It should match ".ext", not only "ext". So "a.tgz" is not a "gz" file. Xin



On Mon, 14 Jan 2002, Yichen Xie wrote:

> Does "extension" mean ftpcopy should match suffix ".ext" or just "ext"? 
> In other words, should a.tgz match as a gz file? Thanks.
> 
> Yichen
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Symlinks
Date: Tue, 15 Jan 2002 14:26:07 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2466

Are we supposed to do anything with symlinks on the ftp site?  Or do we 
just deal with directories and files?

-Sandeep
-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Symlinks
Date: Tue, 15 Jan 2002 14:42:45 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2467

You need to handle symbolic links. Please refer to FAQ Q9 for details. Xin


On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:

> Are we supposed to do anything with symlinks on the ftp site?  Or do we 
> just deal with directories and files?
> 
> -Sandeep
> -- 
> ---------------------------------------------
> Sandeep V. Tamhankar			
> M.S. Student
> Computer Science
> Email: 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Failure to establish data connection
Date: Tue, 15 Jan 2002 15:47:44 -0800
Lines: 50
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2468

Hi,

> I followed the steps mentioned in the handout to establish a data
> connection. I have the following questions:
> 
> 1. Should we use port 0 for the client? Or should we get a random port
> that is larger than the 1024? Is there a way to know what are the nonused
> ports so that we can use one of them?

You should specify the port number as 0 when bind() is called.

> 2. For listen command, should the second parameter be 1 (since we will
> have one outside connection with the server)?

You could.

> 3. About the 8 steps to establish a data connection, which are
> mentioned in the handout. Why we need step 4 in the establishing of the
> connection? Since the port of the socket is already defined in the bind()
> step. And when we read it back using the getsockname(), we will always
> get the same port number that we set in the bind step? So I am not finding
> a use for it.

At step 2, we specify the port number as 0. When bind() is called, the
kernel chooses an ephemeral port. Then we use getsockname() to determine
the assigned port nubmer. The port number we get from getsockname() will
not be 0.
 
> 4. I did all the steps till 7 ( which is sending a ftp CMD over control
> connection). For the PORT command, the server reply was successful.
> However, when I sent the command "LIST \r\n" to start reading the
> directory, what I got is Transient negative reply, with the following
> message from the server:
> 
> Can't build data connection: Connection refused

Please make sure that you are using the right port number.

Best,

Xin

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links
Date: Tue, 15 Jan 2002 16:03:40 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2469

Symbolic links should be handled similarly as regular files. You need to
transfer files until the target directory level. One problem with symbolic
links is that you won't be able to come back up by simply do "CWD". Please
refer to FAQ Q9 for details. Best, Xin


On Tue, 15 Jan 2002, Russell Greene wrote:

> How are we supposed to handle symbolic links to directories?  For example if
> the local dir is /AAA/BBB/ and the remote dir is /CCC/ and a symbolic link
> (localed in the /CCC/ directory) takes us to /ZZZ/ should we create the ZZZ
> directory as a subdirectory of /AAA/BBB/ or create a /ZZZ/ directory
> seperate from /AAA/BBB/?.  As I write this it seems to me that the second
> option is the one that needs to be followed because if /ZZZ/ is in a
> different place on the local machine, the symbolic link won't work after
> being copied.  Any thoughts?   Thanks.
> 
> --Russ
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Adrian Graham" 
Newsgroups: su.class.cs244a
Subject: Re: Enrollment on the EE website
Date: Tue, 15 Jan 2002 18:04:30 -0800
Lines: 29
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: zoostation.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2470

I've tried to register using the link below yesterday and today. However, I
just get a "This service is currently not available" error. Should I just
try later? Or do I need to do something else?

Thanks,

Adrian
"SUL\AIR Laneroom Cluster"  wrote in message


Hello everyone,
Please make sure you enroll to the cs244a class, not the ee384.
the correct url is
https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=cs244a
if you receive a message saying 'you're already subscribed' then everything
is fine.
Regards,
Antonios.
Dhawal Kumar wrote:
From lecture number 2 I came to know that I have to enroll on EE website.
Can I get a link to the same?
I tried
https://audi.stanford.edu/cgi-bin/students.cgi?action=register&cc=ee384
and got an error that my student number is not in the class list.
Do I need to enroll somewhere else?
Thanks,
Dhawal Kumar


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Pete Belknap" 
Newsgroups: su.class.cs244a
Subject: Transferring files you don't have permissions for
Date: Tue, 15 Jan 2002 19:23:16 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2471

The handout says that if we try to transfer a file which we don't have
permissions for, we should just give up and continue from there.  I ran the
ftpd server and got a 550 error for this case.

Is 550 the only error code for permissions violations?  Can we assume that a
550 always means this (i.e. 550 would be the only error code that we don't
terminate on) ?

thanks,
Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: strstr and purify
Date: Tue, 15 Jan 2002 19:53:33 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine31.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2472

I am using the code in the handout to parse a LIST response, and I keep
getting an UMR error on the strstr functions.  Does anyone have an idea
what the problem might be?

Thanks,
Clayton
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links
Date: Tue, 15 Jan 2002 21:10:36 -0800
Lines: 50
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2473

Well, we download 'regular files', and recurse into subdirectories.  And 
you seem to be saying to do both here.  So just to confirm:

* If a symlink is pointing to a regular file, download the file, and 
name it with the name of the symlink.
* If a symlink points to a directory (and assuming we still have some 
depth to cover), go into it searching for files.  If a matching file is 
found, create a directory with the name of the symlink that we'd 
followed and create the file in that directory.

Please confirm or clarify.  Thanks!

-Sandeep

Xin Wang wrote:

> Symbolic links should be handled similarly as regular files. You need to
> transfer files until the target directory level. One problem with symbolic
> links is that you won't be able to come back up by simply do "CWD". Please
> refer to FAQ Q9 for details. Best, Xin
> 
> 
> On Tue, 15 Jan 2002, Russell Greene wrote:
> 
> 
>>How are we supposed to handle symbolic links to directories?  For example if
>>the local dir is /AAA/BBB/ and the remote dir is /CCC/ and a symbolic link
>>(localed in the /CCC/ directory) takes us to /ZZZ/ should we create the ZZZ
>>directory as a subdirectory of /AAA/BBB/ or create a /ZZZ/ directory
>>seperate from /AAA/BBB/?.  As I write this it seems to me that the second
>>option is the one that needs to be followed because if /ZZZ/ is in a
>>different place on the local machine, the symbolic link won't work after
>>being copied.  Any thoughts?   Thanks.
>>
>>--Russ
>>
>>
>>
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Deleting the local tree
Date: Tue, 15 Jan 2002 22:26:55 -0800
Lines: 22
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2474

Are we supposed to delete (recursively) everything in local_dir/* ?  I 
read that we're supposed to overwrite any file that may exist, but I 
have a vague recollection that we are supposed to clean out the local 
directory before downloading files.  Or I'm just confused.

On the one hand, this seems like a good idea for the scenario where you 
want to download file 'a', and a directory by that name exists locally. 
  I'm guessing that doing an 'open' for write will result in an error 
since doing so would involve invalidating a whole directory tree (and 
thus we'd have to delete the whole tree starting at local directory 'a' 
anyway before starting the download).

Thoughts?

-Sandeep
-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links
Date: Wed, 16 Jan 2002 00:07:57 -0800
Lines: 65
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2475

Correct.


On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:

> Well, we download 'regular files', and recurse into subdirectories.  And 
> you seem to be saying to do both here.  So just to confirm:
> 
> * If a symlink is pointing to a regular file, download the file, and 
> name it with the name of the symlink.
> * If a symlink points to a directory (and assuming we still have some 
> depth to cover), go into it searching for files.  If a matching file is 
> found, create a directory with the name of the symlink that we'd 
> followed and create the file in that directory.
> 
> Please confirm or clarify.  Thanks!
> 
> -Sandeep
> 
> Xin Wang wrote:
> 
> > Symbolic links should be handled similarly as regular files. You need to
> > transfer files until the target directory level. One problem with symbolic
> > links is that you won't be able to come back up by simply do "CWD". Please
> > refer to FAQ Q9 for details. Best, Xin
> > 
> > 
> > On Tue, 15 Jan 2002, Russell Greene wrote:
> > 
> > 
> >>How are we supposed to handle symbolic links to directories?  For example if
> >>the local dir is /AAA/BBB/ and the remote dir is /CCC/ and a symbolic link
> >>(localed in the /CCC/ directory) takes us to /ZZZ/ should we create the ZZZ
> >>directory as a subdirectory of /AAA/BBB/ or create a /ZZZ/ directory
> >>seperate from /AAA/BBB/?.  As I write this it seems to me that the second
> >>option is the one that needs to be followed because if /ZZZ/ is in a
> >>different place on the local machine, the symbolic link won't work after
> >>being copied.  Any thoughts?   Thanks.
> >>
> >>--Russ
> >>
> >>
> >>
> >>
> > 
> 
> 
> 
> -- 
> ---------------------------------------------
> Sandeep V. Tamhankar			
> M.S. Student
> Computer Science
> Email: 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Deleting the local tree
Date: Wed, 16 Jan 2002 00:22:15 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2476

You should not delete everything in local_dir/*. Just image any FTP
clients you have used, they copy the files you want to your local
directory. They may overwrite your local file if the file names happen to
be the same, but they will not destroy everthing in your local directory.
Xin


On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:

> Are we supposed to delete (recursively) everything in local_dir/* ?  I 
> read that we're supposed to overwrite any file that may exist, but I 
> have a vague recollection that we are supposed to clean out the local 
> directory before downloading files.  Or I'm just confused.
> 
> On the one hand, this seems like a good idea for the scenario where you 
> want to download file 'a', and a directory by that name exists locally. 
>   I'm guessing that doing an 'open' for write will result in an error 
> since doing so would involve invalidating a whole directory tree (and 
> thus we'd have to delete the whole tree starting at local directory 'a' 
> anyway before starting the download).
> 
> Thoughts?
> 
> -Sandeep
> -- 
> ---------------------------------------------
> Sandeep V. Tamhankar			
> M.S. Student
> Computer Science
> Email: 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Pete Belknap" 
Newsgroups: su.class.cs244a
Subject: Re: strstr and purify
Date: Wed, 16 Jan 2002 00:27:32 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2477

Make sure you don't free the memory allocated by the string you pass to
ftpparse(). I was doing that and getting a bunch of memory errors, including
FMR's and UMR's.

Pete

"Clayton Pierce Jones"  wrote in message

> I am using the code in the handout to parse a LIST response, and I keep
> getting an UMR error on the strstr functions.  Does anyone have an idea
> what the problem might be?
>
> Thanks,
> Clayton


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: File Permission And Parent Directory
Date: Wed, 16 Jan 2002 14:12:31 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2478

Hi,

Imagine that we have a directory with only one file in it. The file has
the same extension that we want to copy. We created the directory on the
client machine, and we tried to copy the file in it. However, we
discovered then that we don't have permission to do so. Should we remove
the created direcotry (As if it is doesn't contain any required file)?

I am facing the above situation since the provided ftpparser doesn't
parse the permission bits. So, the easiest way to check the permission
would be to wait and see if there would be any error in sending the file
through the network. At that time, we may have already created the parent
directory for the file, and then we have to face the above mentioned
problem if we couldn't copy the file.

Thanks,

Moh'd





.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: File Permission And Parent Directory
Date: Wed, 16 Jan 2002 14:19:24 -0800
Lines: 34
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2479

You can solve this by changing the order of things. First issue the RETR
command, see if the return is positive completion, then create the directory
you need if you haven't already, then start the transfer. That is what I did
anyway.

Bret

"Moh'd Saleem Saleem Alrawashdeh"  wrote in message

> Hi,
>
> Imagine that we have a directory with only one file in it. The file has
> the same extension that we want to copy. We created the directory on the
> client machine, and we tried to copy the file in it. However, we
> discovered then that we don't have permission to do so. Should we remove
> the created direcotry (As if it is doesn't contain any required file)?
>
> I am facing the above situation since the provided ftpparser doesn't
> parse the permission bits. So, the easiest way to check the permission
> would be to wait and see if there would be any error in sending the file
> through the network. At that time, we may have already created the parent
> directory for the file, and then we have to face the above mentioned
> problem if we couldn't copy the file.
>
> Thanks,
>
> Moh'd
>
>
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Script output file sizes
Date: Wed, 16 Jan 2002 20:54:42 -0800
Lines: 110
Distribution: su
Message-ID: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2480

First of all, thanks for providing the test scripts, it is great to have
such a simple and thorough way to test our code.  However, I have a couple
questions regarding  "Test F.2".



My test script output includes this:



test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm

exit line said: "OK: 1900222 bytes copied"

---

diff between expected tree and your tree:

1,8d0

< ./kermit/c/all.tar.Z 2606058765 72935

< ./kermit/c/amiga.tar.Z 1017293024 135815

< ./kermit/c/dg.tar.Z 2560791635 406317

< ./kermit/c/mac.tar.Z 2459747744 496551

< ./kermit/c/os2.tar.Z 2771080586 163808

< ./kermit/c/os9.tar.Z 370278796 49438

< ./kermit/c/unix.tar.Z 308104872 349221

< ./kermit/c/vms.tar.Z 53741681 64755



I used ncftp to go to the site to see what was going on, and a "ls -l" in
the kermit/c directory shows this:



ncftp /pub/comm/kermit/c > ls -l

lrwxrwxrwx   1 0        0           37   Apr 14  1995
00README-Legal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs

-r--r--r--   1 0        0          935   Jan 17  1989   README

-r--r--r--   1 0        0        72935   Jan 16  1989   all.tar.Z

-r--r--r--   1 0        0       135815   Jan 16  1989   amiga.tar.Z

-r--r--r--   1 0        0       406317   Jan 16  1989   dg.tar.Z

-r--r--r--   1 0        0       496551   Jan 16  1989   mac.tar.Z

-r--r--r--   1 0        0       163808   Jan 16  1989   os2.tar.Z

-r--r--r--   1 0        0        49438   Jan 16  1989   os9.tar.Z

-r--r--r--   1 0        0       349221   Jan 16  1989   unix.tar.Z

-r--r--r--   1 0        0        64755   Jan 16  1989   vms.tar.Z



It looks like the files sizes are the ones I got, not the large ones from
the script.  Am I mistaken here? The same thing happened on the MS FTP site
(snippets below).



test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie exit line
said: "OK: 5113 bytes copied"

---

diff between expected tree and your tree:

3,4d2

< ./ie-public/fixes/usa/ie401/importexportfavorites-fix/readme.txt
2939125283 643 <
../ie-public/fixes/usa/ie50/importexportfavorites-fix/readme.txt 2939125283
643



ncftp ...ortexportfavorites-fix > ls -l

dr-xr-xr-x   1 owner    group            0 Feb 25  2000 alpha

-r-xr-xr-x   1 owner    group          643 Sep 24  1999 readme.txt

dr-xr-xr-x   1 owner    group            0 Feb 25  2000 x86





Thanks for your time,

Bret





.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Script file sizes (normal line endings this time)
Date: Wed, 16 Jan 2002 20:56:01 -0800
Lines: 65
Distribution: su
Message-ID: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2481

I noticed my first post had some weird carriage returns, so I am reposting
to make it easier to read.  Sorry for the trouble.

First of all, thanks for providing the test scripts, it is great to have
such a simple and thorough way to test our code.  However, I have a couple
questions regarding  "Test F.2".

My test script output includes this:

test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: "OK: 1900222 bytes copied"
---
diff between expected tree and your tree:
1,8d0
< ./kermit/c/all.tar.Z 2606058765 72935
< ./kermit/c/amiga.tar.Z 1017293024 135815
< ./kermit/c/dg.tar.Z 2560791635 406317
< ./kermit/c/mac.tar.Z 2459747744 496551
< ./kermit/c/os2.tar.Z 2771080586 163808
< ./kermit/c/os9.tar.Z 370278796 49438
< ./kermit/c/unix.tar.Z 308104872 349221
< ./kermit/c/vms.tar.Z 53741681 64755

I used ncftp to go to the site to see what was going on, and a "ls -l" in
the kermit/c directory shows this:

ncftp /pub/comm/kermit/c > ls -l
lrwxrwxrwx   1 0        0           37   Apr 14  1995
00README-Legal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs
-r--r--r--   1 0        0          935   Jan 17  1989   README
-r--r--r--   1 0        0        72935   Jan 16  1989   all.tar.Z
-r--r--r--   1 0        0       135815   Jan 16  1989   amiga.tar.Z
-r--r--r--   1 0        0       406317   Jan 16  1989   dg.tar.Z
-r--r--r--   1 0        0       496551   Jan 16  1989   mac.tar.Z
-r--r--r--   1 0        0       163808   Jan 16  1989   os2.tar.Z
-r--r--r--   1 0        0        49438   Jan 16  1989   os9.tar.Z
-r--r--r--   1 0        0       349221   Jan 16  1989   unix.tar.Z
-r--r--r--   1 0        0        64755   Jan 16  1989   vms.tar.Z

It looks like the files sizes are the ones I got, not the large ones from
the script.  Am I mistaken here? The same thing happened on the MS FTP site
(snippets below).

test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie exit line
said: "OK: 5113 bytes copied"
---
diff between expected tree and your tree:
3,4d2
< ./ie-public/fixes/usa/ie401/importexportfavorites-fix/readme.txt
2939125283 643 <
../ie-public/fixes/usa/ie50/importexportfavorites-fix/readme.txt 2939125283
643

ncftp ...ortexportfavorites-fix > ls -l
dr-xr-xr-x   1 owner    group            0 Feb 25  2000 alpha
-r-xr-xr-x   1 owner    group          643 Sep 24  1999 readme.txt
dr-xr-xr-x   1 owner    group            0 Feb 25  2000 x86


Thanks for your time,
Bret




.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: Script file sizes (normal line endings this time)
Date: Wed, 16 Jan 2002 21:45:10 -0800
Lines: 74
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2482

Nevermind, it woked this time, it must have been my numbers that were wrong.

Bret

"Bret Taylor"  wrote in message

> I noticed my first post had some weird carriage returns, so I am reposting
> to make it easier to read.  Sorry for the trouble.
>
> First of all, thanks for providing the test scripts, it is great to have
> such a simple and thorough way to test our code.  However, I have a couple
> questions regarding  "Test F.2".
>
> My test script output includes this:
>
> test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> exit line said: "OK: 1900222 bytes copied"
> ---
> diff between expected tree and your tree:
> 1,8d0
> < ./kermit/c/all.tar.Z 2606058765 72935
> < ./kermit/c/amiga.tar.Z 1017293024 135815
> < ./kermit/c/dg.tar.Z 2560791635 406317
> < ./kermit/c/mac.tar.Z 2459747744 496551
> < ./kermit/c/os2.tar.Z 2771080586 163808
> < ./kermit/c/os9.tar.Z 370278796 49438
> < ./kermit/c/unix.tar.Z 308104872 349221
> < ./kermit/c/vms.tar.Z 53741681 64755
>
> I used ncftp to go to the site to see what was going on, and a "ls -l" in
> the kermit/c directory shows this:
>
> ncftp /pub/comm/kermit/c > ls -l
> lrwxrwxrwx   1 0        0           37   Apr 14  1995
> 00README-Legal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs
> -r--r--r--   1 0        0          935   Jan 17  1989   README
> -r--r--r--   1 0        0        72935   Jan 16  1989   all.tar.Z
> -r--r--r--   1 0        0       135815   Jan 16  1989   amiga.tar.Z
> -r--r--r--   1 0        0       406317   Jan 16  1989   dg.tar.Z
> -r--r--r--   1 0        0       496551   Jan 16  1989   mac.tar.Z
> -r--r--r--   1 0        0       163808   Jan 16  1989   os2.tar.Z
> -r--r--r--   1 0        0        49438   Jan 16  1989   os9.tar.Z
> -r--r--r--   1 0        0       349221   Jan 16  1989   unix.tar.Z
> -r--r--r--   1 0        0        64755   Jan 16  1989   vms.tar.Z
>
> It looks like the files sizes are the ones I got, not the large ones from
> the script.  Am I mistaken here? The same thing happened on the MS FTP
site
> (snippets below).
>
> test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie exit line
> said: "OK: 5113 bytes copied"
> ---
> diff between expected tree and your tree:
> 3,4d2
> < ./ie-public/fixes/usa/ie401/importexportfavorites-fix/readme.txt
> 2939125283 643 <
> ./ie-public/fixes/usa/ie50/importexportfavorites-fix/readme.txt 2939125283
> 643
>
> ncftp ...ortexportfavorites-fix > ls -l
> dr-xr-xr-x   1 owner    group            0 Feb 25  2000 alpha
> -r-xr-xr-x   1 owner    group          643 Sep 24  1999 readme.txt
> dr-xr-xr-x   1 owner    group            0 Feb 25  2000 x86
>
>
> Thanks for your time,
> Bret
>
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: "Potential Leak"
Date: Thu, 17 Jan 2002 00:26:29 -0800
Lines: 37
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2483

I just ran my code through purify (in text-mode so that I can submit a 
purify.output file to the test script), and I get some conflicting info 
near the bottom of the output:


Purify: Searching for all memory leaks...

Memory leaked: 0 bytes (0%); potentially leaked: 0 bytes (0%)

Purify Heap Analysis (combining suppressed and unsuppressed blocks)
                          Blocks        Bytes
               Leaked          0            0
   Potentially Leaked          1         8200
               In-Use        135         5433
   ----------------------------------------
      Total Allocated        136        13633

<snip>

   * 0 bytes leaked.
   * 0 bytes potentially leaked.


So twice it tells me I have 0 bytes of potential mem leak, and once it 
says I have 1 block / 8200 bytes of potential mem leak.  And there is no 
other warning/error anywhere else in the report indicating where this 
"potential" memleak could be.  Can I ignore this, or does this have some 
meaning?

-Sandeep
-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Garrett Smith 
Newsgroups: su.class.cs244a
Subject: snprintf
Date: Thu, 17 Jan 2002 10:32:41 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: ubik.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20020105
X-Accept-Language: en-us, fr
Xref: nntp.stanford.edu su.class.cs244a:2484

Is it ok to use snprintf()?  It is in the ANSI C99 spec, but doesn't get 
a prototype from the headers if -ansi is passed to gcc.

Garrett

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga8.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: FTP server replies
Date: Thu, 17 Jan 2002 19:27:46 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2485


  Hi everybody,

  I have a question, when we send a command on the control connection to
the FTP server (for example, when we want to login), there is not known
that how many lines the server is going to send us back. As I noticed, not
allthe servers hold the regulations with multiple line replies according
to the FTP specification. (I mean using '-'). Also, for example, when you
send the USER command to ftp.cs.stanford edu, it replies you with two separate
lines, one starting with code 220 and the other with code 331 (after
sending USER command). So, how are we supposed to read all the lines from socket before
sending next command?

  Yours,
  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Creating directories...
Date: Thu, 17 Jan 2002 19:32:16 -0800
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2486

Hi everyone,

to create directories in C you can use the mkdir() call. The
man page is accessible with:

  man -s 2 mkdir

along the same lines the man pages for the standard C library
can be accessed with:

  man -s 3c printf

A simple 'man printf' won't work as there also is a shell 
command by the name of printf.

  Guido

PS: I post this because man -k/apropos is currently broken on 
    the cluster machines.

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Darren Lewis" 
Newsgroups: su.class.cs244a
Subject: ftp.microsoft.com
Date: Thu, 17 Jan 2002 19:50:46 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: darren.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2487

Does anyone else have problems with ftp.microsoft.com? My sockets appear to
time out, sometimes getting farther along than other times?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Yichen Xie 
Newsgroups: su.class.cs244a
Subject: Re: FTP server replies
Date: Thu, 17 Jan 2002 19:55:27 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dn800cb149.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.5) Gecko/20011014
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2488

The 220 line you get is most likely:

	220 ftp.cs.stanford.edu NcFTPd Server (free educational license) ready.

This is sent upon connection, before you send in the "USER" command.

Amin Firoozshahian wrote:

>   Hi everybody,
> 
>   I have a question, when we send a command on the control connection to
> the FTP server (for example, when we want to login), there is not known
> that how many lines the server is going to send us back. As I noticed, not
> allthe servers hold the regulations with multiple line replies according
> to the FTP specification. (I mean using '-'). Also, for example, when you
> send the USER command to ftp.cs.stanford edu, it replies you with two separate
> lines, one starting with code 220 and the other with code 331 (after
> sending USER command). So, how are we supposed to read all the lines from socket before
> sending next command?
> 
>   Yours,
>   Amin
> 
> 


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: More Qs about symbolic links
Date: Thu, 17 Jan 2002 21:30:30 -0800
Lines: 34
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2489

Hi,

  I can think of some cases that would be extremely difficult to
parse given a symbolic link.  For example, ftpparse may return
a filename:

Star-Trek -> star-trek (this is taken from ftp.faqs.org)

If we know that is a symbolic link, thats pretty straightforward. Now
what about the following perfectly legal name?

goo -> ber -> sam -> wise

The possibilities are:

goo               is a symbolic link to file "ber -> sam -> wise"
goo -> ber        is a symbolic link to file "sam -> wise"
goo -> ber -> sam is a symbolic link to file "wise"

I cannot think of a method to differentiate between symbolic association and
the filename.  Here's an actual reply I contrived.

jim -> bob -> McGee -> ../../../vimbuddy.vim

which is a file "jim -> bob -> McGee" as a symbolic link to ../../../vimbuddy.vim


Are there any assumptions we can make about the format of symbolic links? Or do we
have to be prepared to use a trial-by-error scheme in case we run across
one of these special cases.  Any help would be greatly appreciated. Thanks.


                ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Symbol To Symbol
Date: Thu, 17 Jan 2002 21:47:54 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2490

Hi,
Should we handle the cases of symbol pointing to another symbol, which can
then points to a file or directory. In other words, should we handle more
than one level of indirection?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: "Potential Leak"
Date: 18 Jan 2002 07:29:42 GMT
Lines: 47
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2491


This is fine. The potential leak noted below may be caused by a 
redirection of output to a file. All we will be requiring for this 
assignment is that the line "Memory leaked: w bytes (x%); potentially 
leaked: y bytes (z%)" has all four of w, x, y, z equal to 0. Of course, 
there should also be no UMRs, etc.

Arun

Sandeep Tamhankar  wrote:
: I just ran my code through purify (in text-mode so that I can submit a 
: purify.output file to the test script), and I get some conflicting info 
: near the bottom of the output:


: Purify: Searching for all memory leaks...

: Memory leaked: 0 bytes (0%); potentially leaked: 0 bytes (0%)

: Purify Heap Analysis (combining suppressed and unsuppressed blocks)
:                           Blocks        Bytes
:                Leaked          0            0
:    Potentially Leaked          1         8200
:                In-Use        135         5433
:    ----------------------------------------
:       Total Allocated        136        13633

: <snip>

:    * 0 bytes leaked.
:    * 0 bytes potentially leaked.


: So twice it tells me I have 0 bytes of potential mem leak, and once it 
: says I have 1 block / 8200 bytes of potential mem leak.  And there is no 
: other warning/error anywhere else in the report indicating where this 
: "potential" memleak could be.  Can I ignore this, or does this have some 
: meaning?

: -Sandeep
: -- 
: ---------------------------------------------
: Sandeep V. Tamhankar			
: M.S. Student
: Computer Science
: Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: snprintf
Date: 18 Jan 2002 07:38:06 GMT
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2492

We'd prefer it, if for consistency sake, that the program compiles without
warning with the -ansi flag.

In this case, the function is linked in so it is present in the library. 
To avoid the warning add the declaration:

extern int snprintf(char *s,  size_t  n,  const  char  *format,  /* args*/ 
....);

and everything should compile and link without event.

Arun

Garrett Smith  wrote:
: Is it ok to use snprintf()?  It is in the ANSI C99 spec, but doesn't get 
: a prototype from the headers if -ansi is passed to gcc.

: Garrett

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: FTP server replies
Date: 18 Jan 2002 07:46:43 GMT
Lines: 41
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2493

The best way to know how to deal with the various responses to commands is
to check the RFC, which describes the possible FTP responses, including
any intermediate replies, to a given command.

When reading the data, you need to be careful to detect the end of a
response, and not issue another read, as the read will block until data is
available. One way to do this is to read byte-byte and look for the end of
response character sequence as you do the reads. There are higher
performance ways to do this (i.e., reading blocks at a time), but one has
to be careful as portions of the following response may be read in along
with the end of reply sequence.

Arun

 Yichen Xie  wrote:
: The 220 line you get is most likely:

: 	220 ftp.cs.stanford.edu NcFTPd Server (free educational license) ready.

: This is sent upon connection, before you send in the "USER" command.

: Amin Firoozshahian wrote:

:>   Hi everybody,
:> 
:>   I have a question, when we send a command on the control connection to
:> the FTP server (for example, when we want to login), there is not known
:> that how many lines the server is going to send us back. As I noticed, not
:> allthe servers hold the regulations with multiple line replies according
:> to the FTP specification. (I mean using '-'). Also, for example, when you
:> send the USER command to ftp.cs.stanford edu, it replies you with two separate
:> lines, one starting with code 220 and the other with code 331 (after
:> sending USER command). So, how are we supposed to read all the lines from socket before
:> sending next command?
:> 
:>   Yours,
:>   Amin
:> 
:> 


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: More Qs about symbolic links
Date: 18 Jan 2002 07:51:01 GMT
Lines: 45
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2494

We understand that, since the list response is non standardized, there 
will be numerous special cases such as this. In general, whatever name 
returned to you by ftpparse can be assumed to be the link name (even if it 
contains a ->). We will try not to test against such ambiguous cases as 
you have mentioned; the servers you are required to work with are listed 
in the assignment.

Arun


Martin Casado  wrote:
: Hi,

:   I can think of some cases that would be extremely difficult to
: parse given a symbolic link.  For example, ftpparse may return
: a filename:

: Star-Trek -> star-trek (this is taken from ftp.faqs.org)

: If we know that is a symbolic link, thats pretty straightforward. Now
: what about the following perfectly legal name?

: goo -> ber -> sam -> wise

: The possibilities are:

: goo               is a symbolic link to file "ber -> sam -> wise"
: goo -> ber        is a symbolic link to file "sam -> wise"
: goo -> ber -> sam is a symbolic link to file "wise"

: I cannot think of a method to differentiate between symbolic association and
: the filename.  Here's an actual reply I contrived.

: jim -> bob -> McGee -> ../../../vimbuddy.vim

: which is a file "jim -> bob -> McGee" as a symbolic link to ../../../vimbuddy.vim


: Are there any assumptions we can make about the format of symbolic links? Or do we
: have to be prepared to use a trial-by-error scheme in case we run across
: one of these special cases.  Any help would be greatly appreciated. Thanks.


:                 ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Xiaoyan Cheng 
Newsgroups: su.class.cs244a
Subject: server close data connection
Date: Fri, 18 Jan 2002 00:04:52 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2495

"The server will automatically close the connection from its end once
all of the data has been sent."

Does this imply that every time when we send a LIST or RETR command over
the control connection, we need to re-establish a data
Connection(starting from step 1, client creates another socket) ...
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Symbol To Symbol
Date: 18 Jan 2002 08:14:17 GMT
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2496

You do not need to worry about doing anything special to handle this case. 
We will avoid testing your code using ambiguous or otherwise exotic cases.

Arun

Moh'd Saleem Saleem Alrawashdeh  wrote:

: Hi,
: Should we handle the cases of symbol pointing to another symbol, which can
: then points to a file or directory. In other words, should we handle more
: than one level of indirection?

: Thanks,

: Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: server close data connection
Date: 18 Jan 2002 08:15:28 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2497

Yes, a new data connection needs to opened for each file/LIST response
(this includes issuing a PORT command etc.)

Arun

Xiaoyan Cheng  wrote:

: "The server will automatically close the connection from its end once
: all of the data has been sent."

: Does this imply that every time when we send a LIST or RETR command over
: the control connection, we need to re-establish a data
: Connection(starting from step 1, client creates another socket) ...
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: FTP server replies
Date: Fri, 18 Jan 2002 00:17:21 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2498

Along the same lines...

After writing a command to the server,

Should we poll the socket for replies as in the echodemo example?

or just read from it once, and only keep reading again if the reply code
indicates that we should wait for another reply before sending next
command?


Amin Firoozshahian wrote:
> 
>   Hi everybody,
> 
>   I have a question, when we send a command on the control connection to
> the FTP server (for example, when we want to login), there is not known
> that how many lines the server is going to send us back. As I noticed, not
> allthe servers hold the regulations with multiple line replies according
> to the FTP specification. (I mean using '-'). Also, for example, when you
> send the USER command to ftp.cs.stanford edu, it replies you with two separate
> lines, one starting with code 220 and the other with code 331 (after
> sending USER command). So, how are we supposed to read all the lines from socket before
> sending next command?
> 
>   Yours,
>   Amin

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: FTP server replies
Date: 18 Jan 2002 09:01:16 GMT
Lines: 38
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2499

Use the strategy I mentioned in the previous posting to monitor the 
response from the server over the control connection. If you have any 
other questions, feel free to ask.

Arun

Luis Robles  wrote:
: Along the same lines...

: After writing a command to the server,

: Should we poll the socket for replies as in the echodemo example?

: or just read from it once, and only keep reading again if the reply code
: indicates that we should wait for another reply before sending next
: command?


: Amin Firoozshahian wrote:
:> 
:>   Hi everybody,
:> 
:>   I have a question, when we send a command on the control connection to
:> the FTP server (for example, when we want to login), there is not known
:> that how many lines the server is going to send us back. As I noticed, not
:> allthe servers hold the regulations with multiple line replies according
:> to the FTP specification. (I mean using '-'). Also, for example, when you
:> send the USER command to ftp.cs.stanford edu, it replies you with two separate
:> lines, one starting with code 220 and the other with code 331 (after
:> sending USER command). So, how are we supposed to read all the lines from socket before
:> sending next command?
:> 
:>   Yours,
:>   Amin

: -- 
:        Luis Robles
: 
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: UMR in gethostbyname
Date: 18 Jan 2002 10:29:31 GMT
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2500

When I pass "ftp.fedworld.gov" into gethostbyname(...),
purify reports a UMR in the library code.

Believe it or not, it only happens with that host. I've checked
and made sure that the string is null terminated.

Does anyone else have this problem, and can we ignore UMR's
reported in the standard library code?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Passing a number in as the hostname
Date: 18 Jan 2002 10:32:37 GMT
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2501

If any number is passed in as the hostname, for example:

ftpcopy exe 0 1323535

inet_addr( ... ) successfully "parses" the string "1323535" as 
the ip address, and when connect is called with the resulting
address, the program just blocks.

Should we or can we fix this behavior?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Eric Watkins 
Newsgroups: su.class.cs244a
Subject: Re: server close data connection
Date: Fri, 18 Jan 2002 02:40:16 -0800
Lines: 10
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: e-man.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2502

> Yes, a new data connection needs to opened for each file/LIST response
> (this includes issuing a PORT command etc.)

But can we (should we) skip steps 1 and 3?  Those seem like things that
should only be done once.  After you have that first socket set up to
receive the incoming connections, then you need to repeat all the other
steps, including accept(), which will get you the new data socket.  Is
this right?

-Eric
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: ftp.microsoft.com--should we consider this
Date: Fri, 18 Jan 2002 12:55:01 -0800
Lines: 37
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2503

when i run my client program for ftp.microsoft.com i get a response code
of "421" for the USER command itself. note that the rfc says:
A reply is defined to contain the 3-digit code, followed by Space
<SP>, followed by one line of text (where some maximum line length
      has been specified), and terminated by the Telnet end-of-line
      code.
(multiline reply will have hyphen after the code on the first line).

but the code returned by microsoft has some other character (apart from
'-' and ' ') after the 3 digits.

anyone else notice this? should we worry about such cases too?

thanks,
ashmi


On Thu, 17 Jan 2002, Darren Lewis wrote:

> Does anyone else have problems with ftp.microsoft.com? My sockets appear to
> time out, sometimes getting farther along than other times?
>
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Yichen Xie" 
Newsgroups: su.class.cs244a
Subject: Re: Passing a number in as the hostname
Date: Fri, 18 Jan 2002 13:17:40 -0800
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dnab4046be.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2504

I found it easier to use only gethostbyname(), since it can handle both
dotted decimal addresses and normal names. 1323535 should translate to
0.20.50.15, which apparently isn't up on the network. That might explain why
it blocks (and eventually time out), just as what you would expect when you
try to access some website that is down.

"Derrick Wen-Shiuan Tong"  wrote in message

> If any number is passed in as the hostname, for example:
>
> ftpcopy exe 0 1323535
>
> inet_addr( ... ) successfully "parses" the string "1323535" as
> the ip address, and when connect is called with the resulting
> address, the program just blocks.
>
> Should we or can we fix this behavior?
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: recursively descending into subdir
Date: Fri, 18 Jan 2002 13:42:01 -0800
Lines: 32
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2505

hi all,
my program hangs as soon as i try to recusively descend into, and list,
the contents of the first subdirectory in the default remote dir.
as far as i can see, i havent done anything silly in the recursion because
the following things work:

-cwd into that dir (say D1) returns 250 ---works
-data connection is set up
-LIST returns 226 (is it necessary this returns 250?)
-on reading the contents returned by LIST though, i get nothing, but this
is not what is causing the program to hang...
-after this since i CWD'd into this directory D1, i try to cwd out of it
by "CWD .."
and <kaboom>.

any ideas?
thanks,
ashmi


###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!rxuan
From: Ruozhong Xuan 
Newsgroups: su.class.cs244a
Subject: submit_test.pl result question
Date: Fri, 18 Jan 2002 14:07:45 -0800
Lines: 38
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2506

Hi,
    I used the submit_test.pl to request a test, and in a section,I got
the below:
===============

    * Test F.2: Does not transfer all of the matching files in the first
level down to the
      last specified level [out of 1]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: "OK:1900222 bytes copied"
---
diff between expected tree and your tree:
1,8d0
< ./kermit/c/all.tar.Z
< ./kermit/c/amiga.tar.Z
< ./kermit/c/dg.tar.Z
< ./kermit/c/mac.tar.Z
< ./kermit/c/os2.tar.Z
< ./kermit/c/os9.tar.Z
< ./kermit/c/unix.tar.Z
< ./kermit/c/vms.tar.Z
10d1
< ./xcomm.tar.Z



But when I manually tried out:
ftpcopy Z 2 204.123.2.2 test /pub/comm
I found all those files have been transfered to "test" directory under the
correct subdirectory. So I am a little confused by the test result. Any
comment?

Thanks.
Ruozhong Xuan


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!rgustin
From: Reid Gustin 
Newsgroups: su.class.cs244a
Subject: Re: recursively descending into subdir
Date: Fri, 18 Jan 2002 14:32:42 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2507

On Fri, 18 Jan 2002, Ashmi wrote:
> -cwd into that dir (say D1) returns 250 ---works
> -data connection is set up
> -LIST returns 226 (is it necessary this returns 250?)
> -on reading the contents returned by LIST though, i get nothing, but this
> is not what is causing the program to hang...
> -after this since i CWD'd into this directory D1, i try to cwd out of it
> by "CWD .."
> and <kaboom>.

I don't know if this is actually the problem you're seeing, but if D1 is
actually a symbolic link to someplace completely different, you could run
into trouble with the 'CWD ..'.  Take the example where D1 points to a
directory, the parent of which you don't have permissions for.  You try
'CWD ..' and it would say you couldn't do that.  It's also a little bit
dangerous, as 'CWD ..' could take you *anywhere* in the system, not
necesarily back to the tree where you were transferring files, in the
case of symbolic links.

As for what LIST can return, it's legal (from the RFC) for it to return
any of: 125, 150, 226, 250, 421, 425, 426, 451, 450, 500, 501, 502, 530.
There's a spot in the RFC under the heading 'Command-Reply Sequences' that
talks about the possible responses to a given command.

Reid

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Reply Buffer over Control and Data connection...
Date: Fri, 18 Jan 2002 14:32:50 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine37.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2508

Hi,

I have two questions regarding receiving the response in replyBuffers over
the control connection and the data connection.

1) Over the control connection, it seems that the max buffer is 1024
bytes. When I use a buffer theis size. I do get the reply back. However,
the status: 1xx, 2xx, 3xx, 4xx, 5xx, is not necessarily the first
information in the buffer. Is that expected?
If yes, then how would I know what the server's reply is?

2) For data connection the server's reply could be very big. But would the
server still reply in packets of 1024 bytes?
I mean that the buffer that we pass to read(s, ...) should be 1024 bytes
and then we need to just keep reading such buffers and then concatenate
the data packets?
Is that true? Please explain.

Would really appreciate your help.

Thanks,
PriyankaA

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!rgustin
From: Reid Gustin 
Newsgroups: su.class.cs244a
Subject: Re: ftp.microsoft.com--should we consider this
Date: Fri, 18 Jan 2002 14:36:03 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2509

On Fri, 18 Jan 2002, Ashmi wrote:
> when i run my client program for ftp.microsoft.com i get a response code
> of "421" for the USER command itself. note that the rfc says:
> A reply is defined to contain the 3-digit code, followed by Space
> <SP>, followed by one line of text (where some maximum line length
>       has been specified), and terminated by the Telnet end-of-line
>       code.
> (multiline reply will have hyphen after the code on the first line).
>
> but the code returned by microsoft has some other character (apart from
> '-' and ' ') after the 3 digits.
>
> anyone else notice this? should we worry about such cases too?

I think that in the case of ftp.microsoft.com, if you're seeing this in
the same spot that I was, this means you can't log on (I got it when the
server was full).  I ended up changing my response parser a little bit to
allow for this, but in a way that didn't break the rest of it.  My problem
with their response was that it wasn't properly terminated (or didn't
appear to be), which caused all kinds of parsing problems until I allowed
for it.

Reid

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!rgustin
From: Reid Gustin 
Newsgroups: su.class.cs244a
Subject: Re: Reply Buffer over Control and Data connection...
Date: Fri, 18 Jan 2002 14:40:12 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2510

On Fri, 18 Jan 2002, Priyanka Gupta wrote:
> 1) Over the control connection, it seems that the max buffer is 1024
> bytes. When I use a buffer theis size. I do get the reply back. However,
> the status: 1xx, 2xx, 3xx, 4xx, 5xx, is not necessarily the first
> information in the buffer. Is that expected?
> If yes, then how would I know what the server's reply is?

The status code is defined in the RFC to be the first thing that comes
back over the socket, so my guess is that you're seeing the tail end of a
previous response, and then the new one.  Just a guess, though.

> 2) For data connection the server's reply could be very big. But would the
> server still reply in packets of 1024 bytes?
> I mean that the buffer that we pass to read(s, ...) should be 1024 bytes
> and then we need to just keep reading such buffers and then concatenate
> the data packets?
> Is that true? Please explain.

The data connections are different, in that they can send arbitrary
amounts of data at a time (like files).  You'll notice in the assignment,
it says, "In this assignment you can assume that no NVT string will be
longer than 1024 bytes."  Which means only NVT strings, like the control
connection, and not the data connection.  The server can send you whatever
it wants over the data connection, in arbitrary sized chunks, with no
requirement for newlines at all.

Reid

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Reply Buffer over Control and Data connection...
Date: Fri, 18 Jan 2002 18:02:39 -0800
Lines: 59
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2511

> it says, "In this assignment you can assume that no NVT string will be
> longer than 1024 bytes."  Which means only NVT strings, like the control
> connection, and not the data connection.

so can we assume that all of the following will be no more than 1024
bytes:
1. every single line reply over ctrl connection
2. every line in the response from LIST (over data connection, but the
response of LIST is in NVT ASCII)


thanks,
~ashmi


On Fri, 18 Jan 2002, Reid Gustin wrote:

> On Fri, 18 Jan 2002, Priyanka Gupta wrote:
> > 1) Over the control connection, it seems that the max buffer is 1024
> > bytes. When I use a buffer theis size. I do get the reply back. However,
> > the status: 1xx, 2xx, 3xx, 4xx, 5xx, is not necessarily the first
> > information in the buffer. Is that expected?
> > If yes, then how would I know what the server's reply is?
>
> The status code is defined in the RFC to be the first thing that comes
> back over the socket, so my guess is that you're seeing the tail end of a
> previous response, and then the new one.  Just a guess, though.
>
> > 2) For data connection the server's reply could be very big. But would the
> > server still reply in packets of 1024 bytes?
> > I mean that the buffer that we pass to read(s, ...) should be 1024 bytes
> > and then we need to just keep reading such buffers and then concatenate
> > the data packets?
> > Is that true? Please explain.
>
> The data connections are different, in that they can send arbitrary
> amounts of data at a time (like files).  You'll notice in the assignment,
> it says, "In this assignment you can assume that no NVT string will be
> longer than 1024 bytes."  Which means only NVT strings, like the control
> connection, and not the data connection.  The server can send you whatever
> it wants over the data connection, in arbitrary sized chunks, with no
> requirement for newlines at all.
>
> Reid
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!epic27.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Re: Reply Buffer over Control and Data connection...
Date: Fri, 18 Jan 2002 19:21:04 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Reid Gustin 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2512

On Fri, 18 Jan 2002, Reid Gustin wrote:

> On Fri, 18 Jan 2002, Priyanka Gupta wrote:
> > 1) Over the control connection, it seems that the max buffer is 1024
> > bytes. When I use a buffer theis size. I do get the reply back. However,
> > the status: 1xx, 2xx, 3xx, 4xx, 5xx, is not necessarily the first
> > information in the buffer. Is that expected?
> > If yes, then how would I know what the server's reply is?
>
> The status code is defined in the RFC to be the first thing that comes
> back over the socket, so my guess is that you're seeing the tail end of a
> previous response, and then the new one.  Just a guess, though.

Just wanted to know if someone else is seeing this too. If its just me
then I might have some problem. In such a case I don't know what the
status code is.

 >
> > 2) For data connection the server's reply could be very big. But would the
> > server still reply in packets of 1024 bytes?
> > I mean that the buffer that we pass to read(s, ...) should be 1024 bytes
> > and then we need to just keep reading such buffers and then concatenate
> > the data packets?
> > Is that true? Please explain.
>
> The data connections are different, in that they can send arbitrary
> amounts of data at a time (like files).  You'll notice in the assignment,
> it says, "In this assignment you can assume that no NVT string will be
> longer than 1024 bytes."  Which means only NVT strings, like the control
> connection, and not the data connection.  The server can send you whatever
> it wants over the data connection, in arbitrary sized chunks, with no
> requirement for newlines at all.

I guess my question is that in that case how do we specify the size of the
reply buffer than we are expecting back?

>
> Reid
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic27.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Reply 4xx
Date: Fri, 18 Jan 2002 19:50:20 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2513


Hi,

I wanted to know that if we get a transient negative completion reply,
which means we could retry later. Should we retry again or can we just
treat 4xx and 5xx the same way?

Thanks,
Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: recursively descending into subdir
Date: Fri, 18 Jan 2002 19:53:20 -0800
Lines: 46
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2514

thanks reid, for pointing out the cwd .. thing. that wasnt my problem but
i do need to take care of that too.
ashmi

On Fri, 18 Jan 2002, Reid Gustin wrote:

> On Fri, 18 Jan 2002, Ashmi wrote:
> > -cwd into that dir (say D1) returns 250 ---works
> > -data connection is set up
> > -LIST returns 226 (is it necessary this returns 250?)
> > -on reading the contents returned by LIST though, i get nothing, but this
> > is not what is causing the program to hang...
> > -after this since i CWD'd into this directory D1, i try to cwd out of it
> > by "CWD .."
> > and <kaboom>.
>
> I don't know if this is actually the problem you're seeing, but if D1 is
> actually a symbolic link to someplace completely different, you could run
> into trouble with the 'CWD ..'.  Take the example where D1 points to a
> directory, the parent of which you don't have permissions for.  You try
> 'CWD ..' and it would say you couldn't do that.  It's also a little bit
> dangerous, as 'CWD ..' could take you *anywhere* in the system, not
> necesarily back to the tree where you were transferring files, in the
> case of symbolic links.
>
> As for what LIST can return, it's legal (from the RFC) for it to return
> any of: 125, 150, 226, 250, 421, 425, 426, 451, 450, 500, 501, 502, 530.
> There's a spot in the RFC under the heading 'Command-Reply Sequences' that
> talks about the possible responses to a given command.
>
> Reid
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: failure @ getsockname
Date: Fri, 18 Jan 2002 20:12:14 -0800
Lines: 29
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2515

I'm following the steps indicated in the assignment handout,
for establishing the data connection,

1.)creating a socket

zeroing out a struct sockaddr_in struct
getting the host name via getenv("HOST") - (returns saga3.stanford.edu)
getting the host address via gethostbyname
copying the host_addr field into the struct sockaddr_in


At this point the struct sockaddr_in looks good, so I call
2.)bind w/ struct sockaddr_in's sin_port still = 0 from zeroing it out
earlier
3.)calling listen
4.)and then calling getsockname to determine the port number

But getsockname fails on me w/ the following args:
 
getsockname( dataId, (struct sockaddr *)clientAddr, sizeof(*clientAddr)
)

Should the 3rd arg be the size of the struct sockaddr_in as in the calls
to bind, or is it sizeof(len) where len is socklen_t as on pg 109?

Is there another reason why getsockname would fail,
or is there something faulty before my call to getsockname?

Cheers,
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Wong 
Newsgroups: su.class.cs244a
Subject: ANSI C library functions
Date: 19 Jan 2002 04:25:33 GMT
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable7.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2516

Hi,

I'm having problems determining whether certain functions that I'm using are in the ANSI C spec. Is there a place that contains a list of the ANSI C library functions? For example, I was under the impression that bzero wasn't in the ANSI C spec - memset should be used instead. However, the given example uses bzero so I'm confused as to what functions are allowed and which ones are not.

Thanks,
Victor.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: Reply Buffer over Control and Data connection...
Date: Fri, 18 Jan 2002 21:07:42 -0800
Lines: 54
Distribution: su
Message-ID: 
References: 
	  
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2517

P -

You dont know the "size" of the line you are expecting back,
so you need to loop and read 1 char at a time until you
hit the end of line char...

Take a look @ the FTP Server Replies post earlier

Priyanka Gupta wrote:
> 
> On Fri, 18 Jan 2002, Reid Gustin wrote:
> 
> > On Fri, 18 Jan 2002, Priyanka Gupta wrote:
> > > 1) Over the control connection, it seems that the max buffer is 1024
> > > bytes. When I use a buffer theis size. I do get the reply back. However,
> > > the status: 1xx, 2xx, 3xx, 4xx, 5xx, is not necessarily the first
> > > information in the buffer. Is that expected?
> > > If yes, then how would I know what the server's reply is?
> >
> > The status code is defined in the RFC to be the first thing that comes
> > back over the socket, so my guess is that you're seeing the tail end of a
> > previous response, and then the new one.  Just a guess, though.
> 
> Just wanted to know if someone else is seeing this too. If its just me
> then I might have some problem. In such a case I don't know what the
> status code is.
> 
>  >
> > > 2) For data connection the server's reply could be very big. But would the
> > > server still reply in packets of 1024 bytes?
> > > I mean that the buffer that we pass to read(s, ...) should be 1024 bytes
> > > and then we need to just keep reading such buffers and then concatenate
> > > the data packets?
> > > Is that true? Please explain.
> >
> > The data connections are different, in that they can send arbitrary
> > amounts of data at a time (like files).  You'll notice in the assignment,
> > it says, "In this assignment you can assume that no NVT string will be
> > longer than 1024 bytes."  Which means only NVT strings, like the control
> > connection, and not the data connection.  The server can send you whatever
> > it wants over the data connection, in arbitrary sized chunks, with no
> > requirement for newlines at all.
> 
> I guess my question is that in that case how do we specify the size of the
> reply buffer than we are expecting back?
> 
> >
> > Reid
> >
> >

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: failure @ getsockname
Date: Sat, 19 Jan 2002 00:02:13 -0800
Lines: 58
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2518

I'm assuming you're getting a compile-error/warning because I believe 
your third arg is wrong; it's supposed to be a 'socklen_t *namelen'. 
Below, you're not passing a pointer to the length, but the length itself.

You need to do something like:

len = sizeof(*clientAddr);
getsockname( dataId, (struct sockaddr *)clientAddr, &len)

getsockname will update the len variable with the real length of the 
object; you never use len, but that's what it does.

Hope that solves your problem.

-Sandeep

Luis Robles wrote:

> I'm following the steps indicated in the assignment handout,
> for establishing the data connection,
> 
> 1.)creating a socket
> 
> zeroing out a struct sockaddr_in struct
> getting the host name via getenv("HOST") - (returns saga3.stanford.edu)
> getting the host address via gethostbyname
> copying the host_addr field into the struct sockaddr_in
> 
> 
> At this point the struct sockaddr_in looks good, so I call
> 2.)bind w/ struct sockaddr_in's sin_port still = 0 from zeroing it out
> earlier
> 3.)calling listen
> 4.)and then calling getsockname to determine the port number
> 
> But getsockname fails on me w/ the following args:
>  
> getsockname( dataId, (struct sockaddr *)clientAddr, sizeof(*clientAddr)
> )
> 
> Should the 3rd arg be the size of the struct sockaddr_in as in the calls
> to bind, or is it sizeof(len) where len is socklen_t as on pg 109?
> 
> Is there another reason why getsockname would fail,
> or is there something faulty before my call to getsockname?
> 
> Cheers,
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: ANSI C library functions
Date: Sat, 19 Jan 2002 00:05:32 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2519

Based on the following reply from Arun to Garrett's question about 
snprintf, I'd say that as long as the thing compiles, you're fine:

<begin quote here>

We'd prefer it, if for consistency sake, that the program compiles without
warning with the -ansi flag.

In this case, the function is linked in so it is present in the library.
To avoid the warning add the declaration:

extern int snprintf(char *s,  size_t  n,  const  char  *format,  /* args*/
....);

and everything should compile and link without event.
<end quote here>

Or at least that's my interpretation of Arun's mail.

-Sandeep

Victor Wong wrote:

> Hi,
> 
> I'm having problems determining whether certain functions that I'm using are in the ANSI C spec. Is there a place that contains a list of the ANSI C library functions? For example, I was under the impression that bzero wasn't in the ANSI C spec - memset should be used instead. However, the given example uses bzero so I'm confused as to what functions are allowed and which ones are not.
> 
> Thanks,
> Victor.
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: server close data connection
Date: Sat, 19 Jan 2002 00:23:31 -0800
Lines: 66
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2520

No, this isn't right (though logical).  Each time, you DO need to create 
a new server socket and listen to it.  In my by-hand experiments, trying 
to recycle the server socket ends up making your next LIST command 
block.  And this isn't a problem in your server socket or anything; it's 
a problem in the ftp server.  Here's the experiment I just did:

1. I started up an interactive Tcl shell (Tcl is this super-cool 
scripting language that I've done a lot of work in in the past six 
years) and created a server socket listening on port 9999.
2. I telnet'ed to port 21 of ftp.stanford.edu and issued the following 
commands:
     USER anonymous
     PASS 
     PORT 171,64,15,69,39,15


The ftp server successfully connected to my server socket in that Tcl 
shell, and I'd coded it such that any data it received from a connection 
would just print to stdout...which it did.

Then I issued another set of PORT and LIST commands (the same as above). 
  The PORT command came back with a 200 response, but the LIST command 
didn't come back with anything...it just hung.  I had to kill that session.

Without touching the Tcl shell that was still (allegedly) listening on 
port 9999, I telnet'ed to port 9999 and typed some random text.  The Tcl 
shell's server socket accepted the connection and printed that text.

So, putting this into the context of our ftpcopy program, you can't 
recycle the server socket because bad things happen at the ftp server end.

And from a security standpoint, this is exactly what you'd want: if you 
left your server socket open throughout the life-time of ftpcopy, ANYONE 
could connect to that port and send you data...and you'd take it and 
save it somewhere (once you did the accept, listen, etc.)

Hmm, as usual, my reply ended up being somewhat longer than I originally 
meant it to be...need to learn to be concise. :(

-Sandeep


Eric Watkins wrote:

>>Yes, a new data connection needs to opened for each file/LIST response
>>(this includes issuing a PORT command etc.)
>>
> 
> But can we (should we) skip steps 1 and 3?  Those seem like things that
> should only be done once.  After you have that first socket set up to
> receive the incoming connections, then you need to repeat all the other
> steps, including accept(), which will get you the new data socket.  Is
> this right?
> 
> -Eric
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Ron Luman II" 
Newsgroups: su.class.cs244a
Subject: Re: Stevens Code
Date: Sat, 19 Jan 2002 01:32:19 -0800
Lines: 21
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: xenon.stanford.edu
X-Trace: news.Stanford.EDU 1011432788 9037 171.64.66.201 (19 Jan 2002 09:33:08 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2521

Hmmm,

I must admit, as detailed as the handout is, I cannot find any text which
_unambiguously_ answers the question.   Would it be possible for you to
excerpt the applicable passage?

Thanks, and sorry for the trouble,

--Ron


"Nick McKeown"  wrote in message

> Russ,
> You'll find a detailed description of
> acceptable re-use in Handout #1, under
> the section on the Honor Code.
>
> - NM


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: some questions about maximum sizes
Date: Sat, 19 Jan 2002 08:01:52 -0800
Lines: 30
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2522

hi,
the grading guidelines mention that we should not assume "maximum size for
list of files". fair enough.
but can we assume the following:

1. size of each single-line reply and that of _each_ line of list response
to be < 1024 (because they are NVT ASCII strings)

2. size of each command string sent (on control socket) < 1024 bytes
((because they are NVT ASCII strings)

3. filenames/path have maximum length

anyone know?
thanks,
ashmi


###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!epic2.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Problems with Accept()
Date: Sat, 19 Jan 2002 08:08:12 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2523


Hi,

After setting up my data connection and sending LIST command to the server
I tried to accept the server's connection. But it did not work. This is
what I am trying to do:

servAddrLen = sizeof(servAddr);
new_socket = accept(control_socket, (struct sockaddr *) &servAddr, &serAddrLen);

where servAddr  is the same as was used to establish the control
connection and control_socket is the socket over the control connection. I
tried using the adat_socket instead also, but in vain.

I would appreciate your help.

Thanks,
Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Symlinks again-symlink to current dir?!?
Date: Sat, 19 Jan 2002 09:35:16 -0800
Lines: 47
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2524

hi,
what if we get the following listing:
-Dir1
 |
 -- a.gz
 |
 -- pub -> .

that is there is a directory pub which is actually a symlink to the
current dir. i guess this could occur at any level, and similarly '..'
could be pointed to by a symlink.

this actually occurs for the listing of ftp.gnu.org.
do we take  of this? if yes, would our resulting destination dir structure
be: (for getting files with ext .gz and level 2)

-Dest
 |
 --a.gz
 |
 --pub
   |
   --a.gz
   |
   --pub
     |
     --a.gz

any ideas?
thanks,
ashmi




###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: UMR in gethostbyname
Date: 19 Jan 2002 17:38:13 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine33.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2525

Derrick Wen-Shiuan Tong  writes:

>When I pass "ftp.fedworld.gov" into gethostbyname(...),
>purify reports a UMR in the library code.

>Believe it or not, it only happens with that host. I've checked
>and made sure that the string is null terminated.

>Does anyone else have this problem, and can we ignore UMR's
>reported in the standard library code?


As long as you've verified that you're passing in valid arguments, yes,
you can ignore UMR warnings in external code.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Problems with Accept()
Date: Sat, 19 Jan 2002 09:48:06 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2526

You don't want to do send control_socket to the accept command; you want 
to send a whole new socket that you created a few lines above.  Take a 
look at the eight steps enumerated in the assignment regarding the 
handling of LIST and RETR.

Good luck.

-Sandeep


Priyanka Gupta wrote:

> Hi,
> 
> After setting up my data connection and sending LIST command to the server
> I tried to accept the server's connection. But it did not work. This is
> what I am trying to do:
> 
> servAddrLen = sizeof(servAddr);
> new_socket = accept(control_socket, (struct sockaddr *) &servAddr, &serAddrLen);
> 
> where servAddr  is the same as was used to establish the control
> connection and control_socket is the socket over the control connection. I
> tried using the adat_socket instead also, but in vain.
> 
> I would appreciate your help.
> 
> Thanks,
> Priyanka
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic21.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Re: Problems with Accept()
Date: Sat, 19 Jan 2002 10:04:54 -0800
Lines: 48
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Sandeep Tamhankar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2527


I did do all the steps, but at the last command (accept()) it fails.

On Sat, 19 Jan 2002, Sandeep Tamhankar wrote:

> You don't want to do send control_socket to the accept command; you want
> to send a whole new socket that you created a few lines above.  Take a
> look at the eight steps enumerated in the assignment regarding the
> handling of LIST and RETR.
>
> Good luck.
>
> -Sandeep
>
>
> Priyanka Gupta wrote:
>
> > Hi,
> >
> > After setting up my data connection and sending LIST command to the server
> > I tried to accept the server's connection. But it did not work. This is
> > what I am trying to do:
> >
> > servAddrLen = sizeof(servAddr);
> > new_socket = accept(control_socket, (struct sockaddr *) &servAddr, &serAddrLen);
> >
> > where servAddr  is the same as was used to establish the control
> > connection and control_socket is the socket over the control connection. I
> > tried using the adat_socket instead also, but in vain.
> >
> > I would appreciate your help.
> >
> > Thanks,
> > Priyanka
> >
> >
>
>
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Reply Buffer over Control and Data connection...
Date: 19 Jan 2002 18:14:02 GMT
Lines: 10
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine33.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2528

Ashmi  writes:

>so can we assume that all of the following will be no more than 1024
>bytes:
>1. every single line reply over ctrl connection
>2. every line in the response from LIST (over data connection, but the
>response of LIST is in NVT ASCII)

Yes, both assumptions are reasonable.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Reply Buffer over Control and Data connection...
Date: 19 Jan 2002 18:29:06 GMT
Lines: 32
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine33.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2529

>> > 2) For data connection the server's reply could be very big. But would the
>> > server still reply in packets of 1024 bytes?
>> > I mean that the buffer that we pass to read(s, ...) should be 1024 bytes
>> > and then we need to just keep reading such buffers and then concatenate
>> > the data packets?
>> > Is that true? Please explain.

The connection uses TCP, i.e. it's stream-oriented rather than
datagram-oriented, so the server just writes data without any particular
notion of packets.  (This will be explained in much greater detail in a
couple of lectures, and particularly in the next assignment).


>> The data connections are different, in that they can send arbitrary
>> amounts of data at a time (like files).  You'll notice in the assignment,
>> it says, "In this assignment you can assume that no NVT string will be
>> longer than 1024 bytes."  Which means only NVT strings, like the control
>> connection, and not the data connection.  The server can send you whatever
>> it wants over the data connection, in arbitrary sized chunks, with no
>> requirement for newlines at all.

>I guess my question is that in that case how do we specify the size of the
>reply buffer than we are expecting back?

The server will close its end of the connection once it has finished sending
whatever you've requested on the data connection, so you can loop on
recv() until it returns zero (EOF), grabbing some amount of data from the
connection each time.  The buffer size you decide to use is up to you.
(Obviously, for efficiency's sake, how you handle NVT strings in a LIST
response may not necessarily be how you'll handle the transfer of a large
file).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Reply 4xx
Date: 19 Jan 2002 18:33:30 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine33.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2530



>I wanted to know that if we get a transient negative completion reply,
>which means we could retry later. Should we retry again or can we just
>treat 4xx and 5xx the same way?

As the RFC says, the meaning of "transient" is rather vague.  You can
handle 4xx and 5xx errors the same way in this assignment.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ANSI C library functions
Date: 19 Jan 2002 18:53:50 GMT
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine33.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2531


>I'm having problems determining whether certain functions that I'm using
>are in the ANSI C spec. Is there a place that contains a list of the
>ANSI C library functions? For example, I was under the impression that
>bzero wasn't in the ANSI C spec - memset should be used instead. However,
>the given example uses bzero so I'm confused as to what functions are
>allowed and which ones are not.

The intent of requiring ANSI C is primarily that people use proper function
prototyping (no K&R-style function declarations), no C++, etc.

In fact, a lot of the library functions you're required to use in this program
are not ANSI C library functions--e.g. the socket calls themselves.  There's
no way around this.

Having said that, for the example you mention, given the issue of portability,
I'd suggest using memset rather than bzero since the option is available.
(Keep in mind that portability is one of the design criteria to consider in
the grading guidelines, so you should try to avoid system-specific
implementation where possible--or if some details are system specific,
they might be abstracted out or otherwise documented in your code).

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic19.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Re: ANSI C library functions
Date: Sat, 19 Jan 2002 11:14:31 -0800
Lines: 30
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: epic19.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Matthew Jonathan Holliman 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2532


Just FYI, the FAQ mentions that you could use bzero or bcopy although
technically they are not ANSI C compliant.

On 19 Jan 2002, Matthew Jonathan Holliman wrote:

>
> >I'm having problems determining whether certain functions that I'm using
> >are in the ANSI C spec. Is there a place that contains a list of the
> >ANSI C library functions? For example, I was under the impression that
> >bzero wasn't in the ANSI C spec - memset should be used instead. However,
> >the given example uses bzero so I'm confused as to what functions are
> >allowed and which ones are not.
>
> The intent of requiring ANSI C is primarily that people use proper function
> prototyping (no K&R-style function declarations), no C++, etc.
>
> In fact, a lot of the library functions you're required to use in this program
> are not ANSI C library functions--e.g. the socket calls themselves.  There's
> no way around this.
>
> Having said that, for the example you mention, given the issue of portability,
> I'd suggest using memset rather than bzero since the option is available.
> (Keep in mind that portability is one of the design criteria to consider in
> the grading guidelines, so you should try to avoid system-specific
> implementation where possible--or if some details are system specific,
> they might be abstracted out or otherwise documented in your code).
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: Server replies and end-of-transmission
Date: Sat, 19 Jan 2002 12:00:40 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2533

I've read the two posts on server replies already, but I'm still not sure
about how we're supposed to detect the end of a transmission over the data
connection. Since it's not ASCII it's not terminated by an /r/n; if we keep
on reading when there is no data left the socket blocks (someone please
explain, too, why the assignment states that the read will return 0).

Also, what is the protocol for the LIST command? Does the server write the
completion acknowledgement (226?) after it's done writing on its side, or
does it wait until I have retrieved the information from the socket? I'm
waiting for a confirmation but it never comes.

Thanks a bunch, guys.



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Server replies and end-of-transmission
Date: Sat, 19 Jan 2002 12:14:57 -0800
Lines: 51
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2534

Steven Siuhong Ngai wrote:

> I've read the two posts on server replies already, but I'm still not sure
> about how we're supposed to detect the end of a transmission over the data
> connection. Since it's not ASCII it's not terminated by an /r/n; if we keep
> on reading when there is no data left the socket blocks (someone please
> explain, too, why the assignment states that the read will return 0).


read() returns 0 on eof; so you keep reading until read returns 0.  Then 
you're done.  Because you want to read until eof, you don't have to 
worry about blocking.  Said differently, you have to worry about 
blocking on the control connection because no one's closing the 
connection.  You send something, you read something.  If you read more 
than you're supposed to, you'll block forever and your world will come 
crashing around you.


> Also, what is the protocol for the LIST command? Does the server write the
> completion acknowledgement (226?) after it's done writing on its side, or
> does it wait until I have retrieved the information from the socket? I'm
> waiting for a confirmation but it never comes.
> 
> Thanks a bunch, guys.


Not sure about this one; I'd guess that it sends the acknowledgment 
after it's done writing all the data to the socket since regardless of 
what you do, it knows it's done it's part.  But the safer assumption is 
that the acknowledgment comes after you're done reading, since this 
covers the scenario where the server couldn't write everything because 
some lower-level thing (kernel? IP protocol?) filled up its buffers 
waiting for you to read some of the data.  So, I'd recommend (and this 
is what I did) reading the acknowledgment after reading all the data off 
the data socket.

Hope that helps.


-Sandeep




-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine6.Stanford.EDU!htimam
From:  (Hasan Taufiq Imam)
Newsgroups: su.class.cs244a
Subject: multiline replies
Date: 19 Jan 2002 22:58:57 GMT
Organization: Stanford University, CA 94305, USA
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine6.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2535

I am not sure what is the significance of multiline replies processing.
Control replies are all going to be 1024 bytes. For each command one call to
read should be enough. For our programs sake we just need to process the first
line and the rest of the lines are not important. Can we receive multi-line 
replies with two read()? A single reply from the server should always be read 
with a single read right?
  


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tim Chao 
Newsgroups: su.class.cs244a
Subject: Re: Server replies and end-of-transmission
Date: Sat, 19 Jan 2002 15:24:32 -0800
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: slimtc.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.7 [en] (Win98; I)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2536



Steven Siuhong Ngai wrote:

> Also, what is the protocol for the LIST command? Does the server write the
> completion acknowledgement (226?) after it's done writing on its side, or
> does it wait until I have retrieved the information from the socket? I'm
> waiting for a confirmation but it never comes.

I think I'm having the same problem here -- I've accepted the data connection
and over the control connection I get the completion acknowledgement (226). So
there should be stuff to read from the data socket. But, when I try to retrieve
the LIST response in ASCII, I get an error on read() - a return value of -1.

Any suggestions?

Thanks guys,
Tim

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine23.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Server replies and end-of-transmission
Date: Sat, 19 Jan 2002 15:28:26 -0800
Lines: 77
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine23.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2537

 So, I'd recommend (and this
> is what I did) reading the acknowledgment after reading all the data off
> the data socket.

that's what i am doing too, but ensure that you wait for 125/150 before
you starting to read the data. else you might get out of sync with the
server.

~ashmi

On Sat, 19 Jan 2002, Sandeep Tamhankar wrote:

> Steven Siuhong Ngai wrote:
>
> > I've read the two posts on server replies already, but I'm still not sure
> > about how we're supposed to detect the end of a transmission over the data
> > connection. Since it's not ASCII it's not terminated by an /r/n; if we keep
> > on reading when there is no data left the socket blocks (someone please
> > explain, too, why the assignment states that the read will return 0).
>
>
> read() returns 0 on eof; so you keep reading until read returns 0.  Then
> you're done.  Because you want to read until eof, you don't have to
> worry about blocking.  Said differently, you have to worry about
> blocking on the control connection because no one's closing the
> connection.  You send something, you read something.  If you read more
> than you're supposed to, you'll block forever and your world will come
> crashing around you.
>
>
> > Also, what is the protocol for the LIST command? Does the server write the
> > completion acknowledgement (226?) after it's done writing on its side, or
> > does it wait until I have retrieved the information from the socket? I'm
> > waiting for a confirmation but it never comes.
> >
> > Thanks a bunch, guys.
>
>
> Not sure about this one; I'd guess that it sends the acknowledgment
> after it's done writing all the data to the socket since regardless of
> what you do, it knows it's done it's part.  But the safer assumption is
> that the acknowledgment comes after you're done reading, since this
> covers the scenario where the server couldn't write everything because
> some lower-level thing (kernel? IP protocol?) filled up its buffers
> waiting for you to read some of the data.  So, I'd recommend (and this
> is what I did) reading the acknowledgment after reading all the data off
> the data socket.
>
> Hope that helps.
>
>
> -Sandeep
>
>
>
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!epic21.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Bytes Read over data connection
Date: Sat, 19 Jan 2002 15:31:37 -0800
Lines: 25
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2538


Hi All,

After setting up a data connection, when I do a LIST, I get the following
response over the control connection:

150 Opening ASCII mode data connection for /bin/ls.
226 Transfer complete.

However when I try to read the data over the data connection:

while(1)
{
	if(( bytes_read = read(data_socket, replyBuffer, REPLY_SIZE )) > 0)
	{
		fprintf(stderr, "%s \n\n", replyBuffer);
		break;
	}
}

I get -1 bytes_read, which means its failing. Any idea why this might be?

Thnaks,
Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine23.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: multiline replies
Date: Sat, 19 Jan 2002 15:32:55 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine23.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2539



> Control replies are all going to be 1024 bytes.
i dont think this is correct - each NVT ASCII string is 1024, means each
line of the multiline could be at most 1024.you need to ensure you stop
reading when the multiline gets over or the socket will block.

>A single reply from the server should always be read
> with a single read right?
not necessarily, you might need to read char by char on the control
connection.

~ashmi


On 19 Jan 2002, Hasan
Taufiq Imam wrote:

> I am not sure what is the significance of multiline replies processing.
> Control replies are all going to be 1024 bytes. For each command one call to
> read should be enough. For our programs sake we just need to process the first
> line and the rest of the lines are not important. Can we receive multi-line
> replies with two read()? A single reply from the server should always be read
> with a single read right?
>
>
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: multiline replies
Date: Sat, 19 Jan 2002 15:33:06 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2540

Well you need to make sure you read the rest of the lines in the 
response, or else next time you send a command, you'll end up reading 
the tail end of the prior response.

-Sandeep

Hasan Taufiq Imam wrote:

> I am not sure what is the significance of multiline replies processing.
> Control replies are all going to be 1024 bytes. For each command one call to
> read should be enough. For our programs sake we just need to process the first
> line and the rest of the lines are not important. Can we receive multi-line 
> replies with two read()? A single reply from the server should always be read 
> with a single read right?
>   
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Bytes Read over data connection
Date: Sat, 19 Jan 2002 15:36:22 -0800
Lines: 43
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2541

Probably an interrupt in the system; this would cause a -1 bytes_read, 
and then you have to check if errno == EINTR to decide whether it's a 
real error or not.  (If you do get EINTR, then just 'continue'.)

-Sandeep

Priyanka Gupta wrote:

> Hi All,
> 
> After setting up a data connection, when I do a LIST, I get the following
> response over the control connection:
> 
> 150 Opening ASCII mode data connection for /bin/ls.
> 226 Transfer complete.
> 
> However when I try to read the data over the data connection:
> 
> while(1)
> {
> 	if(( bytes_read = read(data_socket, replyBuffer, REPLY_SIZE )) > 0)
> 	{
> 		fprintf(stderr, "%s \n\n", replyBuffer);
> 		break;
> 	}
> }
> 
> I get -1 bytes_read, which means its failing. Any idea why this might be?
> 
> Thnaks,
> Priyanka
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: Script file sizes (normal line endings this time)
Date: Sat, 19 Jan 2002 15:45:27 -0800
Lines: 84
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2542

I actually have this exact same error (and my numbers match yours) on
server 204.123.2.2.  I went and looked at the server using ftp, and my
numbers seem to match the numbers on their currently.  Is it possible
that this directory is being changed periodically?  Or did you find
something in your program that affected your answer?

Thanks,
Clayton


Bret Taylor wrote:
> 
> Nevermind, it woked this time, it must have been my numbers that were wrong.
> 
> Bret
> 
> "Bret Taylor"  wrote in message
> 
> > I noticed my first post had some weird carriage returns, so I am reposting
> > to make it easier to read.  Sorry for the trouble.
> >
> > First of all, thanks for providing the test scripts, it is great to have
> > such a simple and thorough way to test our code.  However, I have a couple
> > questions regarding  "Test F.2".
> >
> > My test script output includes this:
> >
> > test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> > exit line said: "OK: 1900222 bytes copied"
> > ---
> > diff between expected tree and your tree:
> > 1,8d0
> > < ./kermit/c/all.tar.Z 2606058765 72935
> > < ./kermit/c/amiga.tar.Z 1017293024 135815
> > < ./kermit/c/dg.tar.Z 2560791635 406317
> > < ./kermit/c/mac.tar.Z 2459747744 496551
> > < ./kermit/c/os2.tar.Z 2771080586 163808
> > < ./kermit/c/os9.tar.Z 370278796 49438
> > < ./kermit/c/unix.tar.Z 308104872 349221
> > < ./kermit/c/vms.tar.Z 53741681 64755
> >
> > I used ncftp to go to the site to see what was going on, and a "ls -l" in
> > the kermit/c directory shows this:
> >
> > ncftp /pub/comm/kermit/c > ls -l
> > lrwxrwxrwx   1 0        0           37   Apr 14  1995
> > 00README-Legal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs
> > -r--r--r--   1 0        0          935   Jan 17  1989   README
> > -r--r--r--   1 0        0        72935   Jan 16  1989   all.tar.Z
> > -r--r--r--   1 0        0       135815   Jan 16  1989   amiga.tar.Z
> > -r--r--r--   1 0        0       406317   Jan 16  1989   dg.tar.Z
> > -r--r--r--   1 0        0       496551   Jan 16  1989   mac.tar.Z
> > -r--r--r--   1 0        0       163808   Jan 16  1989   os2.tar.Z
> > -r--r--r--   1 0        0        49438   Jan 16  1989   os9.tar.Z
> > -r--r--r--   1 0        0       349221   Jan 16  1989   unix.tar.Z
> > -r--r--r--   1 0        0        64755   Jan 16  1989   vms.tar.Z
> >
> > It looks like the files sizes are the ones I got, not the large ones from
> > the script.  Am I mistaken here? The same thing happened on the MS FTP
> site
> > (snippets below).
> >
> > test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie exit line
> > said: "OK: 5113 bytes copied"
> > ---
> > diff between expected tree and your tree:
> > 3,4d2
> > < ./ie-public/fixes/usa/ie401/importexportfavorites-fix/readme.txt
> > 2939125283 643 <
> > ./ie-public/fixes/usa/ie50/importexportfavorites-fix/readme.txt 2939125283
> > 643
> >
> > ncftp ...ortexportfavorites-fix > ls -l
> > dr-xr-xr-x   1 owner    group            0 Feb 25  2000 alpha
> > -r-xr-xr-x   1 owner    group          643 Sep 24  1999 readme.txt
> > dr-xr-xr-x   1 owner    group            0 Feb 25  2000 x86
> >
> >
> > Thanks for your time,
> > Bret
> >
> >
> >
> >
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tim Chao 
Newsgroups: su.class.cs244a
Subject: Re: Bytes Read over data connection
Date: Sun, 20 Jan 2002 00:30:02 +0000 (UTC)
Lines: 20
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: slimtc.stanford.edu
X-Trace: news.Stanford.EDU 1011486602 14319 128.12.65.49 (20 Jan 2002 00:30:02 GMT)
X-Complaints-To: 
User-Agent: Xnews/4.11.09
Xref: nntp.stanford.edu su.class.cs244a:2543

Sandeep Tamhankar  wrote in
 

> Probably an interrupt in the system; this would cause a -1 bytes_read, 
> and then you have to check if errno == EINTR to decide whether it's a 
> real error or not.  (If you do get EINTR, then just 'continue'.)
> 
> -Sandeep

Thanks for the lead... Anyhow, I did a check using perror and errno and 
got the error "Transport endpoint not connected". Does this mean that the 
server hasn't connected to me after all? I was able to accept() the 
connection and the server sent me the 226 message saying that the LIST 
transfer was complete.

Any ideas?

Thanks,
Tim

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Bytes Read over data connection
Date: 20 Jan 2002 00:40:23 GMT
Lines: 26
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2544

Tim Chao  writes:

>Sandeep Tamhankar  wrote in
 

>> Probably an interrupt in the system; this would cause a -1 bytes_read, 
>> and then you have to check if errno == EINTR to decide whether it's a 
>> real error or not.  (If you do get EINTR, then just 'continue'.)
>> 
>> -Sandeep

>Thanks for the lead... Anyhow, I did a check using perror and errno and 
>got the error "Transport endpoint not connected". Does this mean that the 
>server hasn't connected to me after all? I was able to accept() the 
>connection and the server sent me the 226 message saying that the LIST 
>transfer was complete.

>Any ideas?

>Thanks,
>Tim


Be sure that you issue the read() on the socket that accept() returned, not
the socket on which you were listening for the FTP server connection.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tim Chao 
Newsgroups: su.class.cs244a
Subject: Re: Bytes Read over data connection
Date: Sat, 19 Jan 2002 16:45:04 -0800
Lines: 9
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: slimtc.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.7 [en] (Win98; I)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2545

>
> Be sure that you issue the read() on the socket that accept() returned, not
> the socket on which you were listening for the FTP server connection.

That was the first thing I checked. I know that it's erroring out on my data
socket and not the control. Also, in that case it would cause the program to
hang and not error right?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tim Chao 
Newsgroups: su.class.cs244a
Subject: Re: Bytes Read over data connection
Date: Sat, 19 Jan 2002 16:47:30 -0800
Lines: 6
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: slimtc.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.7 [en] (Win98; I)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2546

>
> Be sure that you issue the read() on the socket that accept() returned, not
> the socket on which you were listening for the FTP server connection.

Nevermind, i misread you the first time. That worked. Thanks!

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!htimam
From:  (Hasan Taufiq Imam)
Newsgroups: su.class.cs244a
Subject: Re: multiline replies
Date: 20 Jan 2002 00:53:14 GMT
Organization: Stanford University, CA 94305, USA
Lines: 9
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine13.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2547

thanks much.
tell me if I am right:
1. Each line in control connection is terminated by telnet end of line
2. One line can be longer than 1024 bytes
3. Thus a control reply can be longer than 1024 bytes
4. If that is the case what is the difference between date reply and control 
reply. You have to dynamically accumulate control replies also because control
replies can be arbitarily long.
clarifications will be much appreciated. 
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic13.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Test script
Date: Sat, 19 Jan 2002 17:13:47 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2548


Hi All,

I can't locate the test script. Could someone tell where its located?

Thanks a bunch,
Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Cool Networking Term of the Day
Date: Sat, 19 Jan 2002 19:12:03 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2549

Page 288, Peterson and Davie:

"split horizon with poison reverse"

Now get back to work, people; a few days to go and the pain and 
suffering will be over...(for this project at least...)

Good luck.

-Sandeep
-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic0.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Rearding the assignment 2
Date: Sat, 19 Jan 2002 20:17:23 -0800
Lines: 7
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic0.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2550

Hi,
I am not able to access the /afs/ir/class/cs244a/WWW/homeworks/hw2/gethw2
script.
Thanks
Regards
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Rearding the assignment 2
Date: Sat, 19 Jan 2002 20:23:03 -0800
Lines: 32
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2551

If you're really anxious to get started, go to 
http://www.stanford.edu/class/cs244a/homeworks/hw2/gethw2

This script is what you would run if you had permission; just do by hand 
(through your browser or wget) what this script says.

Or wait until the TAs move all this code to someplace reasonable like 
/afs/ir/class/cs244a/homeworks so that we aren't forced to use wget and 
such tools anytime we need anything.

-Sandeep

Shankar Agarwal wrote:

> Hi,
> I am not able to access the /afs/ir/class/cs244a/WWW/homeworks/hw2/gethw2
> script.
> Thanks
> Regards
> Shankar
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Adrian Graham" 
Newsgroups: su.class.cs244a
Subject: code re-use policy
Date: Sun, 20 Jan 2002 21:46:33 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: zoostation.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2552

I'd appreciate a clarification of the code re-use policy. I have read
handout 1, but I'm still unsure of a couple of cases.

(1) Support code written by someone else. For instance, a dynamic array
library. It is clear to me that an FTP library would not be appropriate for
assignment 1, but the policy with regard to support code is less clear.

(2) Support code written by me. For instance, a dynamic array library I
wrote in a previous class.

Thanks,

Adrian


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Slow Test Script
Date: Sat, 19 Jan 2002 22:42:43 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2553

Hi,

I want to make sure that I am not the only one who faces the slow testing.
I sent my test before 3 hours, and till now I didn't get anything. Is this
normal?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Ron Luman II" 
Newsgroups: su.class.cs244a
Subject: Re: code re-use policy
Date: Sat, 19 Jan 2002 23:43:22 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
X-Trace: news.Stanford.EDU 1011512655 17484 171.64.66.201 (20 Jan 2002 07:44:15 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2554

I agree that this is still ambiguous.  I would also like a response as to
whether we may reuse code from:

- Stevens
- echodemo.c example

Thanks,

--Ron


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: Slow Test Script
Date: Sun, 20 Jan 2002 00:02:31 -0800
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2555

You got a response??  I'm waiting on 4 hours now and still nothing.

--Russ


"Moh'd Saleem Saleem Alrawashdeh"  wrote in message

> Hi,
>
> I want to make sure that I am not the only one who faces the slow testing.
> I sent my test before 3 hours, and till now I didn't get anything. Is this
> normal?
>
> Thanks,
>
> Moh'd
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine27.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Assignment1: Questions regarding symbolic links
Date: Sun, 20 Jan 2002 02:26:14 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2556

I have few questions regarding symbolic links:
(1) Do we have to handle the case of cycles due to symbolic links. All
directories in a cycle may be "empty" and we may keep looping if we either
don't create the directories or maintain a possibly huge data structure
depicting how the directories are organized.

(2) If there's a symbolic link 1.gz -> 1.txt and we are supposed to copy
all the files with gz ext, do we copy 1.txt? And in case we are supposed
to copy all files with .txt ext do we need to look at what the link points
to and figure out if it has a .txt ext and copy 1.txt - and in this case
what should be the name of final file.

(3) If there's a symbolic link "l" to a dir "d" do we create a dir with
the name "l" or "d". If we try to create a dir with name "d" there might
already be a file/dir on the local filesystem with that name (this may
not be due to ftpcopy).

Thanks,
Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic12.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Re: Slow Test Script
Date: Sun, 20 Jan 2002 06:35:04 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Russell Greene 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2557


I think there might be something wrong. I haven't received my reply so far
and I had submitted it last night.

On Sun, 20 Jan 2002, Russell Greene wrote:

> You got a response??  I'm waiting on 4 hours now and still nothing.
>
> --Russ
>
>
> "Moh'd Saleem Saleem Alrawashdeh"  wrote in message
> 
> > Hi,
> >
> > I want to make sure that I am not the only one who faces the slow testing.
> > I sent my test before 3 hours, and till now I didn't get anything. Is this
> > normal?
> >
> > Thanks,
> >
> > Moh'd
> >
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Dead Test Script
Date: Sun, 20 Jan 2002 10:10:01 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2558

I have waited over night for the test script to come back with no luck.  Can
someone post the status of the problem and if the course staff knows about
it?  Appreciated.

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: test
Date: Sun, 20 Jan 2002 11:34:41 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2559


Hi, I user the command: /afs/ir/class/cs244a/bin/testhw.pl TA hw1 (TA is
replace with user name of my TA) but was told no command found.
What's wrong with it?

Any help is highly appreciated!

Thanks,
-jinhui

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script
Date: 20 Jan 2002 21:23:36 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2560

"Russell Greene"  writes:

>I have waited over night for the test script to come back with no luck.  Can
>someone post the status of the problem and if the course staff knows about
>it?  Appreciated.

The server should be up and running again, I hope--please let us know if
you encounter more problems.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: test
Date: 20 Jan 2002 21:24:27 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2561

Jinhui Pan  writes:


>Hi, I user the command: /afs/ir/class/cs244a/bin/testhw.pl TA hw1 (TA is
>replace with user name of my TA) but was told no command found.
>What's wrong with it?

The test script is named submit_test.pl.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding the .purify file.
Date: Sun, 20 Jan 2002 13:25:24 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2562

Hi,
When i am trying to submit the hw1 by using
script /usr/class/cs244a/bin/submit TA hw1. It
does not seem to pick up the .purify file. Has someone else seen the same
problem or i am the only one having this problem. Also how can we supress
the errors caused by the gethostbyname. The faq says that we should do
suppress UMR ...;res_search
suppress UMR ...;nss_search
But it doesnot say where we are suppose to add this command.
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Assignment1: Questions regarding symbolic links
Date: 20 Jan 2002 21:38:51 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2563

Dhawal Kumar  writes:

>I have few questions regarding symbolic links:
>(1) Do we have to handle the case of cycles due to symbolic links. All
>directories in a cycle may be "empty" and we may keep looping if we either
>don't create the directories or maintain a possibly huge data structure
>depicting how the directories are organized.

Nope.

>(2) If there's a symbolic link 1.gz -> 1.txt and we are supposed to copy
>all the files with gz ext, do we copy 1.txt? And in case we are supposed
>to copy all files with .txt ext do we need to look at what the link points
>to and figure out if it has a .txt ext and copy 1.txt - and in this case
>what should be the name of final file.

>(3) If there's a symbolic link "l" to a dir "d" do we create a dir with
>the name "l" or "d". If we try to create a dir with name "d" there might
>already be a file/dir on the local filesystem with that name (this may
>not be due to ftpcopy).

Please see the earlier posted responses to these questions; thanks.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine25.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the .purify file.
Date: Sun, 20 Jan 2002 14:30:25 -0800
Lines: 37
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine25.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2564


Shankar,

I guess purify picks the commands to suppress from .purify file (by
default). So, you may add these two suppress commands to .purify file.


Professor and TAs

1. The submission scripts are not picking up the .purify file. Any
comments?

2. One more, I am using PASV to get the server data port. I am not getting
any UMRs from the network calls that I am making. Should I still put these
statements in the .purify file? I guess not.

Thanks
Anurag




On Sun, 20 Jan 2002, Shankar Agarwal wrote:

> Hi,
> When i am trying to submit the hw1 by using
> script /usr/class/cs244a/bin/submit TA hw1. It
> does not seem to pick up the .purify file. Has someone else seen the same
> problem or i am the only one having this problem. Also how can we supress
> the errors caused by the gethostbyname. The faq says that we should do
> suppress UMR ...;res_search
> suppress UMR ...;nss_search
> But it doesnot say where we are suppose to add this command.
> Shankar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine25.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: Re: test
Date: Sun, 20 Jan 2002 14:34:24 -0800
Lines: 22
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine25.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2565



Command is:

/afs/ir/class/cs244a/bin/submit_test.pl hw1 TA.

Order of arguments was incorrect in the printout.



On 20 Jan 2002, Matthew Jonathan Holliman wrote:

> Jinhui Pan  writes:
>
>
> >Hi, I user the command: /afs/ir/class/cs244a/bin/testhw.pl TA hw1 (TA is
> >replace with user name of my TA) but was told no command found.
> >What's wrong with it?
>
> The test script is named submit_test.pl.
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic14.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: ReplyBuffer
Date: Sun, 20 Jan 2002 16:13:49 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2566


Hi,

Sometime when I do a read from the server, it seems that the
server appends some data from the last command to the buffer
also.

As a result lets say I do the following:
1) PORT command write
2) READ : 200 PORT command successful.
3) RETR write
4) RETR read :
200 PORT command successful.
550 atimetokill.pl: No such file or directory.

replyBuffer has some data left over from the last time although that was
already read.

Any ideas?

Thanks,
Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga2.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script
Date: Sun, 20 Jan 2002 16:18:32 -0800
Lines: 27
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2567

I sent my test yesterday at 8 pm. At 12 today morning I got the message
that my test started and that I will recieve the results shortely. Till
now (4:30 pm) no results come back.

It seems that the server is down again. What I suggest if the TAs can tell
us what the command they run, and what directory heirarchy that
is expected. After that, we can do the test by ourselves. It would be much
faster and easier.

Thanks,

Moh'd

 On 20 Jan 2002, Matthew Jonathan Holliman wrote:

> "Russell Greene"  writes:
>
> >I have waited over night for the test script to come back with no luck.  Can
> >someone post the status of the problem and if the course staff knows about
> >it?  Appreciated.
>
> The server should be up and running again, I hope--please let us know if
> you encounter more problems.
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script
Date: Sun, 20 Jan 2002 16:23:01 -0800
Lines: 39
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2568

Try re-submitting your program to the test script.  I had it work for me at
2:30pm (waited about 5 mins).

--Russ


"Moh'd Saleem Saleem Alrawashdeh"  wrote in message

> I sent my test yesterday at 8 pm. At 12 today morning I got the message
> that my test started and that I will recieve the results shortely. Till
> now (4:30 pm) no results come back.
>
> It seems that the server is down again. What I suggest if the TAs can tell
> us what the command they run, and what directory heirarchy that
> is expected. After that, we can do the test by ourselves. It would be much
> faster and easier.
>
> Thanks,
>
> Moh'd
>
>  On 20 Jan 2002, Matthew Jonathan Holliman wrote:
>
> > "Russell Greene"  writes:
> >
> > >I have waited over night for the test script to come back with no luck.
Can
> > >someone post the status of the problem and if the course staff knows
about
> > >it?  Appreciated.
> >
> > The server should be up and running again, I hope--please let us know if
> > you encounter more problems.
> >
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: Script file sizes (normal line endings this time)
Date: Sun, 20 Jan 2002 16:24:48 -0800
Lines: 92
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2569

Just in case someone else runs across this bug...

you might be using strlen when writing your data out to a file.

Good luck
Clayton

Clayton Pierce Jones wrote:
> 
> I actually have this exact same error (and my numbers match yours) on
> server 204.123.2.2.  I went and looked at the server using ftp, and my
> numbers seem to match the numbers on their currently.  Is it possible
> that this directory is being changed periodically?  Or did you find
> something in your program that affected your answer?
> 
> Thanks,
> Clayton
> 
> Bret Taylor wrote:
> >
> > Nevermind, it woked this time, it must have been my numbers that were wrong.
> >
> > Bret
> >
> > "Bret Taylor"  wrote in message
> > 
> > > I noticed my first post had some weird carriage returns, so I am reposting
> > > to make it easier to read.  Sorry for the trouble.
> > >
> > > First of all, thanks for providing the test scripts, it is great to have
> > > such a simple and thorough way to test our code.  However, I have a couple
> > > questions regarding  "Test F.2".
> > >
> > > My test script output includes this:
> > >
> > > test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> > > exit line said: "OK: 1900222 bytes copied"
> > > ---
> > > diff between expected tree and your tree:
> > > 1,8d0
> > > < ./kermit/c/all.tar.Z 2606058765 72935
> > > < ./kermit/c/amiga.tar.Z 1017293024 135815
> > > < ./kermit/c/dg.tar.Z 2560791635 406317
> > > < ./kermit/c/mac.tar.Z 2459747744 496551
> > > < ./kermit/c/os2.tar.Z 2771080586 163808
> > > < ./kermit/c/os9.tar.Z 370278796 49438
> > > < ./kermit/c/unix.tar.Z 308104872 349221
> > > < ./kermit/c/vms.tar.Z 53741681 64755
> > >
> > > I used ncftp to go to the site to see what was going on, and a "ls -l" in
> > > the kermit/c directory shows this:
> > >
> > > ncftp /pub/comm/kermit/c > ls -l
> > > lrwxrwxrwx   1 0        0           37   Apr 14  1995
> > > 00README-Legal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs
> > > -r--r--r--   1 0        0          935   Jan 17  1989   README
> > > -r--r--r--   1 0        0        72935   Jan 16  1989   all.tar.Z
> > > -r--r--r--   1 0        0       135815   Jan 16  1989   amiga.tar.Z
> > > -r--r--r--   1 0        0       406317   Jan 16  1989   dg.tar.Z
> > > -r--r--r--   1 0        0       496551   Jan 16  1989   mac.tar.Z
> > > -r--r--r--   1 0        0       163808   Jan 16  1989   os2.tar.Z
> > > -r--r--r--   1 0        0        49438   Jan 16  1989   os9.tar.Z
> > > -r--r--r--   1 0        0       349221   Jan 16  1989   unix.tar.Z
> > > -r--r--r--   1 0        0        64755   Jan 16  1989   vms.tar.Z
> > >
> > > It looks like the files sizes are the ones I got, not the large ones from
> > > the script.  Am I mistaken here? The same thing happened on the MS FTP
> > site
> > > (snippets below).
> > >
> > > test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie exit line
> > > said: "OK: 5113 bytes copied"
> > > ---
> > > diff between expected tree and your tree:
> > > 3,4d2
> > > < ./ie-public/fixes/usa/ie401/importexportfavorites-fix/readme.txt
> > > 2939125283 643 <
> > > ./ie-public/fixes/usa/ie50/importexportfavorites-fix/readme.txt 2939125283
> > > 643
> > >
> > > ncftp ...ortexportfavorites-fix > ls -l
> > > dr-xr-xr-x   1 owner    group            0 Feb 25  2000 alpha
> > > -r-xr-xr-x   1 owner    group          643 Sep 24  1999 readme.txt
> > > dr-xr-xr-x   1 owner    group            0 Feb 25  2000 x86
> > >
> > >
> > > Thanks for your time,
> > > Bret
> > >
> > >
> > >
> > >
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine26.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script
Date: 21 Jan 2002 00:29:19 GMT
Lines: 20
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine26.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2570

Moh'd Saleem Saleem Alrawashdeh  writes:

>I sent my test yesterday at 8 pm. At 12 today morning I got the message
>that my test started and that I will recieve the results shortely. Till
>now (4:30 pm) no results come back.

>It seems that the server is down again. What I suggest if the TAs can tell
>us what the command they run, and what directory heirarchy that
>is expected. After that, we can do the test by ourselves. It would be much
>faster and easier.

>Thanks,

As far as I can tell, the server is still up.

It's quite possible (actually, probable) that you received the message while
the server was being brought back up and down originally (at about noon),
in which case the test may never have been carried out correctly.  Can you
try resubmitting your code now?

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: ReplyBuffer
Date: Sun, 20 Jan 2002 16:35:56 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2571

Well, my guess is that you're recycling the buffer and using 
strcat/strncat instead of strcpy/strncpy.  Be sure to initialize your 
buffer to the empty string before doing anything with it.

-Sandeep

Priyanka Gupta wrote:

> Hi,
> 
> Sometime when I do a read from the server, it seems that the
> server appends some data from the last command to the buffer
> also.
> 
> As a result lets say I do the following:
> 1) PORT command write
> 2) READ : 200 PORT command successful.
> 3) RETR write
> 4) RETR read :
> 200 PORT command successful.
> 550 atimetokill.pl: No such file or directory.
> 
> replyBuffer has some data left over from the last time although that was
> already read.
> 
> Any ideas?
> 
> Thanks,
> Priyanka
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!wkimball
From: Kimball Chun-Jen Wang 
Newsgroups: su.class.cs244a
Subject: Can't compile echodemo.c
Date: Sun, 20 Jan 2002 16:57:53 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2572

Hello:

	I can't compile echodemo.c or any socket related c program.
I think something is wrong with my .cshrc file.  Can some one tell
me what to include for .cshrc?

Kimball

Here is the error message.
elaine30:~/cs244a/pj1demo> make echodemo
cc   echodemo.o   -o echodemo
Undefined                       first referenced
 symbol                             in file
socket                              echodemo.o
gethostbyname                       echodemo.o
getservbyname                       echodemo.o
inet_addr                           echodemo.o
shutdown                            echodemo.o
connect                             echodemo.o
ld: fatal: Symbol referencing errors. No output written to echodemo
make: *** [echodemo] Error 1
elaine30:~/cs244a/pj1demo>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: Can't compile echodemo.c
Date: Sun, 20 Jan 2002 16:59:26 -0800
Lines: 36
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2573

When you link, you need to use the -lnsl and -lsocket flags.  You should use
the supplied Makefile to avoid worrying about this.  It produces the
following:

gcc -g -O0 -ansi -Wall -DSOLARIS  -c echodemo.c
gcc -g -O0 -ansi -Wall -DSOLARIS  -o echodemo  echodemo.o -lsocket -lnsl

Bret

"Kimball Chun-Jen Wang"  wrote in message

> Hello:
>
> I can't compile echodemo.c or any socket related c program.
> I think something is wrong with my .cshrc file.  Can some one tell
> me what to include for .cshrc?
>
> Kimball
>
> Here is the error message.
> elaine30:~/cs244a/pj1demo> make echodemo
> cc   echodemo.o   -o echodemo
> Undefined                       first referenced
>  symbol                             in file
> socket                              echodemo.o
> gethostbyname                       echodemo.o
> getservbyname                       echodemo.o
> inet_addr                           echodemo.o
> shutdown                            echodemo.o
> connect                             echodemo.o
> ld: fatal: Symbol referencing errors. No output written to echodemo
> make: *** [echodemo] Error 1
> elaine30:~/cs244a/pj1demo>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!wkimball
From: Kimball Chun-Jen Wang 
Newsgroups: su.class.cs244a
Subject: Can't gcc echodemo.c
Date: Sun, 20 Jan 2002 17:11:23 -0800
Lines: 13
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: elaine11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2574

elaine30:~/cs244a/pj1demo> gcc echodemo.c
Undefined                       first referenced
 symbol                             in file
socket                              /var/tmp/ccQmlPRj.o
gethostbyname                       /var/tmp/ccQmlPRj.o
getservbyname                       /var/tmp/ccQmlPRj.o
inet_addr                           /var/tmp/ccQmlPRj.o
shutdown                            /var/tmp/ccQmlPRj.o
connect                             /var/tmp/ccQmlPRj.o
ld: fatal: Symbol referencing errors. No output written to a.out
collect2: ld returned 1 exit status
elaine30:~/cs244a/pj1demo>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: LIST behavior?!
Date: Sun, 20 Jan 2002 17:23:11 -0800
Lines: 44
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2575

It seems to me that of the servers specified in the assignment handout, only
ftp.cs.stanford.edu is
ending lines with \r\n. Look at the output from ftp.fedworld.gov:

[t=116][o=111][t=116][a=97][l=108][ =32][8=56][
=10][d=100][r=114][w=119][x=120][r=114][-=45][x=120][r=114][-=45][x=120][ =3
2][ =32][ =32][8=56][ =32][1=49][0=48][4=52][ =32][ =32][ =32][ =32][ =32][ 
=32][1=49][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32]
[1=49][0=48][2=50][4=52][ =32][J=74][u=117][n=110][ =32][1=49][6=54][ =32][ 
=32][2=50][0=48][0=48][0=48][ =32][.=46][
=10][d=100][r=114][w=119][x=120][r=114][-=45][x=120][r=114][-=45][x=120][ =3
2][ =32][ =32][8=56][ =32][1=49][0=48][4=52][ =32][ =32][ =32][ =32][ =32][ 
=32][1=49][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32][ =32]
[1=49][0=48][2=50][4=52][ =32][J=74][u=117][n=110][ =32][1=49][6=54][ =32][ 
=32][2=50][0=48][0=48][0=48][ =32][.=46][.=46][

only character 10. The parsing strstr of the given code returns NULL on all
these servers. In comparison, here's that of the cs server:

[d=100][r=114][w=119][x=120][r=114][-=45][x=120][r=114][-=45][x=120][ =32][ 
=32][5=53][4=52][ =32][f=102][t=116][p=112][u=117][s=115][e=101][r=114][ =32
][ =32][f=102][t=116][p=112][u=117][s=115][e=101][r=114][s=115][ =32][ =32][


[ =32][ =32][1=49][0=48][2=50][4=52][ =32][N=78][o=111][v=118][ =32][1=49][4
=52][ =32][ =32][2=50][0=48][0=48][0=48][ =32][r=114][o=111][b=98][o=111][t=
116][i=10=13][99][s=115][
=10][d=100][r=114][w=119][x=120][r=114][-=45][x=120][r=114][-=45][x=120][ =3
2][ =32][ =32][2=50][ =32][f=102][t=116][p=112][u=117][s=115][e=101][r=114][

][ =32][f=102][t=116][p=112][u=117][s=115][e=101][r=114][s=115][ =32][ =32][

=32][ =32][ =32][ =32][5=53][1=49][2=50][ =32][J=74][u=117][n=110][ =32][1=4
9][6=54][ =32][ =32][1=49][9=57][9=57][3=51][ =32][t=116][h=104][e=101][o=11
1][r=114]=13][1][
=10]

Note the weird thing going on with the 13 before the 10.
Does anyone have a clue as to what's going on? Anyone noticed this before?

Thanks,
Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine30.Stanford.EDU!wkimball
From: Kimball Chun-Jen Wang 
Newsgroups: su.class.cs244a
Subject: Problem Solved Re: Can't gcc echodemo.c
Date: Sun, 20 Jan 2002 17:33:53 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
  
 
NNTP-Posting-Host: elaine30.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2576

On Sun, 20 Jan 2002, Kimball Chun-Jen Wang wrote:

> elaine30:~/cs244a/pj1demo> gcc echodemo.c
> Undefined                       first referenced
>  symbol                             in file
> socket                              /var/tmp/ccQmlPRj.o
> gethostbyname                       /var/tmp/ccQmlPRj.o
> getservbyname                       /var/tmp/ccQmlPRj.o
> inet_addr                           /var/tmp/ccQmlPRj.o
> shutdown                            /var/tmp/ccQmlPRj.o
> connect                             /var/tmp/ccQmlPRj.o
> ld: fatal: Symbol referencing errors. No output written to a.out
> collect2: ld returned 1 exit status
> elaine30:~/cs244a/pj1demo>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!rxuan
From: Ruozhong Xuan 
Newsgroups: su.class.cs244a
Subject: zero size file?
Date: Sun, 20 Jan 2002 18:11:55 -0800
Lines: 9
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine37.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Xin Wang 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2577

Hi,
   In assignment 1, if a remote file has the correct extension such as
"gz" but has size 0, should we ftp "copy" it and creat  the null size file
in the local?

Thanks.
Ruozhong Xuan


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Problem Solved Re: Can't gcc echodemo.c
Date: Sun, 20 Jan 2002 19:27:13 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
  
 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Kimball Chun-Jen Wang 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2578


You should use the makefile to compile echodemo.. the problem is that you are
not linking wiht -lnsl and -lsocket

            ~~m

> On Sun, 20 Jan 2002, Kimball Chun-Jen Wang wrote:
>
> > elaine30:~/cs244a/pj1demo> gcc echodemo.c
> > Undefined                       first referenced
> >  symbol                             in file
> > socket                              /var/tmp/ccQmlPRj.o
> > gethostbyname                       /var/tmp/ccQmlPRj.o
> > getservbyname                       /var/tmp/ccQmlPRj.o
> > inet_addr                           /var/tmp/ccQmlPRj.o
> > shutdown                            /var/tmp/ccQmlPRj.o
> > connect                             /var/tmp/ccQmlPRj.o
> > ld: fatal: Symbol referencing errors. No output written to a.out
> > collect2: ld returned 1 exit status
> > elaine30:~/cs244a/pj1demo>
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the .purify file.
Date: Sun, 20 Jan 2002 19:33:22 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2579

Hi,

There was a problem with the config file of the submit script, this is why 
it was not taking the .purify file. You will have to run the program 
/usr/class/cs244a/bin/submit TA hw1 again to have your .purify submitted.

I appologize for the inconvenience,

Pablo Molinero

On Sun, 20 Jan 2002, Shankar Agarwal wrote:

> Hi,
> When i am trying to submit the hw1 by using
> script /usr/class/cs244a/bin/submit TA hw1. It
> does not seem to pick up the .purify file. Has someone else seen the same
> problem or i am the only one having this problem. Also how can we supress
> the errors caused by the gethostbyname. The faq says that we should do
> suppress UMR ...;res_search
> suppress UMR ...;nss_search
> But it doesnot say where we are suppose to add this command.
> Shankar
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the .purify file.
Date: Sun, 20 Jan 2002 19:40:57 -0800
Lines: 50
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2580

The submit script's broken:

saga7:~/cs244a/proj1> /usr/class/cs244a/bin/submit akishan hw1
Error: Config line 33: Require: No assignment ID "hw2.1". (Must precede 
require: statement)

A problem statement has been e-mailed to the system
administrator 

-Sandeep

Pablo Molinero Fernandez wrote:

> Hi,
> 
> There was a problem with the config file of the submit script, this is why 
> it was not taking the .purify file. You will have to run the program 
> /usr/class/cs244a/bin/submit TA hw1 again to have your .purify submitted.
> 
> I appologize for the inconvenience,
> 
> Pablo Molinero
> 
> On Sun, 20 Jan 2002, Shankar Agarwal wrote:
> 
> 
>>Hi,
>>When i am trying to submit the hw1 by using
>>script /usr/class/cs244a/bin/submit TA hw1. It
>>does not seem to pick up the .purify file. Has someone else seen the same
>>problem or i am the only one having this problem. Also how can we supress
>>the errors caused by the gethostbyname. The faq says that we should do
>>suppress UMR ...;res_search
>>suppress UMR ...;nss_search
>>But it doesnot say where we are suppose to add this command.
>>Shankar
>>
>>
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the .purify file.
Date: Sun, 20 Jan 2002 19:58:58 -0800
Lines: 64
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Sandeep Tamhankar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2581


> The submit script's broken:

Yeah for me too :-(

elaine15:~/projects/myftp> /usr/class/cs244a/bin/submit
Error: Config line 33: Require: No assignment ID "hw2.1". (Must precede require: statement)

A problem statement has been e-mailed to the system
administrator 
elaine15:~/projects/myftp>

                ~~m

> saga7:~/cs244a/proj1> /usr/class/cs244a/bin/submit akishan hw1
> Error: Config line 33: Require: No assignment ID "hw2.1". (Must precede
> require: statement)
>
> A problem statement has been e-mailed to the system
> administrator 
>
> -Sandeep
>
> Pablo Molinero Fernandez wrote:
>
> > Hi,
> >
> > There was a problem with the config file of the submit script, this is why
> > it was not taking the .purify file. You will have to run the program
> > /usr/class/cs244a/bin/submit TA hw1 again to have your .purify submitted.
> >
> > I appologize for the inconvenience,
> >
> > Pablo Molinero
> >
> > On Sun, 20 Jan 2002, Shankar Agarwal wrote:
> >
> >
> >>Hi,
> >>When i am trying to submit the hw1 by using
> >>script /usr/class/cs244a/bin/submit TA hw1. It
> >>does not seem to pick up the .purify file. Has someone else seen the same
> >>problem or i am the only one having this problem. Also how can we supress
> >>the errors caused by the gethostbyname. The faq says that we should do
> >>suppress UMR ...;res_search
> >>suppress UMR ...;nss_search
> >>But it doesnot say where we are suppose to add this command.
> >>Shankar
> >>
> >>
> >>
> >
>
>
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Dead Test Script (again)
Date: Sun, 20 Jan 2002 21:03:56 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2582

Is it me or is the test script not working again??

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic13.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script (again)
Date: Sun, 20 Jan 2002 21:08:29 -0800
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Russell Greene 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2583


i just submitted and worked fine for me.

thanks, Arvind

On Sun, 20 Jan 2002, Russell Greene wrote:

> Is it me or is the test script not working again??
>
> --Russ
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script (again)
Date: Sun, 20 Jan 2002 21:10:49 -0800
Lines: 24
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2584

Wow.  I try to run the test script and it says that it cannot find any
..purify files in my directory (even though I have both the ftpcopy.purify
file and the purify.output file).  Am I missing anything?

--Russ

"Arvind Bhat"  wrote in message

>
> i just submitted and worked fine for me.
>
> thanks, Arvind
>
> On Sun, 20 Jan 2002, Russell Greene wrote:
>
> > Is it me or is the test script not working again??
> >
> > --Russ
> >
> >
> >
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: SSH timeouts
Date: Sun, 20 Jan 2002 21:19:29 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2585


Hello,

I am an SITN student and remotely connect to Stanford using SSH.
(OpenSSH_2.9p2, SSH protocols 1.5/2.0, OpenSSL 0x0090602f).

I am getting timed out very often (15-20 mins) no matter which
server (myth,epic,elaine) i am connected to. This does not happen
on all the days. But when it happens, it is quite annoying and
makes debugging extremely painful.

I talked to fellow sitn student. But his connection works ok for
same server. Wondering if someone has a solution/work around.

Thanks much,
Arvind

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script (again)
Date: Sun, 20 Jan 2002 21:23:40 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Russell Greene 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2586


i did a touch .purify to get the script going.

On Sun, 20 Jan 2002, Russell Greene wrote:

> Wow.  I try to run the test script and it says that it cannot find any
> .purify files in my directory (even though I have both the ftpcopy.purify
> file and the purify.output file).  Am I missing anything?
>
> --Russ
>
> "Arvind Bhat"  wrote in message
> 
> >
> > i just submitted and worked fine for me.
> >
> > thanks, Arvind
> >
> > On Sun, 20 Jan 2002, Russell Greene wrote:
> >
> > > Is it me or is the test script not working again??
> > >
> > > --Russ
> > >
> > >
> > >
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: Dead Test Script (again)
Date: Sun, 20 Jan 2002 21:52:12 -0800
Lines: 41
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2587

That did the trick.  Thanks!

--Russ


"Arvind Bhat"  wrote in message

>
> i did a touch .purify to get the script going.
>
> On Sun, 20 Jan 2002, Russell Greene wrote:
>
> > Wow.  I try to run the test script and it says that it cannot find any
> > .purify files in my directory (even though I have both the
ftpcopy.purify
> > file and the purify.output file).  Am I missing anything?
> >
> > --Russ
> >
> > "Arvind Bhat"  wrote in message
> > 
> > >
> > > i just submitted and worked fine for me.
> > >
> > > thanks, Arvind
> > >
> > > On Sun, 20 Jan 2002, Russell Greene wrote:
> > >
> > > > Is it me or is the test script not working again??
> > > >
> > > > --Russ
> > > >
> > > >
> > > >
> > >
> >
> >
> >
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga4.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: Can not submit test
Date: Sun, 20 Jan 2002 21:53:54 -0800
Lines: 30
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2588


I still can not submit test. This is the error message:

************************************************************************
SubmitBase: /afs/ir/class/cs244a/submissions
Debug Level = 0
This is your submission 1 (out of 5) for today (1-20-2002)
could not find any file matching .purify in
/afs/ir.stanford.edu/users/j/i/jinhui/cs244a/hw1!!
======================================================================
Your code could not be sumitted because of some error.
If you think there is a mistake, please send an email to your TA
(holliman;  with the relevant information.
======================================================================


Died at /afs/ir/class/cs244a/bin/submit_test.pl line 195.
***********************************************************************

But I do have ftpcopy.purify in my folder. What's wrong with it?


Thanks a lot!
-jinhui






.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: .purify is required now
Date: Sun, 20 Jan 2002 22:15:46 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2589

Hi,

From now on, the submit and submit_test.pl scripts need a .purify before 
they can proceed. The .purify was listed as part of the deliverables in 
the assignment handout, but the scripts were not taking it. Now they do.


Pablo Molinero

.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: submit script working now
Date: Sun, 20 Jan 2002 22:17:33 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2590

Hi,

Because of the change to require the .purify file, the submit script was 
broken for a few minutes (it complained that hw2.1 was not a valid 
assignment). This problem has been solved.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga2.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: submit_test.pl result question
Date: Sun, 20 Jan 2002 22:21:52 -0800
Lines: 50
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Ruozhong Xuan 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2591

Hi,

I am having the same problem. Any idea?

Thanks,

Moh'd
On Fri, 18 Jan 2002, Ruozhong Xuan wrote:

> Hi,
>     I used the submit_test.pl to request a test, and in a section,I got
> the below:
> ===============
>
>     * Test F.2: Does not transfer all of the matching files in the first
> level down to the
>       last specified level [out of 1]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> exit line said: "OK:1900222 bytes copied"
> ---
> diff between expected tree and your tree:
> 1,8d0
> < ./kermit/c/all.tar.Z
> < ./kermit/c/amiga.tar.Z
> < ./kermit/c/dg.tar.Z
> < ./kermit/c/mac.tar.Z
> < ./kermit/c/os2.tar.Z
> < ./kermit/c/os9.tar.Z
> < ./kermit/c/unix.tar.Z
> < ./kermit/c/vms.tar.Z
> 10d1
> < ./xcomm.tar.Z
>
>
>
> But when I manually tried out:
> ftpcopy Z 2 204.123.2.2 test /pub/comm
> I found all those files have been transfered to "test" directory under the
> correct subdirectory. So I am a little confused by the test result. Any
> comment?
>
> Thanks.
> Ruozhong Xuan
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Rearding the assignment 2
Date: Sun, 20 Jan 2002 23:06:57 -0800
Lines: 37
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2592


Hi,

The permissions of the directory for HW2 have been updated, so that you
can begin working on assigment #2 (if you want to start early).

Pablo

On Sat, 19 Jan 2002, Sandeep Tamhankar wrote:

> If you're really anxious to get started, go to 
> http://www.stanford.edu/class/cs244a/homeworks/hw2/gethw2
> 
> This script is what you would run if you had permission; just do by hand 
> (through your browser or wget) what this script says.
> 
> Or wait until the TAs move all this code to someplace reasonable like 
> /afs/ir/class/cs244a/homeworks so that we aren't forced to use wget and 
> such tools anytime we need anything.
> 
> -Sandeep
> 
> Shankar Agarwal wrote:
> 
> > Hi,
> > I am not able to access the /afs/ir/class/cs244a/WWW/homeworks/hw2/gethw2
> > script.
> > Thanks
> > Regards
> > Shankar
> > 
> > 
> 
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: submit script down
Date: Sun, 20 Jan 2002 23:30:18 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2593

I think the submit script is down again... i get the following error:
Died at /usr/class/cs244a/bin/submit_test.pl line 195.

Thanks,
Clayton
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: submit script down
Date: Sun, 20 Jan 2002 23:31:54 -0800
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2594

Sorry, as mentioned in a previous post, touch .purify did the trick

Clayton Pierce Jones wrote:
> 
> I think the submit script is down again... i get the following error:
> Died at /usr/class/cs244a/bin/submit_test.pl line 195.
> 
> Thanks,
> Clayton
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From:  (Dennis Lin)
Newsgroups: su.class.cs244a
Subject: Basic Question: PORT Command
Date: Mon, 21 Jan 2002 07:54:28 GMT
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: cpe-24-221-190-93.ca.sprintbbd.net
X-Newsreader: Forte Free Agent 1.21/32.243
Xref: nntp.stanford.edu su.class.cs244a:2595

Hi everybody,

I've been stuck on this problem for some time now, and I realize that
it is a basic question. 

For the PORT command, I know that we are supposed to send an IP
address and the port we are connecting to.  How do we get this
information to send out?  Seems like getsockname() doesn't really
work.  Thanks.


Dennis
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Basic Question: PORT Command
Date: 21 Jan 2002 08:17:19 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2596

You need to create a new socket with port specified to 0, and then bind to 
this port. Use getsockname() after the bind() to extract the assigned 
port. This is the port you will tell the server to connect to with the 
PORT command.

Arun

Dennis Lin  wrote:
: Hi everybody,

: I've been stuck on this problem for some time now, and I realize that
: it is a basic question. 

: For the PORT command, I know that we are supposed to send an IP
: address and the port we are connecting to.  How do we get this
: information to send out?  Seems like getsockname() doesn't really
: work.  Thanks.


: Dennis
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: quick questions
Date: Mon, 21 Jan 2002 00:19:28 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2597


I have 2 questions about the testscript files:

1. even though i submit purify.output all the time, this report says
purify.output is not submitted. Even the output of submit test script says
purify.output has been taken. please clarify.

2. ftpparse utility is unable to parse the list response from 204.123.2.2.
the list response has some zeros (0). May be it is because of that. Are
there any solutions to this?

anybody?
-anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From:  (Dennis Lin)
Newsgroups: su.class.cs244a
Subject: Re: Basic Question: PORT Command
Date: Mon, 21 Jan 2002 08:24:17 GMT
Lines: 88
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: cpe-24-221-190-93.ca.sprintbbd.net
X-Newsreader: Forte Free Agent 1.21/32.243
Xref: nntp.stanford.edu su.class.cs244a:2598

I really appreciate your help.  

I've tried that, and it still didn't work.  Here is what I have in my
code, and the reply I got from the terminal:


******************** BEGIN CODE *********************************


/**********************************************************************************************************/

/*  - bind an address to port 0 (ephermal port)
/**********************************************************************************************************/

localAddr.sin_family = AF_INET;
localAddr.sin_addr.s_addr = INADDR_ANY;
localAddr.sin_port = 0;

return_message = bind ( socket_in, (struct sockaddr *) &localAddr, 
	sizeof(localAddr));
printf ("Binding socket - return message: %d\n", return_message );
printf ("Bound serverPortNumber: %d\n", serverPortNumber );
printf ("Bound port - %d\n", localAddr.sin_port );

/**********************************************************************************************************/

/*  - call listen()

/**********************************************************************************************************/

return_message = listen ( socket_in, 32 );
printf ("Listening socket - return message: %d\n", return_message );

/**********************************************************************************************************/

/*  - call getsockname() to get port number this listening socket 
/**********************************************************************************************************/

return_message = getsockname ( socket_in, (struct sockaddr *) 
	&retrieveAddr, &socket_length_info);
return_message = getsockname ( socket_in, (struct sockaddr *)
	 &localAddr, &socket_length_info);

printf ("Control connection port: %d, return_message = %d\n",
	 servAddr.sin_port, return_message );

printf ("DEBUG -retrieved local address: %d.%d.%d.%d on PORT %d\n",
		localAddr.sin_addr.S_un.S_un_b.s_b1,
		localAddr.sin_addr.S_un.S_un_b.s_b2,
		localAddr.sin_addr.S_un.S_un_b.s_b3,
		localAddr.sin_addr.S_un.S_un_b.s_b4,
		localAddr.sin_port );



******************** BEGIN REPLY  *********************************
Binding socket - return message: 0
Bound serverPortNumber: 21
Bound port - 0
Listening socket - return message: 0
Control connection port: 21, return_message = 0
DEBUG - printing retrieved local address: 0.0.0.0 on PORT 0

******************** BEGIN CODE *********************************
On 21 Jan 2002 08:17:19 GMT, Arun Upadhyaya Kishan
 wrote:

>You need to create a new socket with port specified to 0, and then bind to 
>this port. Use getsockname() after the bind() to extract the assigned 
>port. This is the port you will tell the server to connect to with the 
>PORT command.
>
>Arun
>
>Dennis Lin  wrote:
>: Hi everybody,
>
>: I've been stuck on this problem for some time now, and I realize that
>: it is a basic question. 
>
>: For the PORT command, I know that we are supposed to send an IP
>: address and the port we are connecting to.  How do we get this
>: information to send out?  Seems like getsockname() doesn't really
>: work.  Thanks.
>
>
>: Dennis

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: Basic Question: PORT Command
Date: Mon, 21 Jan 2002 00:16:44 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2599

If you call getsockname on the control socket

size = sizeof(local_addr);
getsockname(control_socket, (struct sockaddr *) &local_addr, &size);

you should be able to access the 32 bit local IP address with

ntohl(local_addr.sin_addr.s_addr)

Bret

"Dennis Lin"  wrote in message

> Hi everybody,
>
> I've been stuck on this problem for some time now, and I realize that
> it is a basic question.
>
> For the PORT command, I know that we are supposed to send an IP
> address and the port we are connecting to.  How do we get this
> information to send out?  Seems like getsockname() doesn't really
> work.  Thanks.
>
>
> Dennis


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: Regarding Post of 1/16/02, 8:56: Script File Sizes
Date: Mon, 21 Jan 2002 00:31:43 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2600

Some random questions:

-Have any of you gotten the gotten the same bad results on that part of the
test that Bret mentioned? I've manually visited the sites where my tests
fail, and I can't figure out what's wrong. Any helpful stories of bugs
fixed?

-Did either of you perhaps have any problems with the exit line reports? My
test results read "exit line: ' ' " where I think they ought to have read
"OK: etc." Should they read "OK..."? Are points being deducted on that basis
(this is a question probably for a TA)?

-When the test lists diffs, do >'s mean that your tree had certain files and
that the correct one didn't (ie you shouldn't have had them)?

Thanks.
Steve



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Transferring files you don't have permissions for
Date: 21 Jan 2002 08:36:20 GMT
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2601

Yes, you can assume for a RETR comamnd 550 means permission denied. You 
should not create a 0-length file in this case, but should continue 
processing.

If you want to be ambitious, you can handle a temporary unavailable code 
and retry the file a couple times, though this is not required.

Arun

 Pete Belknap  wrote:
: The handout says that if we try to transfer a file which we don't have
: permissions for, we should just give up and continue from there.  I ran the
: ftpd server and got a 550 error for this case.

: Is 550 the only error code for permissions violations?  Can we assume that a
: 550 always means this (i.e. 550 would be the only error code that we don't
: terminate on) ?

: thanks,
: Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Basic Question: PORT Command
Date: 21 Jan 2002 08:40:33 GMT
Lines: 94
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2602

Make sure that the third argument to getsockname() is a variable of type 
socklen_t intiialized to sizeof(struct sockaddr).

Arun

Dennis Lin  wrote:
: I really appreciate your help.  

: I've tried that, and it still didn't work.  Here is what I have in my
: code, and the reply I got from the terminal:


: ******************** BEGIN CODE *********************************


: /**********************************************************************************************************/

: /*  - bind an address to port 0 (ephermal port)
: /**********************************************************************************************************/

: localAddr.sin_family = AF_INET;
: localAddr.sin_addr.s_addr = INADDR_ANY;
: localAddr.sin_port = 0;

: return_message = bind ( socket_in, (struct sockaddr *) &localAddr, 
: 	sizeof(localAddr));
: printf ("Binding socket - return message: %d\n", return_message );
: printf ("Bound serverPortNumber: %d\n", serverPortNumber );
: printf ("Bound port - %d\n", localAddr.sin_port );

: /**********************************************************************************************************/

: /*  - call listen()

: /**********************************************************************************************************/

: return_message = listen ( socket_in, 32 );
: printf ("Listening socket - return message: %d\n", return_message );

: /**********************************************************************************************************/

: /*  - call getsockname() to get port number this listening socket 
: /**********************************************************************************************************/

: return_message = getsockname ( socket_in, (struct sockaddr *) 
: 	&retrieveAddr, &socket_length_info);
: return_message = getsockname ( socket_in, (struct sockaddr *)
: 	 &localAddr, &socket_length_info);

: printf ("Control connection port: %d, return_message = %d\n",
: 	 servAddr.sin_port, return_message );

: printf ("DEBUG -retrieved local address: %d.%d.%d.%d on PORT %d\n",
: 		localAddr.sin_addr.S_un.S_un_b.s_b1,
: 		localAddr.sin_addr.S_un.S_un_b.s_b2,
: 		localAddr.sin_addr.S_un.S_un_b.s_b3,
: 		localAddr.sin_addr.S_un.S_un_b.s_b4,
: 		localAddr.sin_port );



: ******************** BEGIN REPLY  *********************************
: Binding socket - return message: 0
: Bound serverPortNumber: 21
: Bound port - 0
: Listening socket - return message: 0
: Control connection port: 21, return_message = 0
: DEBUG - printing retrieved local address: 0.0.0.0 on PORT 0

: ******************** BEGIN CODE *********************************
: On 21 Jan 2002 08:17:19 GMT, Arun Upadhyaya Kishan
:  wrote:

:>You need to create a new socket with port specified to 0, and then bind to 
:>this port. Use getsockname() after the bind() to extract the assigned 
:>port. This is the port you will tell the server to connect to with the 
:>PORT command.
:>
:>Arun
:>
:>Dennis Lin  wrote:
:>: Hi everybody,
:>
:>: I've been stuck on this problem for some time now, and I realize that
:>: it is a basic question. 
:>
:>: For the PORT command, I know that we are supposed to send an IP
:>: address and the port we are connecting to.  How do we get this
:>: information to send out?  Seems like getsockname() doesn't really
:>: work.  Thanks.
:>
:>
:>: Dennis

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From:  (Dennis Lin)
Newsgroups: su.class.cs244a
Subject: Re: Basic Question: PORT Command
Date: Mon, 21 Jan 2002 08:42:28 GMT
Lines: 105
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: cpe-24-221-190-93.ca.sprintbbd.net
X-Newsreader: Forte Free Agent 1.21/32.243
Xref: nntp.stanford.edu su.class.cs244a:2603

Hi guys,

Wow.  Thanks for all the help you've all given me.  This solved it!!


Dennis

On 21 Jan 2002 08:40:33 GMT, Arun Upadhyaya Kishan
 wrote:

>Make sure that the third argument to getsockname() is a variable of type 
>socklen_t intiialized to sizeof(struct sockaddr).
>
>Arun
>
>Dennis Lin  wrote:
>: I really appreciate your help.  
>
>: I've tried that, and it still didn't work.  Here is what I have in my
>: code, and the reply I got from the terminal:
>
>
>: ******************** BEGIN CODE *********************************
>
>
>: /**********************************************************************************************************/
>
>: /*  - bind an address to port 0 (ephermal port)
>: /**********************************************************************************************************/
>
>: localAddr.sin_family = AF_INET;
>: localAddr.sin_addr.s_addr = INADDR_ANY;
>: localAddr.sin_port = 0;
>
>: return_message = bind ( socket_in, (struct sockaddr *) &localAddr, 
>: 	sizeof(localAddr));
>: printf ("Binding socket - return message: %d\n", return_message );
>: printf ("Bound serverPortNumber: %d\n", serverPortNumber );
>: printf ("Bound port - %d\n", localAddr.sin_port );
>
>: /**********************************************************************************************************/
>
>: /*  - call listen()
>
>: /**********************************************************************************************************/
>
>: return_message = listen ( socket_in, 32 );
>: printf ("Listening socket - return message: %d\n", return_message );
>
>: /**********************************************************************************************************/
>
>: /*  - call getsockname() to get port number this listening socket 
>: /**********************************************************************************************************/
>
>: return_message = getsockname ( socket_in, (struct sockaddr *) 
>: 	&retrieveAddr, &socket_length_info);
>: return_message = getsockname ( socket_in, (struct sockaddr *)
>: 	 &localAddr, &socket_length_info);
>
>: printf ("Control connection port: %d, return_message = %d\n",
>: 	 servAddr.sin_port, return_message );
>
>: printf ("DEBUG -retrieved local address: %d.%d.%d.%d on PORT %d\n",
>: 		localAddr.sin_addr.S_un.S_un_b.s_b1,
>: 		localAddr.sin_addr.S_un.S_un_b.s_b2,
>: 		localAddr.sin_addr.S_un.S_un_b.s_b3,
>: 		localAddr.sin_addr.S_un.S_un_b.s_b4,
>: 		localAddr.sin_port );
>
>
>
>: ******************** BEGIN REPLY  *********************************
>: Binding socket - return message: 0
>: Bound serverPortNumber: 21
>: Bound port - 0
>: Listening socket - return message: 0
>: Control connection port: 21, return_message = 0
>: DEBUG - printing retrieved local address: 0.0.0.0 on PORT 0
>
>: ******************** BEGIN CODE *********************************
>: On 21 Jan 2002 08:17:19 GMT, Arun Upadhyaya Kishan
>:  wrote:
>
>:>You need to create a new socket with port specified to 0, and then bind to 
>:>this port. Use getsockname() after the bind() to extract the assigned 
>:>port. This is the port you will tell the server to connect to with the 
>:>PORT command.
>:>
>:>Arun
>:>
>:>Dennis Lin  wrote:
>:>: Hi everybody,
>:>
>:>: I've been stuck on this problem for some time now, and I realize that
>:>: it is a basic question. 
>:>
>:>: For the PORT command, I know that we are supposed to send an IP
>:>: address and the port we are connecting to.  How do we get this
>:>: information to send out?  Seems like getsockname() doesn't really
>:>: work.  Thanks.
>:>
>:>
>:>: Dennis
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga2.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Problem with TestScript
Date: Mon, 21 Jan 2002 00:53:39 -0800
Lines: 40
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2604

Hi,

I have two questions about the Test script.

First: What is the expected behavior of the program when it faced a linked
file in the same directory of the file. This is tested by:

 * Test L.5: Special cases: Symbolic-link-to-file [out of 0.5]
[SUMMARY] NOT OK
[RESULT]  Symbolic link to file in same directory did not transfer
(wrong!)

Can you please tell me what the behavior should be? Should we copy also
the link file when it points to a file in the same directory. I did this
and the test still not working.


Second: I passed all the other tests. However, I always failed this test:

/***********************************************************/
* Test D: Connects to server but does not transfer any matching files in
the specified levels [out of 3.5]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie
exit line said: "ERROR: Transiant Negative reply from server "
/***********************************************************/

However, when I run the program manually for the given command:

ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie

It ALWAYS works! I sent the test three times and it gave the above error
message. I have no way to debug it since it is not occuring with me! Any
suggestion please.

Thanks,

Moh'd

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!araik
From: Araik Grigoryan 
Newsgroups: su.class.cs244a
Subject: Re: quick questions
Date: Mon, 21 Jan 2002 00:57:07 -0800
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2605



> 1. even though i submit purify.output all the time, this report says
> purify.output is not submitted. Even the output of submit test script says
> purify.output has been taken. please clarify.

Sorry, that's for the TAs to fix.

> 2. ftpparse utility is unable to parse the list response from 204.123.2.2.
> the list response has some zeros (0). May be it is because of that. Are
> there any solutions to this?

I do know the possible answer to this ... This FTP server sends data that
is terminated with "\n" and NOT with "\r\n". I sent an email to Nick
mentioning this but did not get a response yet. Meanwhile, you may want to
put in a hack to take care of this. The fix does not involve messing with
ftpparse, btw, just your own code.

araik

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!araik
From: Araik Grigoryan 
Newsgroups: su.class.cs244a
Subject: file separator
Date: Mon, 21 Jan 2002 01:00:11 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2606


Hi,

Does anyone know if there is some global string defined in a C library
that contains a character that separates file names, e.g. "/" or "\"?
Thanks.

araik

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga2.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Test Functionality
Date: Mon, 21 Jan 2002 01:12:58 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2607

Hi all,

Anyone can tell me what is the required functionality of this test:

 * Test I: User interface specification:  violates output format [out of
0.5]


Thanks,

Moh'd

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Submit Script
Date: Mon, 21 Jan 2002 10:27:38 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2608

Has anyone received the error when running the submit script?

Submission succeeded
Error: log: open: Permission denied


--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: Problem with TestScript
Date: Mon, 21 Jan 2002 10:31:03 -0800
Lines: 57
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2609

I can tell you a little more about the ftp.microsoft.com issue.  Your
program is aborting because the server is full.  I can only speak from
experience but I am having the same problems.  However, I seem to almost
always be able to connect to microsoft from a saga machine but never from my
home machine (I get the server full error from my home machine when manually
trying to open a connection).  Since the tests are run from the leland
cluster though, that doesn't support my theory.  I guess I'm just letting
you know that you aren't the only one having problems with Microsoft ;)

--Russ



"Moh'd Saleem Saleem Alrawashdeh"  wrote in message

> Hi,
>
> I have two questions about the Test script.
>
> First: What is the expected behavior of the program when it faced a linked
> file in the same directory of the file. This is tested by:
>
>  * Test L.5: Special cases: Symbolic-link-to-file [out of 0.5]
> [SUMMARY] NOT OK
> [RESULT]  Symbolic link to file in same directory did not transfer
> (wrong!)
>
> Can you please tell me what the behavior should be? Should we copy also
> the link file when it points to a file in the same directory. I did this
> and the test still not working.
>
>
> Second: I passed all the other tests. However, I always failed this test:
>
> /***********************************************************/
> * Test D: Connects to server but does not transfer any matching files in
> the specified levels [out of 3.5]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie
> exit line said: "ERROR: Transiant Negative reply from server "
> /***********************************************************/
>
> However, when I run the program manually for the given command:
>
> ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie
>
> It ALWAYS works! I sent the test three times and it gave the above error
> message. I have no way to debug it since it is not occuring with me! Any
> suggestion please.
>
> Thanks,
>
> Moh'd
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!araik
From: Araik Grigoryan 
Newsgroups: su.class.cs244a
Subject: RE: non-NVT ASCII replies (fwd)
Date: Mon, 21 Jan 2002 10:58:05 -0800
Lines: 45
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2610


Here's what I got from Antonios.

---------- Forwarded message ----------
Date: Mon, 21 Jan 2002 09:22:50 -0800
From: Antonios Hondroulis 
To: Araik Grigoryan 
Subject: RE: non-NVT ASCII replies


Hi Araik,

No, only what the RFC says is what we expect you to handle.

Please try to post on the newsgroup your questions. This way
it's more beneficial for everyone.

Thanks,
Antonios.

-----Original Message-----
From: Araik Grigoryan 
Sent: Sunday, January 20, 2002 5:44 PM
To: 
Subject: non-NVT ASCII replies



Hi Nick,

Got a question ... I ran my program with the test script and there is one
test (ftpcopy Z 2 204.123.2.2 localdir /pub/comm) that gives me a problem
because the replies that it returns are not terminated with "\r\n" but
only with "\n". According to the assignment, we don't have to deal with
this but I wanted to confirm with you. Btw, it's not a bug in my code, I
confirmed it with another student (which still does not really prove
anything, but I am pretty certain that the replies are in non-NVT ASCII
format). So should we follow the assignment and ignore this case or put in
a hack to handle it?

Thanks,

araik


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Submit Script
Date: Mon, 21 Jan 2002 11:02:59 -0800
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:2611

Hi,

If you saw an error like this, the error is minor and it did not prevent 
you from submitting your code, in other words, we received your files 
correctly. This problem has been solved, so you will not see it again, 
if you submit now.

Pablo

Russell Greene wrote:

> Has anyone received the error when running the submit script?
> 
> Submission succeeded
> Error: log: open: Permission denied
> 
> 
> --Russ
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!rxuan
From: Ruozhong Xuan 
Newsgroups: su.class.cs244a
Subject: Re: submit_test.pl result question
Date: Mon, 21 Jan 2002 11:12:40 -0800
Lines: 61
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: elaine37.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Moh'd Saleem Saleem Alrawashdeh 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2612

Hi Moh'd:
    I don't know why that happens. In my test result, I failed at most of
the items from D to F. But when I manually tried out, I got the expected
result, and my local file tree got created as expected.

--ruozhong xuan


 On Sun, 20 Jan 2002, Moh'd Saleem Saleem Alrawashdeh wrote:

> Hi,
>
> I am having the same problem. Any idea?
>
> Thanks,
>
> Moh'd
> On Fri, 18 Jan 2002, Ruozhong Xuan wrote:
>
> > Hi,
> >     I used the submit_test.pl to request a test, and in a section,I got
> > the below:
> > ===============
> >
> >     * Test F.2: Does not transfer all of the matching files in the first
> > level down to the
> >       last specified level [out of 1]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> > exit line said: "OK:1900222 bytes copied"
> > ---
> > diff between expected tree and your tree:
> > 1,8d0
> > < ./kermit/c/all.tar.Z
> > < ./kermit/c/amiga.tar.Z
> > < ./kermit/c/dg.tar.Z
> > < ./kermit/c/mac.tar.Z
> > < ./kermit/c/os2.tar.Z
> > < ./kermit/c/os9.tar.Z
> > < ./kermit/c/unix.tar.Z
> > < ./kermit/c/vms.tar.Z
> > 10d1
> > < ./xcomm.tar.Z
> >
> >
> >
> > But when I manually tried out:
> > ftpcopy Z 2 204.123.2.2 test /pub/comm
> > I found all those files have been transfered to "test" directory under the
> > correct subdirectory. So I am a little confused by the test result. Any
> > comment?
> >
> > Thanks.
> > Ruozhong Xuan
> >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: quick questions
Date: Mon, 21 Jan 2002 11:33:32 -0800
Lines: 35
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:2613



Araik Grigoryan wrote:

> 
>>1. even though i submit purify.output all the time, this report says
>>purify.output is not submitted. Even the output of submit test script says
>>purify.output has been taken. please clarify.
>>
> 
> Sorry, that's for the TAs to fix.
> 


This problem has been fixed.

Pablo


> 
>>2. ftpparse utility is unable to parse the list response from 204.123.2.2.
>>the list response has some zeros (0). May be it is because of that. Are
>>there any solutions to this?
>>
> 
> I do know the possible answer to this ... This FTP server sends data that
> is terminated with "\n" and NOT with "\r\n". I sent an email to Nick
> mentioning this but did not get a response yet. Meanwhile, you may want to
> put in a hack to take care of this. The fix does not involve messing with
> ftpparse, btw, just your own code.
> 
> araik
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Sriram Viji" 
Newsgroups: su.class.cs244a
Subject: error creating data sockets
Date: Mon, 21 Jan 2002 11:48:29 -0800
Lines: 28
Distribution: su
Message-ID: 
NNTP-Posting-Host: glaurung.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2615.200
X-MIMEOLE: Produced By Microsoft MimeOLE V5.00.2615.200
Xref: nntp.stanford.edu su.class.cs244a:2614


hi,

i'm having problems opening data sockets. when i try to do a recursive list,
it works fine for a while and then runs into problems trying to open a new
socket, am i running into problems trying to open too many sockets?

data connection involves:
getting a socket using the socket() command,  <- this is where i get errors
later on
binding it using bind()
listening on that using listen()
getting the port number using getsockname()
sending PORT over the control connection
accepting the connection using accept()
sending LIST over the control connection
reading the data and when we're done with this the server closes the
connection

do we also have to close the connection from our end using shutdown() and
close()?
do we have to create a socket for each new list command?
the purify version gives me an error, it says 128 file descriptors in use

thanks,
sriram


.

Path: shelby.stanford.edu!nntp.stanford.edu!cardinal5.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: Question about mkdir
Date: Mon, 21 Jan 2002 13:11:47 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: cardinal5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2615

I have implemented most of the things, but I have one question left.
According to the, "You should overwrite any files which already exist
without prompting". And when you want to construct a directory, but there
is one directory with the same name, I think we should over write the old
directory.
But if we use mkdir to construct the new directory, we can't overwrite the
old directory. It can only incur and error and exit.
How to handle this problem? Do we need to delete the old directory(and
all the files under it), and then use mkdir to construct the new
directory?
Thanks

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine21.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: non-NVT ASCII replies (fwd)
Date: 21 Jan 2002 21:13:52 GMT
Lines: 48
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine21.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2616


This server is correctly giving me NVT ASCII when I try.

Matthew


>---------- Forwarded message ----------
>Date: Mon, 21 Jan 2002 09:22:50 -0800
>From: Antonios Hondroulis 
>To: Araik Grigoryan 
>Subject: RE: non-NVT ASCII replies


>Hi Araik,

>No, only what the RFC says is what we expect you to handle.

>Please try to post on the newsgroup your questions. This way
>it's more beneficial for everyone.

>Thanks,
>Antonios.

>-----Original Message-----
>From: Araik Grigoryan 
>Sent: Sunday, January 20, 2002 5:44 PM
>To: 
>Subject: non-NVT ASCII replies



>Hi Nick,

>Got a question ... I ran my program with the test script and there is one
>test (ftpcopy Z 2 204.123.2.2 localdir /pub/comm) that gives me a problem
>because the replies that it returns are not terminated with "\r\n" but
>only with "\n". According to the assignment, we don't have to deal with
>this but I wanted to confirm with you. Btw, it's not a bug in my code, I
>confirmed it with another student (which still does not really prove
>anything, but I am pretty certain that the replies are in non-NVT ASCII
>format). So should we follow the assignment and ignore this case or put in
>a hack to handle it?

>Thanks,

>araik


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Antonios Hondroulis 
Newsgroups: su.class.cs244a
Subject: Re: error creating data sockets
Date: Mon, 21 Jan 2002 13:15:21 -0800
Organization: Stanford University
Lines: 40
Distribution: su
Message-ID: 
References: 
Reply-To: 
NNTP-Posting-Host: sul-hd-adam.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; x-mac-type="54455854"; x-mac-creator="4D4F5353"
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.77 (Macintosh; U; PPC)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2617



Sriram,

 You're running out of file-descriptors by not closing your sockets.
 Please take a look at the code in page 113 of Stevens.
 Also, please keep in mind that the file descriptor returned by
 accept is different from the one you used to LISTEN to.

 Regards,
 Antonios.

Sriram Viji wrote:

> hi,
>
> i'm having problems opening data sockets. when i try to do a recursive list,
> it works fine for a while and then runs into problems trying to open a new
> socket, am i running into problems trying to open too many sockets?
>
> data connection involves:
> getting a socket using the socket() command,  <- this is where i get errors
> later on
> binding it using bind()
> listening on that using listen()
> getting the port number using getsockname()
> sending PORT over the control connection
> accepting the connection using accept()
> sending LIST over the control connection
> reading the data and when we're done with this the server closes the
> connection
>
> do we also have to close the connection from our end using shutdown() and
> close()?
> do we have to create a socket for each new list command?
> the purify version gives me an error, it says 128 file descriptors in use
>
> thanks,
> sriram

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine21.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Question about mkdir
Date: 21 Jan 2002 21:16:54 GMT
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine21.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2618

Fang Sun  writes:

>I have implemented most of the things, but I have one question left.
>According to the, "You should overwrite any files which already exist
>without prompting". And when you want to construct a directory, but there
>is one directory with the same name, I think we should over write the old
>directory.
>But if we use mkdir to construct the new directory, we can't overwrite the
>old directory. It can only incur and error and exit.
>How to handle this problem? Do we need to delete the old directory(and
>all the files under it), and then use mkdir to construct the new
>directory?
>Thanks


You can check errno to determine the reason for mkdir() failing; if it's
because the directory exists, you shouldn't treat the case as an error.

.

Path: shelby.stanford.edu!nntp.stanford.edu!cardinal5.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: Re: Question about mkdir
Date: Mon, 21 Jan 2002 14:42:57 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: cardinal5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2619

Thanks.
And after checking the errno, we should delete the directory if it exists,
right? And if the directory includes files and subdirectories, we should
delete them all recursively.
Is that right?


 On 21 Jan 2002, Matthew Jonathan Holliman
wrote:

> Fang Sun  writes:
>
> >I have implemented most of the things, but I have one question left.
> >According to the, "You should overwrite any files which already exist
> >without prompting". And when you want to construct a directory, but there
> >is one directory with the same name, I think we should over write the old
> >directory.
> >But if we use mkdir to construct the new directory, we can't overwrite the
> >old directory. It can only incur and error and exit.
> >How to handle this problem? Do we need to delete the old directory(and
> >all the files under it), and then use mkdir to construct the new
> >directory?
> >Thanks
>
>
> You can check errno to determine the reason for mkdir() failing; if it's
> because the directory exists, you shouldn't treat the case as an error.
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine31.Stanford.EDU!htimam
From:  (Hasan Taufiq Imam)
Newsgroups: su.class.cs244a
Subject: file extensions.
Date: 21 Jan 2002 23:11:39 GMT
Organization: Stanford University, CA 94305, USA
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine31.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2620

xxx.tar.gz an acceptable file if user specification is just
..gz. Need clarification.
Thanks,
Hasan  

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Debashis Sahoo" 
Newsgroups: su.class.cs244a
Subject: broken pipe
Date: Mon, 21 Jan 2002 15:46:33 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: dsahoo.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MIMEOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2621

I am receiving following error from the automatic test script.
I am getting broken pipe while calling write to socket in this case.
Do we have to catch SIGPIPE signal in this case?
How do we handle this case ...any ideas?

    * Test I: User interface specification:  violates output format [out of
0.5]

[SUMMARY] NOT OK
[RESULT]
test used: ftpcopy gz 2 epic11.stanford.edu localdir
exit line said: ""

Thanks,
Sahoo


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic18.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: Default remoteBaseDir ??
Date: Mon, 21 Jan 2002 16:00:22 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic18.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2622


This one for the TA's:

   Can we assume the default remote server's base directory specified be
always '/' or can default directory be any other directory in general?
   The ftp servers mentioned in the problem statement all default to '/',
so is this a rule or mere coincidence??

-gaurav

*************************************************************************
Gaurav Garg                          * Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             *       Escondido Village, Stanford
Department of Electrical Engineering *       CA-94305
Stanford University                  * Ph:650-498-1208
*************************************************************************

-Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Pete Belknap" 
Newsgroups: su.class.cs244a
Subject: Symbolic Links and File Extensions
Date: Mon, 21 Jan 2002 16:23:24 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2623

What should we do when the symbolic link has a diff extension than the file
it points to?

So, for example, if we want to download .txt files and we have the link

link.lnk  ->  readme.txt

Should we download it?  Of course, the opposite question is, if we are
interested in files of extension .lnk, should we download the link?  It
seems to me that the answer to my first question should be no and the second
should be yes...

thanks,
Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: ftpcopy Z 2 204.123.2.2 localdir /pub/comm ??
Date: Mon, 21 Jan 2002 16:45:37 -0800
Lines: 67
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2624

Hey Folks,

I'm failing the test: ftpcopy Z 2 204.123.2.2 localdir /pub/comm

In the test_submit script...

When I run my program (which runs fine on other sites) on this
site I get the following last few lines of OUTPUT:

-----> CWD c
 
<--- 250 CWD command successful.
 
-----> PORT 171,64,15,142,236,108
 
<--- 200 PORT command successful.
 
-----> LIST
 
<--- 150 Opening ASCII mode data connection for /bin/ls.
 
<--- 226 Transfer complete.
 
<--- total 1714
 
<--- lrwxrwxrwx   1 0        0             37 Apr 14  1995 00README-Le
gal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs
 
*** file Link: 00README-Legal-Rules-Regs -> ../../../../00README-Legal
-Rules-Regs
<--- -r--r--r--   1 0        0            935 Jan 17  1989 README
 
<--- -r--r--r--   1 0        0          72935 Jan 16  1989 all.tar.Z
 
-----> PORT 171,64,15,142,236,109
 
<--- 200 PORT command successful.
 
-----> TYPE I
 
<--- 200 Type set to I.
 
-----> RETR all.tar.Z
 
<--- 150 Opening BINARY mode data connection for all.tar.Z (72935 byte
s).
 
******* Here it just hangs indefinitely *****

In my readReply code, when I encounter a "1**" reply, I do another
read from the control socket, and I think its probably hanging
waiting for a reply from the server that never comes...

The assignment handout states that a "1**" indicates a positive
preliminary reply, but to "expect another reply before sending the next
command" -

Should I be reading the other (promised) reply line from the connection
socket
when I do a readReply, or before sending the next command?

What I currently have works for the other sites, but this
one is acting up...

Comments? Suggestions?

Cheers,
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga0.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: ftpcopy Z 2 204.123.2.2 localdir /pub/comm ??
Date: Mon, 21 Jan 2002 17:02:27 -0800
Lines: 89
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga0.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2625

hi

I am passing the test. But the test script says it is not.

Any ideas?

thanks
abhishek

On Mon, 21 Jan 2002, Luis Robles wrote:

> Hey Folks,
>
> I'm failing the test: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
>
> In the test_submit script...
>
> When I run my program (which runs fine on other sites) on this
> site I get the following last few lines of OUTPUT:
>
> -----> CWD c
>
> <--- 250 CWD command successful.
>
> -----> PORT 171,64,15,142,236,108
>
> <--- 200 PORT command successful.
>
> -----> LIST
>
> <--- 150 Opening ASCII mode data connection for /bin/ls.
>
> <--- 226 Transfer complete.
>
> <--- total 1714
>
> <--- lrwxrwxrwx   1 0        0             37 Apr 14  1995 00README-Le
> gal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs
>
> *** file Link: 00README-Legal-Rules-Regs -> ../../../../00README-Legal
> -Rules-Regs
> <--- -r--r--r--   1 0        0            935 Jan 17  1989 README
>
> <--- -r--r--r--   1 0        0          72935 Jan 16  1989 all.tar.Z
>
> -----> PORT 171,64,15,142,236,109
>
> <--- 200 PORT command successful.
>
> -----> TYPE I
>
> <--- 200 Type set to I.
>
> -----> RETR all.tar.Z
>
> <--- 150 Opening BINARY mode data connection for all.tar.Z (72935 byte
> s).
>
> ******* Here it just hangs indefinitely *****
>
> In my readReply code, when I encounter a "1**" reply, I do another
> read from the control socket, and I think its probably hanging
> waiting for a reply from the server that never comes...
>
> The assignment handout states that a "1**" indicates a positive
> preliminary reply, but to "expect another reply before sending the next
> command" -
>
> Should I be reading the other (promised) reply line from the connection
> socket
> when I do a readReply, or before sending the next command?
>
> What I currently have works for the other sites, but this
> one is acting up...
>
> Comments? Suggestions?
>
> Cheers,
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Dir and files with same names...
Date: Mon, 21 Jan 2002 17:35:56 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2626


 I wonder what I should do with dirs and files with the same name..

1) There may be a file with the same name as the specified local
directory...
   Then shall I delete the file and then mkdir?
   Or shall I just report error and exit?
  ( I think the latter would make sense..)

2) It is stated that dir's and file's with the same name be overwritten..
   Then, does it mean that I have to delete the original file or dir
   when there's a file with the same name as the dir I'm going to make,
  or when there's a dir with the same name as the file I'm going to create..



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine31.Stanford.EDU!htimam
From:  (Hasan Taufiq Imam)
Newsgroups: su.class.cs244a
Subject: Other files in directory.
Date: 22 Jan 2002 01:42:36 GMT
Organization: Stanford University, CA 94305, USA
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine31.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2627

Please, clarify. 
In case of non-empty local directory do we delete files or directories
as required or do we delete everything. 
In other words, can the local directory contain any files or directories other
than the tree copied from the remote server. 

I still don't know if the user specifies extension gz, is tar.gz a valid 
extension or not.
Thanks.  
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!araik
From: Araik Grigoryan 
Newsgroups: su.class.cs244a
Subject: Re: non-NVT ASCII replies (fwd)
Date: Mon, 21 Jan 2002 17:43:46 -0800
Lines: 59
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2628


On the data connection or the cotrol connection or both? The control
connection seems to be fine, it's the data connection that returns non-NVT
ASCII replies. Matt, I think you mentioned that you are using your own
ftpcopy program. Are you sure you are not specifically programming around
this problem?

On 21 Jan 2002, Matthew Jonathan Holliman wrote:

>
> This server is correctly giving me NVT ASCII when I try.
>
> Matthew
>
>
> >---------- Forwarded message ----------
> >Date: Mon, 21 Jan 2002 09:22:50 -0800
> >From: Antonios Hondroulis 
> >To: Araik Grigoryan 
> >Subject: RE: non-NVT ASCII replies
>
>
> >Hi Araik,
>
> >No, only what the RFC says is what we expect you to handle.
>
> >Please try to post on the newsgroup your questions. This way
> >it's more beneficial for everyone.
>
> >Thanks,
> >Antonios.
>
> >-----Original Message-----
> >From: Araik Grigoryan 
> >Sent: Sunday, January 20, 2002 5:44 PM
> >To: 
> >Subject: non-NVT ASCII replies
>
>
>
> >Hi Nick,
>
> >Got a question ... I ran my program with the test script and there is one
> >test (ftpcopy Z 2 204.123.2.2 localdir /pub/comm) that gives me a problem
> >because the replies that it returns are not terminated with "\r\n" but
> >only with "\n". According to the assignment, we don't have to deal with
> >this but I wanted to confirm with you. Btw, it's not a bug in my code, I
> >confirmed it with another student (which still does not really prove
> >anything, but I am pretty certain that the replies are in non-NVT ASCII
> >format). So should we follow the assignment and ignore this case or put in
> >a hack to handle it?
>
> >Thanks,
>
> >araik
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!araik
From: Araik Grigoryan 
Newsgroups: su.class.cs244a
Subject: Re: broken pipe
Date: Mon, 21 Jan 2002 17:47:45 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2629


Yep, had that problem ... Basically, epic11 does not accept FTP
connections but when it tells you that it doesn't, it's reply string is
not in standard reply format, i.e. "xyz Some text", it just gives you the
text without any numerical code and closes the connection. You just have
to program around that and you'll be fine.

On Mon, 21 Jan 2002, Debashis Sahoo wrote:

> I am receiving following error from the automatic test script.
> I am getting broken pipe while calling write to socket in this case.
> Do we have to catch SIGPIPE signal in this case?
> How do we handle this case ...any ideas?
>
>     * Test I: User interface specification:  violates output format [out of
> 0.5]
>
> [SUMMARY] NOT OK
> [RESULT]
> test used: ftpcopy gz 2 epic11.stanford.edu localdir
> exit line said: ""
>
> Thanks,
> Sahoo
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Jason Ahmad 
Newsgroups: su.class.cs244a
Subject: error cleanup
Date: Mon, 21 Jan 2002 18:46:33 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic13.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2630

What action, exactly, are we to take on general error cases (can't
allocate socket, can't make connection etc)? It is clear we are to write
a brief explaination to stderr and exit, but are we to free malloced
memory and make sure all sockets are closed and whatnot?

Thanks,
Jason
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: ftpcopy Z 2 204.123.2.2 localdir /pub/comm ??
Date: Mon, 21 Jan 2002 19:08:55 -0800
Lines: 79
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.5) Gecko/20011012
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2631

You're not supposed to wait for the next reply right away.  You're 
supposed to read all the data off the data socket; then the ftp server 
will acknowledge that the transfer is complete by issuing the final 
response...so, read all the data, THEN read from the control socket for 
the final response.

-Sandeep

Luis Robles wrote:

> Hey Folks,
> 
> I'm failing the test: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> 
> In the test_submit script...
> 
> When I run my program (which runs fine on other sites) on this
> site I get the following last few lines of OUTPUT:
> 
> -----> CWD c
>  
> <--- 250 CWD command successful.
>  
> -----> PORT 171,64,15,142,236,108
>  
> <--- 200 PORT command successful.
>  
> -----> LIST
>  
> <--- 150 Opening ASCII mode data connection for /bin/ls.
>  
> <--- 226 Transfer complete.
>  
> <--- total 1714
>  
> <--- lrwxrwxrwx   1 0        0             37 Apr 14  1995 00README-Le
> gal-Rules-Regs -> ../../../../00README-Legal-Rules-Regs
>  
> *** file Link: 00README-Legal-Rules-Regs -> ../../../../00README-Legal
> -Rules-Regs
> <--- -r--r--r--   1 0        0            935 Jan 17  1989 README
>  
> <--- -r--r--r--   1 0        0          72935 Jan 16  1989 all.tar.Z
>  
> -----> PORT 171,64,15,142,236,109
>  
> <--- 200 PORT command successful.
>  
> -----> TYPE I
>  
> <--- 200 Type set to I.
>  
> -----> RETR all.tar.Z
>  
> <--- 150 Opening BINARY mode data connection for all.tar.Z (72935 byte
> s).
>  
> ******* Here it just hangs indefinitely *****
> 
> In my readReply code, when I encounter a "1**" reply, I do another
> read from the control socket, and I think its probably hanging
> waiting for a reply from the server that never comes...
> 
> The assignment handout states that a "1**" indicates a positive
> preliminary reply, but to "expect another reply before sending the next
> command" -
> 
> Should I be reading the other (promised) reply line from the connection
> socket
> when I do a readReply, or before sending the next command?
> 
> What I currently have works for the other sites, but this
> one is acting up...
> 
> Comments? Suggestions?
> 
> Cheers,
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga0.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: empty files
Date: Mon, 21 Jan 2002 19:18:01 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga0.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2632


hi

I am trying to transfer empty files using ftpd and i am being successful
in doing so i.e no file created.
However, the test script says that I transferred 7 bytes on two
occassions. Can anyone suggest how di I test this? as i said, using ftpd
says i am OK.

thanks
abhishek

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!ambu
From: Ambika Balakrishnan 
Newsgroups: su.class.cs244a
Subject: location of testing dir
Date: Mon, 21 Jan 2002 20:14:24 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2633


Hello,

In the handout "Testing homework 1" there are 6 examples and I dont know
how to run examples 1 and 2, which refer to a permissionsDir and an
exampleDir. Where are these directories? Am I supposed to create them? Am
I supposed to run the supplied ftpd for these examples?

Thanks,
Ambika

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Submit Script
Date: Mon, 21 Jan 2002 20:36:04 -0800
Lines: 7
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Pablo Molinero Fernandez 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2634

Hi,
The submission script is still giving some problem. I am having the
ftpcopy.purify in the directory but its saying that .purify not found.
Please look into the matter.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: LIST response in IMAGE type?
Date: Mon, 21 Jan 2002 20:33:25 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2635

 When I set the TYPE to I and sent the LIST command,
 the response doesn't seem to be in NVT-ASCII format.
 Although I may experiment with it and make things work,
 I'm afraid LIST response in IMAGE type may have no standard.

 But, I think switching between NVT-ASCII for LIST and IMAGE for RETR is too
cumbersome...
 I wish that I can use TYPE I exclusively...

 Could anybody help?



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Submit Fails
Date: Mon, 21 Jan 2002 20:43:29 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2636

Hi,
I am trying to submit hw1 and it has ftpcopy.purify in that directory but
it still says that no .purify is present.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Mon, 21 Jan 2002 21:02:46 -0800
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2637


It seems to me that ftpcopy.purify and .purify are two different files.

Alex

On Mon, 21 Jan 2002, Shankar Agarwal wrote:

> Hi,
> I am trying to submit hw1 and it has ftpcopy.purify in that directory but
> it still says that no .purify is present.
> Thanks
> Shankar
>
>

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Test Result
Date: Mon, 21 Jan 2002 21:33:37 -0800
Lines: 31
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2638

Hi,

I am posting this again since I didn't get a reply for my first post. Can
anyone please tell me what this test means?

===============

    * Test L.5: Special cases: Symbolic-link-to-file [out of 0.5]

[SUMMARY] NOT OK
[RESULT]  Symbolic link to file in same directory did not transfer
(wrong!)
exit line said: ""
---
Your output to stdout was:

OK: 14 bytes copied

---


===============

I appreciate any clarification of what special case that this test is
testing?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: Re: non-NVT ASCII replies (fwd)
Date: Mon, 21 Jan 2002 21:51:28 -0800
Lines: 65
Distribution: su
Message-ID: 
References: 
	  
NNTP-Posting-Host: saga20.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2639

I've run into the same thing, but i'm apt to blame myself. Using ftpd,
I've found that I am sometimes getting list responses off the data
connection that end in "\r\n" and that sometimes end in "\n". This
inconsistency makes me think there's a bug in my code. Suggestions on
this would be quite helpful.

Araik Grigoryan wrote:
> 
> On the data connection or the cotrol connection or both? The control
> connection seems to be fine, it's the data connection that returns non-NVT
> ASCII replies. Matt, I think you mentioned that you are using your own
> ftpcopy program. Are you sure you are not specifically programming around
> this problem?
> 
> On 21 Jan 2002, Matthew Jonathan Holliman wrote:
> 
> >
> > This server is correctly giving me NVT ASCII when I try.
> >
> > Matthew
> >
> >
> > >---------- Forwarded message ----------
> > >Date: Mon, 21 Jan 2002 09:22:50 -0800
> > >From: Antonios Hondroulis 
> > >To: Araik Grigoryan 
> > >Subject: RE: non-NVT ASCII replies
> >
> >
> > >Hi Araik,
> >
> > >No, only what the RFC says is what we expect you to handle.
> >
> > >Please try to post on the newsgroup your questions. This way
> > >it's more beneficial for everyone.
> >
> > >Thanks,
> > >Antonios.
> >
> > >-----Original Message-----
> > >From: Araik Grigoryan 
> > >Sent: Sunday, January 20, 2002 5:44 PM
> > >To: 
> > >Subject: non-NVT ASCII replies
> >
> >
> >
> > >Hi Nick,
> >
> > >Got a question ... I ran my program with the test script and there is one
> > >test (ftpcopy Z 2 204.123.2.2 localdir /pub/comm) that gives me a problem
> > >because the replies that it returns are not terminated with "\r\n" but
> > >only with "\n". According to the assignment, we don't have to deal with
> > >this but I wanted to confirm with you. Btw, it's not a bug in my code, I
> > >confirmed it with another student (which still does not really prove
> > >anything, but I am pretty certain that the replies are in non-NVT ASCII
> > >format). So should we follow the assignment and ignore this case or put in
> > >a hack to handle it?
> >
> > >Thanks,
> >
> > >araik
> >
> >
> >
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lin Lu 
Newsgroups: su.class.cs244a
Subject: how to handle 
Date: Mon, 21 Jan 2002 22:30:55 -0800
Lines: 30
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2640


I have several questions:

1.What are we supposed to do with 


2.This is from course web page.  Could anyone clarify why the results
should like that?

ftpcopy ext 1 


    12345 is the port at which the local ftp server is listening 
    The permissionsDir looks like this: 

    .
    ./a_zero.ext
    ./d_normal.ext
    ./c_dir.ext
    ./e_dir.ext
    ./e_dir.ext/f_dir.ext
    ./g_filelink.ext -> ./d_normal.ext
    ./h_dirlink.ext -> ./e_dir.ext

    After the transfer the results in localdir should look like this: 

    .
    ./a_zero.ext
    ./d_normal.ext
    ./g_filelink.ext
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Jian Deng 
Newsgroups: su.class.cs244a
Subject: Link to Dir
Date: 22 Jan 2002 06:58:53 GMT
Lines: 28
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2641

I have question on the following test case
special cases: Symbolic-link-to-directory [out of 0.5]

[SUMMARY] NOT OK
[RESULT]  Symbolic link to directory in same directory did transfer (wrong!)
exit line said: ""
---
Your output to stdout was:
OK: 14 bytes copied


Is this test case testing a transfer should not
happen to a symbolic link to the same dir that has already
been transfered?

I am wondering how can we prevent that with out parsing the
actual output of the list which varies with ftp servers.. 
and PWD will not give you the actual path.

Does this test case also imply we should handle cycles and not transfer
dirs already been transfered?

Your clarification is appreciated.



 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Zero byte files
Date: Tue, 22 Jan 2002 00:07:03 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2642


I think I was confused and tried NOT TO transfer zero-byte files...
But after submitting, the test result seems to be saying that I should
transfer zero-byte files...

I think the specification didn't say anything explicitly.. (or just my
fault.. )
Could anybody confirm this?
( Should I  transfer zero-byte files or not ? )



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: Zero byte files
Date: Tue, 22 Jan 2002 01:01:19 -0800
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2643

I think we are supposed to transfer files that match on
the remote FTP server - even if they are size 0,

But, not create any 0 length files due to file-permission or
transfer problems...

BAEHOPIL wrote:
> 
> I think I was confused and tried NOT TO transfer zero-byte files...
> But after submitting, the test result seems to be saying that I should
> transfer zero-byte files...
> 
> I think the specification didn't say anything explicitly.. (or just my
> fault.. )
> Could anybody confirm this?
> ( Should I  transfer zero-byte files or not ? )

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!ambu
From: Ambika Balakrishnan 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Tue, 22 Jan 2002 02:16:02 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Alex Khomenko 
cc: Shankar Agarwal 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2644


> It seems to me that ftpcopy.purify and .purify are two different files.
Yes:

ftpcopy.purify is the purified version of your ftpcopy executable whereas
..purify is the file containing purify options such as the two "suppress
UMR" commands in the purify FAQ. This file is read by purify on startup.

Also remember to enable log files and disable windows either in your
Makefile or in your .purify so that the script can generate its own purify
output file instead of a graphical display.

--
Ambika


>
> Alex
>
> On Mon, 21 Jan 2002, Shankar Agarwal wrote:
>
> > Hi,
> > I am trying to submit hw1 and it has ftpcopy.purify in that directory but
> > it still says that no .purify is present.
> > Thanks
> > Shankar
> >
> >
>
> --
> Knowledge is a deadly friend when no one sets the rules.
> The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: About level
Date: Tue, 22 Jan 2002 02:13:33 -0800
Lines: 31
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2645


It seems that I'm having some confusion about level number..

suppose this structure..

Root - F0
        - D1 - F1
        - D2 - D3 -F2
        - D4 -  D5 - D6-F3

My understanding was that F0, D1, D2,D4 are level 0 entities,
  F2,D3,D5 are level 1, F2,D6 are level 2...

So, if ftpcopy is called with level=1,  F0, D1,F1 will be copied..
 ( assuming extension is right and there are no other files )
With level=2, F0,D1,F1,D2,D3,F2 will be copied...

But, I got the following result, which seems to be saying that I copied more
than necessary...

diff between expected tree and your tree:
19,22d18
< ./examples/extex.ch
< ./examples/wc-dos.ch
< ./examples/wmer-os2.ch
< ./examples/wmerg-pc.ch

Am I having some misunderstanding here?



.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: ftpcopy Z 2 204.123.2.2 localdir /pub/comm ??
Date: 22 Jan 2002 15:31:32 GMT
Organization: Stanford University, CA 94305, USA
Lines: 7
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2646

I am reading the replys from the control socket before reading the data
and I am passing the test scripts. Have I just gotten lucky?
Also can I take it from your response we should be reading the entire LIST
reply before parsing it? 

-mark

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Tue, 22 Jan 2002 07:53:21 -0800
Lines: 47
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: elaine14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Ambika Balakrishnan 
cc: Alex Khomenko 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2647

Hi,
Is there any particular name that i should  give to this file and how do i
tell the compiler to pick up this file. I mean do i have to set some
environment variable where i can tell the name of the file containing the
supress statements.
Thanks
Regards
Shankar
On Tue, 22 Jan 2002, Ambika Balakrishnan wrote:

>
> > It seems to me that ftpcopy.purify and .purify are two different files.
> Yes:
>
> ftpcopy.purify is the purified version of your ftpcopy executable whereas
> .purify is the file containing purify options such as the two "suppress
> UMR" commands in the purify FAQ. This file is read by purify on startup.
>
> Also remember to enable log files and disable windows either in your
> Makefile or in your .purify so that the script can generate its own purify
> output file instead of a graphical display.
>
> --
> Ambika
>
>
> >
> > Alex
> >
> > On Mon, 21 Jan 2002, Shankar Agarwal wrote:
> >
> > > Hi,
> > > I am trying to submit hw1 and it has ftpcopy.purify in that directory but
> > > it still says that no .purify is present.
> > > Thanks
> > > Shankar
> > >
> > >
> >
> > --
> > Knowledge is a deadly friend when no one sets the rules.
> > The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: non-NVT ASCII replies (fwd)
Date: Tue, 22 Jan 2002 09:06:58 -0800
Lines: 42
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2648


I found that the response to LIST command is different for TYPE I and TYPE
A..
( LIST response is not automatically set to NVT-ASCII mode )
When TYPE A, server returns NVT-ASCII, but when TYPE=I, it does not..
So, if we're changing between TYPE A and TYPE I for RETR, there may be some
inconsistency...
I think the inconsistency may be because of this, but I'm not sure...

-------------------------------------------
"Timothy Tay Chao"  wrote in message

> I've run into the same thing, but i'm apt to blame myself. Using ftpd,
> I've found that I am sometimes getting list responses off the data
> connection that end in "\r\n" and that sometimes end in "\n". This
> inconsistency makes me think there's a bug in my code. Suggestions on
> this would be quite helpful.
>
> Araik Grigoryan wrote:
> >
> > On the data connection or the cotrol connection or both? The control
> > connection seems to be fine, it's the data connection that returns
non-NVT
> > ASCII replies. Matt, I think you mentioned that you are using your own
> > ftpcopy program. Are you sure you are not specifically programming
around
> > this problem?
> >
> > On 21 Jan 2002, Matthew Jonathan Holliman wrote:
> >
> > >
> > > This server is correctly giving me NVT ASCII when I try.
> > >
> > > Matthew
> > >

> > > >araik
> > >
> > >
> > >


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic3.Stanford.EDU!anhoe
From: An-hoe Shih 
Newsgroups: su.class.cs244a
Subject: argument of PORT command
Date: Tue, 22 Jan 2002 09:23:01 -0800
Lines: 27
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2649

Hi all,

   For the PORT command, how do we specify the argument? For example, if
the local host address is 171.64.15.36 and the port to use is 0xD84F, do
we just convert each byte to a character and place a ',' in the middle?

   I tried to do this:
171 = '\253'
64 = '@'
15 = '\017'
36 = '$'
0xDB = \330
0x4F = '0'

So I send to server the following command line:
PORT 

However, the server kept givinig me the error: 501 Syntax error in
parameters.

Can anybody tell me what did I do wrong?


Thanks,

An-Hoe

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: Re: argument of PORT command
Date: Tue, 22 Jan 2002 10:06:12 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: An-hoe Shih 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2650


You need to print each byte as a decimal integer, not character.

Alex

On Tue, 22 Jan 2002, An-hoe Shih wrote:

> Hi all,
>
>    For the PORT command, how do we specify the argument? For example, if
> the local host address is 171.64.15.36 and the port to use is 0xD84F, do
> we just convert each byte to a character and place a ',' in the middle?
>
>    I tried to do this:
> 171 = '\253'
> 64 = '@'
> 15 = '\017'
> 36 = '$'
> 0xDB = \330
> 0x4F = '0'
>
> So I send to server the following command line:
> PORT 
>
> However, the server kept givinig me the error: 501 Syntax error in
> parameters.
>
> Can anybody tell me what did I do wrong?
>
>
> Thanks,
>
> An-Hoe
>
>

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: LIST response in IMAGE type?
Date: Tue, 22 Jan 2002 10:26:58 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2651

Very simple; and this might be Araik's problem as well:

make sure you're using 'TYPE A' when getting directory listings -- as in 
ASCII transfer, not IMAGE.

-Sandeep

BAEHOPIL wrote:

>  When I set the TYPE to I and sent the LIST command,
>  the response doesn't seem to be in NVT-ASCII format.
>  Although I may experiment with it and make things work,
>  I'm afraid LIST response in IMAGE type may have no standard.
> 
>  But, I think switching between NVT-ASCII for LIST and IMAGE for RETR is too
> cumbersome...
>  I wish that I can use TYPE I exclusively...
> 
>  Could anybody help?
> 
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Tue, 22 Jan 2002 10:35:06 -0800
Lines: 79
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2652

Ok, let's try this one more time:

1. cd to your project directory
2. cat > .purify
3. Type the following two lines (don't wait for a prompt or anything; 
just type)
suppress UMR ...;res_search
suppress UMR ...;nss_search

4. Hit Ctrl-d to close the file.

You now have a valid .purify file.  It won't show up with 'ls' because 
files that begin with "." are hidden.  To list hidden files do an 'ls -a'.

Good luck.

-Sandeep

Shankar Agarwal wrote:

> Hi,
> Is there any particular name that i should  give to this file and how do i
> tell the compiler to pick up this file. I mean do i have to set some
> environment variable where i can tell the name of the file containing the
> supress statements.
> Thanks
> Regards
> Shankar
> On Tue, 22 Jan 2002, Ambika Balakrishnan wrote:
> 
> 
>>>It seems to me that ftpcopy.purify and .purify are two different files.
>>>
>>Yes:
>>
>>ftpcopy.purify is the purified version of your ftpcopy executable whereas
>>.purify is the file containing purify options such as the two "suppress
>>UMR" commands in the purify FAQ. This file is read by purify on startup.
>>
>>Also remember to enable log files and disable windows either in your
>>Makefile or in your .purify so that the script can generate its own purify
>>output file instead of a graphical display.
>>
>>--
>>Ambika
>>
>>
>>
>>>Alex
>>>
>>>On Mon, 21 Jan 2002, Shankar Agarwal wrote:
>>>
>>>
>>>>Hi,
>>>>I am trying to submit hw1 and it has ftpcopy.purify in that directory but
>>>>it still says that no .purify is present.
>>>>Thanks
>>>>Shankar
>>>>
>>>>
>>>>
>>>--
>>>Knowledge is a deadly friend when no one sets the rules.
>>>The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
>>>
>>>
>>>
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: ftpcopy Z 2 204.123.2.2 localdir /pub/comm ??
Date: Tue, 22 Jan 2002 10:44:52 -0800
Lines: 38
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2653

In my opinion you did get lucky...well, it's more than an opinion; some 
of my friends ran into this very blocking problem.

Here's my reasoning:  you tell the ftp server that you want a directory 
listing; it tells you, 'Sure, I'll get started on that."  But if you 
don't read from the data socket, and the directory listing is 
sufficiently large, he won't be able to write the whole listing to his 
end of the socket (because some buffer has filled up).  So he won't send 
the 226.  However, if you read from the data connection (i.e. read ALL 
the data), then the server will finish writing to its end of the socket 
and issue the 226.

It wouldn't hurt to parse the directory data a line at a time; but you 
can't act on it (i.e. download a file, change to a subdirectory) until 
you've finished reading all the data from the current listing.

-Sandeep

Mark Joseph Dolan wrote:

> I am reading the replys from the control socket before reading the data
> and I am passing the test scripts. Have I just gotten lucky?
> Also can I take it from your response we should be reading the entire LIST
> reply before parsing it? 
> 
> -mark
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine42.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Got the following error message.
Date: Tue, 22 Jan 2002 10:45:52 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine42.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2654

Hi,
I got the follwoing error message when submitting the hw1. Does this mean
that submisssion was successful or i need to submit again.
Submission succeeded.
Error: log: open: Permission denied

A problem statement has been e-mailed to the system
administrator 

Thanks
Regards
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine42.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Tue, 22 Jan 2002 10:47:56 -0800
Lines: 90
Distribution: su
Message-ID: 
References: 
 
 
 
 
NNTP-Posting-Host: elaine42.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Sandeep Tamhankar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2655


Hi Sandeep,
Thanks a lot. I was misintrepreting .purify as some file with .purify in
the end. Thats why all this confusion. Anyways finally i was able to
submit the code.
Regards
Shankar
On Tue, 22 Jan 2002, Sandeep Tamhankar wrote:

> Ok, let's try this one more time:
>
> 1. cd to your project directory
> 2. cat > .purify
> 3. Type the following two lines (don't wait for a prompt or anything;
> just type)
> suppress UMR ...;res_search
> suppress UMR ...;nss_search
>
> 4. Hit Ctrl-d to close the file.
>
> You now have a valid .purify file.  It won't show up with 'ls' because
> files that begin with "." are hidden.  To list hidden files do an 'ls -a'.
>
> Good luck.
>
> -Sandeep
>
> Shankar Agarwal wrote:
>
> > Hi,
> > Is there any particular name that i should  give to this file and how do i
> > tell the compiler to pick up this file. I mean do i have to set some
> > environment variable where i can tell the name of the file containing the
> > supress statements.
> > Thanks
> > Regards
> > Shankar
> > On Tue, 22 Jan 2002, Ambika Balakrishnan wrote:
> >
> >
> >>>It seems to me that ftpcopy.purify and .purify are two different files.
> >>>
> >>Yes:
> >>
> >>ftpcopy.purify is the purified version of your ftpcopy executable whereas
> >>.purify is the file containing purify options such as the two "suppress
> >>UMR" commands in the purify FAQ. This file is read by purify on startup.
> >>
> >>Also remember to enable log files and disable windows either in your
> >>Makefile or in your .purify so that the script can generate its own purify
> >>output file instead of a graphical display.
> >>
> >>--
> >>Ambika
> >>
> >>
> >>
> >>>Alex
> >>>
> >>>On Mon, 21 Jan 2002, Shankar Agarwal wrote:
> >>>
> >>>
> >>>>Hi,
> >>>>I am trying to submit hw1 and it has ftpcopy.purify in that directory but
> >>>>it still says that no .purify is present.
> >>>>Thanks
> >>>>Shankar
> >>>>
> >>>>
> >>>>
> >>>--
> >>>Knowledge is a deadly friend when no one sets the rules.
> >>>The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
> >>>
> >>>
> >>>
> >>
> >
>
>
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine25.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: non-NVT ASCII replies (fwd)
Date: 22 Jan 2002 18:59:43 GMT
Lines: 18
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine25.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2656

Araik Grigoryan  writes:


>On the data connection or the cotrol connection or both? The control
>connection seems to be fine, it's the data connection that returns non-NVT
>ASCII replies. Matt, I think you mentioned that you are using your own
>ftpcopy program. Are you sure you are not specifically programming around
>this problem?

The server sends NVT ASCII over both data and control connections; I wasn't
doing anything special to handle this particular host.

I could reproduce the behaviour you mentioned if I set the transfer mode
to binary before doing LIST.  As the RFC states, the transfer mode must
be set to ASCII before you issue LIST--so be sure to check that this is the
case.

Matthew
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: Got the following error message.
Date: Tue, 22 Jan 2002 11:14:07 -0800
Lines: 42
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2657

(This was taken from an earlier posting)

Hi,

If you saw an error like this, the error is minor and it did not prevent
you from submitting your code, in other words, we received your files
correctly. This problem has been solved, so you will not see it again,
if you submit now.

Pablo

Russell Greene wrote:

> Has anyone received the error when running the submit script?
>
> Submission succeeded
> Error: log: open: Permission denied
>
>
> --Russ
>
>
>


"Shankar Agarwal"  wrote in message

> Hi,
> I got the follwoing error message when submitting the hw1. Does this mean
> that submisssion was successful or i need to submit again.
> Submission succeeded.
> Error: log: open: Permission denied
>
> A problem statement has been e-mailed to the system
> administrator 
>
> Thanks
> Regards
> Shankar
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Tue, 22 Jan 2002 11:39:47 -0800
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: madrid.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Windows; U; Windows NT 5.1; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2658

As mentioned in the assignment handout, you have to have two 
purify-related files, .purify and a purify.output, in your homework 
directory before you can submit your code.

Pablo


Shankar Agarwal wrote:

> Hi,
> I am trying to submit hw1 and it has ftpcopy.purify in that directory but
> it still says that no .purify is present.
> Thanks
> Shankar
> 
> 


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: Regarding Post of 1/16/02, 8:56: Script File Sizes
Date: 22 Jan 2002 20:01:20 GMT
Lines: 43
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth10.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2659

I'm having similar problems. Tests D and F read:

exit line: ""

but my program always prints something out, either an "OK: ..." or "ERROR:
...." I've manually tried running my program one of the tested sites
(ftp.cs.stanford.edu) and everything works as expected, so I don't know
what to do.

I'm also having problems with tests I and L. I think the FTP server on
epic is messed up, because I get:

Your output to stdout was:
TIMEOUT! killed program: /afs/ir/class/cs244a/submissions/restricted/hw1/hondroul/oceans//compile.sun/ftpcopy ext 1  /afs/ir/class/cs244a/submissions/restricted/hw1/hondroul/oceans//compile.sun/download/perms .   21715


Anyone know what's going on?

- Derrick



Steven Siuhong Ngai  wrote:
: Some random questions:

: -Have any of you gotten the gotten the same bad results on that part of the
: test that Bret mentioned? I've manually visited the sites where my tests
: fail, and I can't figure out what's wrong. Any helpful stories of bugs
: fixed?

: -Did either of you perhaps have any problems with the exit line reports? My
: test results read "exit line: ' ' " where I think they ought to have read
: "OK: etc." Should they read "OK..."? Are points being deducted on that basis
: (this is a question probably for a TA)?

: -When the test lists diffs, do >'s mean that your tree had certain files and
: that the correct one didn't (ie you shouldn't have had them)?

: Thanks.
: Steve



.

Path: shelby.stanford.edu!nntp.stanford.edu!myth8.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Got the following error message.
Date: Tue, 22 Jan 2002 12:20:56 -0800
Lines: 64
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2660


Hi,
I jsut submitted and ran into this error message.
Hope i am ok and does not need to resubmit. Just
wanted to convey because earlier mail said this
problem was fixed ?

Thanks, Arvind

-------------------------
Submission succeeded.
Error: log: open: Permission denied

A problem statement has been e-mailed to the system
administrator 



On Tue, 22 Jan 2002, Russell Greene wrote:

> (This was taken from an earlier posting)
>
> Hi,
>
> If you saw an error like this, the error is minor and it did not prevent
> you from submitting your code, in other words, we received your files
> correctly. This problem has been solved, so you will not see it again,
> if you submit now.
>
> Pablo
>
> Russell Greene wrote:
>
> > Has anyone received the error when running the submit script?
> >
> > Submission succeeded
> > Error: log: open: Permission denied
> >
> >
> > --Russ
> >
> >
> >
>
>
> "Shankar Agarwal"  wrote in message
> 
> > Hi,
> > I got the follwoing error message when submitting the hw1. Does this mean
> > that submisssion was successful or i need to submit again.
> > Submission succeeded.
> > Error: log: open: Permission denied
> >
> > A problem statement has been e-mailed to the system
> > administrator 
> >
> > Thanks
> > Regards
> > Shankar
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Debashis Sahoo" 
Newsgroups: su.class.cs244a
Subject: User Interface violation ...
Date: Tue, 22 Jan 2002 12:42:36 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: dsahoo.stanford.edu
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2661

Hi,
    I received following error.
What should be the output? Any ideas.....?

    * Test I: User interface specification:  violates output format [out of
0.5]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy gz 2 epic11.stanford.edu localdir
exit line said: "ERROR : You can't FTP into this machine. FTP into
transfer.stanford.edu."
---
Program output:
ERROR : You can't FTP into this machine. FTP into transfer.stanford.edu.

---

Thanks,
Sahoo


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Joong Ho Won" 
Newsgroups: su.class.cs244a
Subject: proper RETR and LIST behavior
Date: Tue, 22 Jan 2002 13:09:31 -0800
Lines: 25
Distribution: su
Message-ID: 
NNTP-Posting-Host: cecily.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2919.6700
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700
Xref: nntp.stanford.edu su.class.cs244a:2662

Hello,

I wonder if RETR command can take an absoulte/relative path to a file as an
argument. i.e.,

RETR /a/path/to/a/file

rather than having just a filename, i.e.,

RETR file

RFC doesn't seem to specify this case.

Some ftp(ex. ftp.stanford.edu ) seems not to allow the former case, but
others(ex. transfer.stanford.edu ) seems to.

I'd like to make it clear for LIST for the same case as well, though it
seems to be allowed seeing RFC.

Thanks.

Joong Ho Won



.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Test Result
Date: Tue, 22 Jan 2002 13:14:24 -0800
Lines: 48
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Moh'd Saleem Saleem Alrawashdeh 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2663

It is just testing symbolic links to files. Please see the earlier posted
responses to how to handle symbolic links. Best, Xin


On Mon, 21 Jan 2002, Moh'd Saleem Saleem Alrawashdeh wrote:


> Hi,
> 
> I am posting this again since I didn't get a reply for my first post. Can
> anyone please tell me what this test means?
> 
> ===============
> 
>     * Test L.5: Special cases: Symbolic-link-to-file [out of 0.5]
> 
> [SUMMARY] NOT OK
> [RESULT]  Symbolic link to file in same directory did not transfer
> (wrong!)
> exit line said: ""
> ---
> Your output to stdout was:
> 
> OK: 14 bytes copied
> 
> ---
> 
> 
> ===============
> 
> I appreciate any clarification of what special case that this test is
> testing?
> 
> Thanks,
> 
> Moh'd
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: how to handle 
Date: Tue, 22 Jan 2002 13:35:18 -0800
Lines: 47
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Lin Lu 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2664

Hi,

> 1.What are we supposed to do with 

As the handout says, if the name is of the form 
should be used as the port number of the ftp server; otherwise, the
default ftp service port should be used.

> 2.This is from course web page.  Could anyone clarify why the results
> should like that?
> 
> ftpcopy ext 1 
> 
> 
>     12345 is the port at which the local ftp server is listening 
>     The permissionsDir looks like this: 
> 
>     .
>     ./a_zero.ext
>     ./d_normal.ext
>     ./c_dir.ext
>     ./e_dir.ext
>     ./e_dir.ext/f_dir.ext
>     ./g_filelink.ext -> ./d_normal.ext
>     ./h_dirlink.ext -> ./e_dir.ext
> 
>     After the transfer the results in localdir should look like this: 
> 
>     .
>     ./a_zero.ext
>     ./d_normal.ext
>     ./g_filelink.ext

Could you please specify where you don't understand?

Best,

Xin

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Zero byte files
Date: Tue, 22 Jan 2002 14:28:05 -0800
Lines: 19
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2665

Exactly. You are supposed to handle zero bype files in this way. Xin


On Tue, 22 Jan 2002, Luis Robles wrote:

> I think we are supposed to transfer files that match on
> the remote FTP server - even if they are size 0,
> 
> But, not create any 0 length files due to file-permission or
> transfer problems...

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: About level
Date: Tue, 22 Jan 2002 14:30:08 -0800
Lines: 46
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: BAEHOPIL 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2666

Please refer to the examples of levels in review session 1. Xin


On Tue, 22 Jan 2002, BAEHOPIL wrote:

> 
> It seems that I'm having some confusion about level number..
> 
> suppose this structure..
> 
> Root - F0
>         - D1 - F1
>         - D2 - D3 -F2
>         - D4 -  D5 - D6-F3
> 
> My understanding was that F0, D1, D2,D4 are level 0 entities,
>   F2,D3,D5 are level 1, F2,D6 are level 2...
> 
> So, if ftpcopy is called with level=1,  F0, D1,F1 will be copied..
>  ( assuming extension is right and there are no other files )
> With level=2, F0,D1,F1,D2,D3,F2 will be copied...
> 
> But, I got the following result, which seems to be saying that I copied more
> than necessary...
> 
> diff between expected tree and your tree:
> 19,22d18
> < ./examples/extex.ch
> < ./examples/wc-dos.ch
> < ./examples/wmer-os2.ch
> < ./examples/wmerg-pc.ch
> 
> Am I having some misunderstanding here?
> 
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: ftpcopy.purify reports FIU
Date: Tue, 22 Jan 2002 14:31:24 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2667

When I run ftpcopy.purify I get a FIU due to gethostbyname

FIU: file descriptor 3: "/etc/.name_service_door", O_RDONLY
File info: ?r--r--r--  1 root root ...
This file descriptor was allocated from:
    __open64      [libc.so.1]
    _open64       [libc.so.1]
    _nsc_trydoorcall [libc.so.1]
    ...
    ...
    gethostbyname_r [gethostbyname_r.c]

What should I do?

I have one more question:
Do we have to issue a quit on the control connection before closing or
it's OK without that.

Thanks,
Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit test
Date: Tue, 22 Jan 2002 15:10:32 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2668

Please do a "touch .purify"
before submitting.

Regards,
Antonios.

Jinhui Pan wrote:

> I still can not submit test. This is the error message:
>
> ************************************************************************
> SubmitBase: /afs/ir/class/cs244a/submissions
> Debug Level = 0
> This is your submission 1 (out of 5) for today (1-20-2002)
> could not find any file matching .purify in
> /afs/ir.stanford.edu/users/j/i/jinhui/cs244a/hw1!!
> ======================================================================
> Your code could not be sumitted because of some error.
> If you think there is a mistake, please send an email to your TA
> (holliman;  with the relevant information.
> ======================================================================
>
> Died at /afs/ir/class/cs244a/bin/submit_test.pl line 195.
> ***********************************************************************
>
> But I do have ftpcopy.purify in my folder. What's wrong with it?
>
> Thanks a lot!
> -jinhui

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: file separator
Date: Tue, 22 Jan 2002 15:34:46 -0800
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2669

Araik,

I'm not aware of a preprocessor definition for the file name separators.

http://www.scs.carleton.ca/~courses/101/Notes2001/pathnames.html
Here's a link that names a couple of differences between
Win and UNIX filenames, in case you're interested.

Regards,
Antonios.

Araik Grigoryan wrote:

> Hi,
>
> Does anyone know if there is some global string defined in a C library
> that contains a character that separates file names, e.g. "/" or "\"?
> Thanks.
>
> araik

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Test Functionality
Date: Tue, 22 Jan 2002 15:46:53 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2670


Hi Moh'd,

Your output at stderr should follow exactly the format described
in the assignment, both in the successful and the unsuccessful cases (OK or
ERR
cases).

Regards,
Antonios.


Moh'd Saleem Saleem Alrawashdeh wrote:

> Hi all,
>
> Anyone can tell me what is the required functionality of this test:
>
>  * Test I: User interface specification:  violates output format [out of
> 0.5]
>
> Thanks,
>
> Moh'd

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: file extensions.
Date: Tue, 22 Jan 2002 15:52:58 -0800
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2671

if you specify as your extention "gz" then that includes
the xxx.tar.gz since it is a zipped file!

Regards,
Antonios.

Hasan Taufiq Imam wrote:

> xxx.tar.gz an acceptable file if user specification is just
> .gz. Need clarification.
> Thanks,
> Hasan

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: broken pipe
Date: Tue, 22 Jan 2002 16:00:18 -0800
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2672


Debashis,

even though the site responds in a non-standard way,
your code should detect that the connection has already
been closed and not reach the point where you try to
write to a closed-connection socket (when you do that
you receive a SIGPIPE signal). It is up to you to decide
a) whether you want to detect such non-ftp servers by
counting on the SIGPIPE signal being sent and handling it
or b) by reading what the server responded and figuring out
it's not a correct ftpserver (by the standard).
The error message you receive in the report
didn't handle that and didn't manage to print the appropriate ERR code.

Regards,
Antonios.

Debashis Sahoo wrote:

> I am receiving following error from the automatic test script.
> I am getting broken pipe while calling write to socket in this case.
> Do we have to catch SIGPIPE signal in this case?
> How do we handle this case ...any ideas?
>
>     * Test I: User interface specification:  violates output format [out of
> 0.5]
>
> [SUMMARY] NOT OK
> [RESULT]
> test used: ftpcopy gz 2 epic11.stanford.edu localdir
> exit line said: ""
>
> Thanks,
> Sahoo

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Default remoteBaseDir ??
Date: Tue, 22 Jan 2002 16:03:07 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2673

Gaurav,
you can try sending a "PWD" command to receive the current
working directory the moment you first connect to the server.
for more info on PWD please read the rfc.

Regards,
Antonios.

Gaurav Garg wrote:

> This one for the TA's:
>
>    Can we assume the default remote server's base directory specified be
> always '/' or can default directory be any other directory in general?
>    The ftp servers mentioned in the problem statement all default to '/',
> so is this a rule or mere coincidence??
>
> -gaurav
>
> *************************************************************************
> Gaurav Garg                          * Contact Info: 20A Comstock Circle
> 1st Yr, Graduate Student             *       Escondido Village, Stanford
> Department of Electrical Engineering *       CA-94305
> Stanford University                  * Ph:650-498-1208
> *************************************************************************
>
> -Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Debashis Sahoo" 
Newsgroups: su.class.cs244a
Subject: Re: broken pipe
Date: Tue, 22 Jan 2002 16:26:30 -0800
Lines: 61
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: dsahoo.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2674

When I printed following error message it says NOT OK.
I received following error message from the automatic test script.
Should we print any error message in this case?

   * Test I: User interface specification:  violates output format [out of
0.5]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy gz 2 epic11.stanford.edu localdir
exit line said: "ERROR : You can't FTP into this machine. FTP into
transfer.stanford.edu."
---
Program output:
ERROR : You can't FTP into this machine. FTP into transfer.stanford.edu.

---

Thanks,
Debashis

"SUL\AIR Laneroom Cluster"  wrote in message

>
> Debashis,
>
> even though the site responds in a non-standard way,
> your code should detect that the connection has already
> been closed and not reach the point where you try to
> write to a closed-connection socket (when you do that
> you receive a SIGPIPE signal). It is up to you to decide
> a) whether you want to detect such non-ftp servers by
> counting on the SIGPIPE signal being sent and handling it
> or b) by reading what the server responded and figuring out
> it's not a correct ftpserver (by the standard).
> The error message you receive in the report
> didn't handle that and didn't manage to print the appropriate ERR code.
>
> Regards,
> Antonios.
>
> Debashis Sahoo wrote:
>
> > I am receiving following error from the automatic test script.
> > I am getting broken pipe while calling write to socket in this case.
> > Do we have to catch SIGPIPE signal in this case?
> > How do we handle this case ...any ideas?
> >
> >     * Test I: User interface specification:  violates output format [out
of
> > 0.5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]
> > test used: ftpcopy gz 2 epic11.stanford.edu localdir
> > exit line said: ""
> >
> > Thanks,
> > Sahoo
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: IMPORTANT CLARIFICATION: failure on ftp.slac.stanford.edu tests
Date: 23 Jan 2002 00:28:46 GMT
Lines: 27
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2675

Hello all,

A lot of people have been sending in emails to the staff stating that the
submit test script is reporting failures on various ftp.slac.stanford.edu,
such as TIMEOUT or exit line incorrect, or tree incorrect, etc.

Note that the script determines these errors by examining the output of a
single program run. In particular, if this test does not terminate in
TIMEOUT time, it is highly likely that all the remaining tests will fail.  
Since this is a relatively deep recursion, the time limit is imposed to
penalize programs with poor performance. Alternatively, programs that hang
or recurse will force the program to be abnormally terminated as well.

If you see a TIMEOUT error for a given test, please look into what could 
be causing this. Two things to consider:

a) Do you read data from the socket when there is no data available (i.e., 
expecting a reply when there should be none)? This will hang your process

b) During directory recursion, do you ignore the "." and ".." entries? 
Some, not all, ftp servers return these as directory names as part of the 
LIST response. Among these is ftp.slac.stanford.edu.

Good luck,

Arun

.

Path: shelby.stanford.edu!nntp.stanford.edu!spevack.Stanford.EDU!spevack
From: Max Spevack 
Newsgroups: su.class.cs244a
Subject: Not getting response from QUIT
Date: Tue, 22 Jan 2002 16:29:24 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: spevack.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2676

Hi,

This may seem silly, but for some reason I can't get my program to print
out the response from the server after I issue the QUIT command.  Other
server responses are printed out fine, but QUIT never seems to come
through.

Any ideas?

Thanks,
Max


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links and File Extensions
Date: Tue, 22 Jan 2002 16:33:57 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2677

Pete,
a symbolic link to a file of interest (because of the file's extention) should
always be followed and treated as a local file, i.e. should be copied. The idea
of the ftpclient
is to make local copies of particular files of interest. Symbolic links aren't
particularly interesting files since copying them locally may create dangling
references.

So, the answer is yes to the first, no to the second!

Regards,
Antonios.


Pete Belknap wrote:

> What should we do when the symbolic link has a diff extension than the file
> it points to?
>
> So, for example, if we want to download .txt files and we have the link
>
> link.lnk  ->  readme.txt
>
> Should we download it?  Of course, the opposite question is, if we are
> interested in files of extension .lnk, should we download the link?  It
> seems to me that the answer to my first question should be no and the second
> should be yes...
>
> thanks,
> Pete

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: broken pipe
Date: Tue, 22 Jan 2002 16:37:25 -0800
Lines: 68
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2678

Please try to print "ERROR: " before your message. Please notice the exact
format required from the assignment.

Regards,
Antonios.

Debashis Sahoo wrote:

> When I printed following error message it says NOT OK.
> I received following error message from the automatic test script.
> Should we print any error message in this case?
>
>    * Test I: User interface specification:  violates output format [out of
> 0.5]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy gz 2 epic11.stanford.edu localdir
> exit line said: "ERROR : You can't FTP into this machine. FTP into
> transfer.stanford.edu."
> ---
> Program output:
> ERROR : You can't FTP into this machine. FTP into transfer.stanford.edu.
>
> ---
>
> Thanks,
> Debashis
>
> "SUL\AIR Laneroom Cluster"  wrote in message
> 
> >
> > Debashis,
> >
> > even though the site responds in a non-standard way,
> > your code should detect that the connection has already
> > been closed and not reach the point where you try to
> > write to a closed-connection socket (when you do that
> > you receive a SIGPIPE signal). It is up to you to decide
> > a) whether you want to detect such non-ftp servers by
> > counting on the SIGPIPE signal being sent and handling it
> > or b) by reading what the server responded and figuring out
> > it's not a correct ftpserver (by the standard).
> > The error message you receive in the report
> > didn't handle that and didn't manage to print the appropriate ERR code.
> >
> > Regards,
> > Antonios.
> >
> > Debashis Sahoo wrote:
> >
> > > I am receiving following error from the automatic test script.
> > > I am getting broken pipe while calling write to socket in this case.
> > > Do we have to catch SIGPIPE signal in this case?
> > > How do we handle this case ...any ideas?
> > >
> > >     * Test I: User interface specification:  violates output format [out
> of
> > > 0.5]
> > >
> > > [SUMMARY] NOT OK
> > > [RESULT]
> > > test used: ftpcopy gz 2 epic11.stanford.edu localdir
> > > exit line said: ""
> > >
> > > Thanks,
> > > Sahoo
> >

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Other files in directory.
Date: Tue, 22 Jan 2002 17:11:41 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2679


you should copy all permissible files in a remote directory,
OVERWRITING any old ones in the local directory. Your
code should make sure no extra files exist in the new, local tree you're
trying to create.


Regards,
Antonios.


Hasan Taufiq Imam wrote:

> Please, clarify.
> In case of non-empty local directory do we delete files or directories
> as required or do we delete everything.
> In other words, can the local directory contain any files or directories other
> than the tree copied from the remote server.
>
> I still don't know if the user specifies extension gz, is tar.gz a valid
> extension or not.
> Thanks.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Sundar Maddali 
Newsgroups: su.class.cs244a
Subject: CheckSum Errors
Date: Tue, 22 Jan 2002 17:13:30 -0800
Lines: 34
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic0.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2680

Hi,

   when I run submit_test against my code i get checksum errors and all
files not transferred errors,( when the program is run against
204.123.2.2 server especially) . Also later in the report it says that
this process got killed due to TIMEOUT. 
could it be because of the program being killed, i am getting the
checksum errors.
 
I wanted to know what is the TIMEOUT value that the testing program is
configured to, so that i can try with that settings locally. 

Thanks
~
Lakshman Maddali
-----------------------------------------------------------------------------
 Test G: Transfers all of the matching files, but bad file checksum [out
of 1]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: ""
---
diff between expected tree and your tree:
4c4,8
< ./kermit/c/mac.tar.Z 3663175723 5840
---
> ./kermit/c/mac.tar.Z 2459747744 496551
> ./kermit/c/os2.tar.Z 2771080586 163808
> ./kermit/c/os9.tar.Z 370278796 49438
> ./kermit/c/unix.tar.Z 308104872 349221
> ./kermit/c/vms.tar.Z 53741681 64755

-----------------------------------------------------------------------------
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: error cleanup
Date: Tue, 22 Jan 2002 17:13:49 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2681

Yes, your code should handle exceptional cases as normal.
ie. deallocate memory and close sockets when done w/ them, etc.
Of course, please don't forget the required output message!
Regards,
Antonios.

Jason Ahmad wrote:

> What action, exactly, are we to take on general error cases (can't
> allocate socket, can't make connection etc)? It is clear we are to write
> a brief explaination to stderr and exit, but are we to free malloced
> memory and make sure all sockets are closed and whatnot?
>
> Thanks,
> Jason

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: location of testing dir
Date: Tue, 22 Jan 2002 17:17:14 -0800
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2682

Hello Ambika,
those directories are your responsibility to find and use.
You can create them from scratch or use preexisting ones.
Please be careful not to overwrite any useful data with your
ftp client, as your code is required to overwrite existing files!

Regards,
Antonios.

Ambika Balakrishnan wrote:

> Hello,
>
> In the handout "Testing homework 1" there are 6 examples and I dont know
> how to run examples 1 and 2, which refer to a permissionsDir and an
> exampleDir. Where are these directories? Am I supposed to create them? Am
> I supposed to run the supplied ftpd for these examples?
>
> Thanks,
> Ambika

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Link to Dir
Date: Tue, 22 Jan 2002 17:27:14 -0800
Lines: 39
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2683


Hi Jian,

Yes, your code should be written in such a way that no
file is downloaded more than once. Cases with symlinks
to the current directory or to parent directories that you've
already processed are troublesome cases that you need
to take care of, in order to avoid downloading twice a file.

Regards,
Antonios.

Jian Deng wrote:

> I have question on the following test case
> special cases: Symbolic-link-to-directory [out of 0.5]
>
> [SUMMARY] NOT OK
> [RESULT]  Symbolic link to directory in same directory did transfer (wrong!)
> exit line said: ""
> ---
> Your output to stdout was:
> OK: 14 bytes copied
>
> Is this test case testing a transfer should not
> happen to a symbolic link to the same dir that has already
> been transfered?
>
> I am wondering how can we prevent that with out parsing the
> actual output of the list which varies with ftp servers..
> and PWD will not give you the actual path.
>
> Does this test case also imply we should handle cycles and not transfer
> dirs already been transfered?
>
> Your clarification is appreciated.
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: User Interface violation ...
Date: Tue, 22 Jan 2002 17:29:43 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2684

Please check your posting from yesterday, 'cause
I've already suggested what could be wrong.
Regards,
Antonios.

Debashis Sahoo wrote:

> Hi,
>     I received following error.
> What should be the output? Any ideas.....?
>
>     * Test I: User interface specification:  violates output format [out of
> 0.5]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy gz 2 epic11.stanford.edu localdir
> exit line said: "ERROR : You can't FTP into this machine. FTP into
> transfer.stanford.edu."
> ---
> Program output:
> ERROR : You can't FTP into this machine. FTP into transfer.stanford.edu.
>
> ---
>
> Thanks,
> Sahoo

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: proper RETR and LIST behavior
Date: Tue, 22 Jan 2002 17:33:12 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2685


Joong,
here's the part of the RFC that explains RETR:

       RETRIEVE (RETR)

            This command causes the server-DTP to transfer a copy of the
            file, specified in the pathname, to the server- or user-DTP
            at the other end of the data connection.  The status and
            contents of the file at the server site shall be unaffected.


Please keep in mind that a pathname and a filename are indeed different!
Regards,
Antonios.

Joong Ho Won wrote:

> Hello,
>
> I wonder if RETR command can take an absoulte/relative path to a file as an
> argument. i.e.,
>
> RETR /a/path/to/a/file
>
> rather than having just a filename, i.e.,
>
> RETR file
>
> RFC doesn't seem to specify this case.
>
> Some ftp(ex. ftp.stanford.edu ) seems not to allow the former case, but
> others(ex. transfer.stanford.edu ) seems to.
>
> I'd like to make it clear for LIST for the same case as well, though it
> seems to be allowed seeing RFC.
>
> Thanks.
>
> Joong Ho Won

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: ftpcopy.purify reports FIU
Date: Tue, 22 Jan 2002 17:40:41 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2686

Dhawal,

out of courtesy to the server, please do use QUIT as it is mentioned
in the RFC :-)

The FileInUse problem sounds like nothing to do with your code.
It could be a transient name-server error, but I'm not sure.
Please check the args to gethostbyname to make sure.

Regards,
Antonios.

Dhawal Kumar wrote:

> When I run ftpcopy.purify I get a FIU due to gethostbyname
>
> FIU: file descriptor 3: "/etc/.name_service_door", O_RDONLY
> File info: ?r--r--r--  1 root root ...
> This file descriptor was allocated from:
>     __open64      [libc.so.1]
>     _open64       [libc.so.1]
>     _nsc_trydoorcall [libc.so.1]
>     ...
>     ...
>     gethostbyname_r [gethostbyname_r.c]
>
> What should I do?
>
> I have one more question:
> Do we have to issue a quit on the control connection before closing or
> it's OK without that.
>
> Thanks,
> Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Not getting response from QUIT
Date: Tue, 22 Jan 2002 17:43:01 -0800
Lines: 38
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2687


Hi Max,

here's what the RFC says about QUIT:

         LOGOUT (QUIT)

            This command terminates a USER and if file transfer is not
            in progress, the server closes the control connection.  If
            file transfer is in progress, the connection will remain
            open for result response and the server will then close it.
            If the user-process is transferring files for several USERs
            but does not wish to close and then reopen connections for
            each, then the REIN command should be used instead of QUIT.

            An unexpected close on the control connection will cause the
            server to take the effective action of an abort (ABOR) and a
            logout (QUIT).

Please check that your transfers are over before doing a quit.

Regards,
Antonios.

Max Spevack wrote:

> Hi,
>
> This may seem silly, but for some reason I can't get my program to print
> out the response from the server after I issue the QUIT command.  Other
> server responses are printed out fine, but QUIT never seems to come
> through.
>
> Any ideas?
>
> Thanks,
> Max

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links and File Extensions
Date: Tue, 22 Jan 2002 17:43:12 -0800
Lines: 60
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2688

Excerpt from a previous posting:

<snip>
Correct.


On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:

> Well, we download 'regular files', and recurse into subdirectories.  And
> you seem to be saying to do both here.  So just to confirm:
>
> * If a symlink is pointing to a regular file, download the file, and
> name it with the name of the symlink.

</snip>

This means if we are interested in txt files and we download link.lnk
(as you suggest) the name on the local dir should be link.lnk whose
extension is not txt. Is this acceptable? I thought that all files on
local dir should have the ext specified.

Dhawal Kumar



On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:

> Pete,
> a symbolic link to a file of interest (because of the file's extention) should
> always be followed and treated as a local file, i.e. should be copied. The idea
> of the ftpclient
> is to make local copies of particular files of interest. Symbolic links aren't
> particularly interesting files since copying them locally may create dangling
> references.
>
> So, the answer is yes to the first, no to the second!
>
> Regards,
> Antonios.
>
>
> Pete Belknap wrote:
>
> > What should we do when the symbolic link has a diff extension than the file
> > it points to?
> >
> > So, for example, if we want to download .txt files and we have the link
> >
> > link.lnk  ->  readme.txt
> >
> > Should we download it?  Of course, the opposite question is, if we are
> > interested in files of extension .lnk, should we download the link?  It
> > seems to me that the answer to my first question should be no and the second
> > should be yes...
> >
> > thanks,
> > Pete
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: empty files
Date: Tue, 22 Jan 2002 17:46:02 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma14.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2689


Abhishek,
please refer to theposting regarding empty files (Tue, 22nd)
for useful comments.

Regards,
Antonios.

Abhishek Das wrote:

> hi
>
> I am trying to transfer empty files using ftpd and i am being successful
> in doing so i.e no file created.
> However, the test script says that I transferred 7 bytes on two
> occassions. Can anyone suggest how di I test this? as i said, using ftpd
> says i am OK.
>
> thanks
> abhishek
>
> Abhishek Das
> Graduate Research Assistant
> Computer Systems Lab
> Stanford University
>
> Address:-
> Escondido Village 33B
> Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine31.Stanford.EDU!htimam
From:  (Hasan Taufiq Imam)
Newsgroups: su.class.cs244a
Subject: Removing directory recursively.
Date: 23 Jan 2002 01:46:34 GMT
Organization: Stanford University, CA 94305, USA
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine31.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2690

If a directory with the same name as a remote directory exists in your 
local dir you are supposed to recursively remove it and then create the 
directory. After investigating quite a bit, I could not find any graceful 
way to remove directory recursively. Other than recursive call to rmdir,
remove,chdir. So I opted for system() call. 
Hopefully this is an acceptable solution.

Hasan
.

Path: shelby.stanford.edu!nntp.stanford.edu!fable12.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: ftpd deamon
Date: Tue, 22 Jan 2002 18:02:21 -0800
Lines: 40
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2691


This is a question on ftpd provided for testing:

* The ftpd provides a port number on which to connect to for running ass1.

* Now is this port number in network byte order?

The issue is that if this number is not in network byte order then we can
have problem in following case.

Suppose there are two machines:

Machine A: little endian

Machine B: big endian (same as network byte order)

Now we run ftpd on machine A. It will throw out a port which is used on
machine B for running ass1, assumed to be in A's byte order, i.e. little
endian.

Now the machine B does not know what byte order does machine A comply to.
So there is no way for it to know to convert from what to network(a simple
htons() wont work).

So is it true that the provided byte order is infact network byte order!!
I could not test it myself as I dont have access to a little endian
machine, and in case it is not can we assume that our code wont be tested
against the cases like one specified above??


-gaurav
*************************************************************************
Gaurav Garg                          * Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             *       Escondido Village, Stanford
Department of Electrical Engineering *       CA-94305
Stanford University                  * Ph:650-498-1208
*************************************************************************

-Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Any graceful way of figuring out open descriptors
Date: Tue, 22 Jan 2002 18:36:35 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2692

My program closes all the descriptors if it terminates normally. But in
case of unexpected termination - say due to signal I am unable to close
the open descriptors. Is there a way of figuring out the open descriptors
in such a case?

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga11.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: TEST script problem
Date: Tue, 22 Jan 2002 19:44:07 -0800
Lines: 101
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2693

hi

This is the second time I got this problem out of all my prior
submissions. Please take a look. I had earlier mailed Antonios regarding
this.

I have checked that the tree made my me and as reported below should be
the expected tree (by connecting to the machine). Also the checksum part
is correct since I have compared the actual file sizes and the bytes I
copied to create the files in my machine. BTW, i am working from
saga22.stanford.edu.

thanks
abhishek

THE PROBLEM:-

===============

    * Test D: Connects to server but does not transfer any matching files
in the specified
      levels [out of 3.5]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: ""
---
Program output:

---

===============

    * Test F.1: Connects, but does not transfer all the matching files in
the top level [out of 1]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: ""
---
diff between expected tree and your tree:
0a1,2
> ./xcmalt.tar.Z
> ./xcomm.tar.Z

---


===============

    * Test F.2: Does not transfer all of the matching files in the first
level down to the
      last specified level [out of 1]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: ""
---
diff between expected tree and your tree:
0a1,10
> ./kermit/c/all.tar.Z
> ./kermit/c/amiga.tar.Z
> ./kermit/c/dg.tar.Z
> ./kermit/c/mac.tar.Z
> ./kermit/c/os2.tar.Z
> ./kermit/c/os9.tar.Z
> ./kermit/c/unix.tar.Z
> ./kermit/c/vms.tar.Z
> ./xcmalt.tar.Z
> ./xcomm.tar.Z

---


===============

    * Test G: Transfers all of the matching files, but bad file checksum
[out of 1]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: ""
---
diff between expected tree and your tree:
0a1,10
> ./kermit/c/all.tar.Z 2606058765 72935
> ./kermit/c/amiga.tar.Z 1017293024 135815
> ./kermit/c/dg.tar.Z 2560791635 406317
> ./kermit/c/mac.tar.Z 2459747744 496551
> ./kermit/c/os2.tar.Z 2771080586 163808
> ./kermit/c/os9.tar.Z 370278796 49438
> ./kermit/c/unix.tar.Z 308104872 349221
> ./kermit/c/vms.tar.Z 53741681 64755
> ./xcmalt.tar.Z 536156530 96993
> ./xcomm.tar.Z 2692844690 64389

---

===============


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Hasan Taufiq Imam 
Newsgroups: su.class.cs244a
Subject: ftp.microsoft.com
Date: Tue, 22 Jan 2002 19:51:27 -0800
Lines: 7
Distribution: su
Message-ID: 
Reply-To: 
NNTP-Posting-Host: elaine31.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2694

I have been submitting my code for testing and everything was passing.
Now suddenly transfer from ftp.microsoft.com is failing.
So is for symbolic link to file?
Both of them can be due to timeout of 15s. The script does not tell
us if symbolic link to file is being copied from ftp.mirosoft.com
also. Can somebody with access to the script,please let me know if 
this is the case!
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Problem Set #2 Question1
Date: Tue, 22 Jan 2002 19:57:01 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2695

 Hi,

   I was wondering if, in allocating subnets for University A, do we account
for future growth?  Such as, "we may add another router off of R2" or can
we assume the number of routers and interfaces will remain static? As
always, any comments, tips, thoughs, suggestions etc. are greatly appreciated.
:-)

                    ~~martin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: Re: CheckSum Errors
Date: Tue, 22 Jan 2002 20:11:06 -0800
Lines: 69
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth8.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2696

I'm having the same problem with the submit script for the 204.123.2.2
server, except I don't get a TIMEOUT error.

-----------------------------------------------------------------------------
* Test F.2: Does not transfer all of the matching files in the first
level down to the 
      last specified level [out of 1]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
exit line said: ""
---
diff between expected tree and your tree:
5,8d4
< ./kermit/c/os2.tar.Z
< ./kermit/c/os9.tar.Z
< ./kermit/c/unix.tar.Z
< ./kermit/c/vms.tar.Z

------------------------------------------------------------------------------

However, when i try my program myself, I do get the files that the
script says I'm missing. When I logged onto the server manually (not
using my program), I notced that it was extremely  slow. It took me >3
seconds to get responses to LIST and CWD commands. I'm tempted to blame
the script TIMEOUT, but since I'm not getting a TIMEOUT in my results
I'm a bit confused. Is it possible for the script to be timing out, but
not telling me?

Thanks,
Tim


Lakshman Sundar Maddali wrote:
> 
> Hi,
> 
>    when I run submit_test against my code i get checksum errors and all
> files not transferred errors,( when the program is run against
> 204.123.2.2 server especially) . Also later in the report it says that
> this process got killed due to TIMEOUT.
> could it be because of the program being killed, i am getting the
> checksum errors.
> 
> I wanted to know what is the TIMEOUT value that the testing program is
> configured to, so that i can try with that settings locally.
> 
> Thanks
> ~
> Lakshman Maddali
> -----------------------------------------------------------------------------
>  Test G: Transfers all of the matching files, but bad file checksum [out
> of 1]
> 
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> exit line said: ""
> ---
> diff between expected tree and your tree:
> 4c4,8
> < ./kermit/c/mac.tar.Z 3663175723 5840
> ---
> > ./kermit/c/mac.tar.Z 2459747744 496551
> > ./kermit/c/os2.tar.Z 2771080586 163808
> > ./kermit/c/os9.tar.Z 370278796 49438
> > ./kermit/c/unix.tar.Z 308104872 349221
> > ./kermit/c/vms.tar.Z 53741681 64755
> 
> -----------------------------------------------------------------------------
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Re: TEST script problem
Date: Tue, 22 Jan 2002 20:47:25 -0800
Lines: 113
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2697

I'm having a similar problem. The test script says I am failing the test,
but when I run my program manually, I did get all of the tree with each file
having the correct file size. I'm not sure what happened!

-Henry



"Abhishek Das"  wrote in message

> hi
>
> This is the second time I got this problem out of all my prior
> submissions. Please take a look. I had earlier mailed Antonios regarding
> this.
>
> I have checked that the tree made my me and as reported below should be
> the expected tree (by connecting to the machine). Also the checksum part
> is correct since I have compared the actual file sizes and the bytes I
> copied to create the files in my machine. BTW, i am working from
> saga22.stanford.edu.
>
> thanks
> abhishek
>
> THE PROBLEM:-
>
> ===============
>
>     * Test D: Connects to server but does not transfer any matching files
> in the specified
>       levels [out of 3.5]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> exit line said: ""
> ---
> Program output:
>
> ---
>
> ===============
>
>     * Test F.1: Connects, but does not transfer all the matching files in
> the top level [out of 1]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> exit line said: ""
> ---
> diff between expected tree and your tree:
> 0a1,2
> > ./xcmalt.tar.Z
> > ./xcomm.tar.Z
>
> ---
>
>
> ===============
>
>     * Test F.2: Does not transfer all of the matching files in the first
> level down to the
>       last specified level [out of 1]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> exit line said: ""
> ---
> diff between expected tree and your tree:
> 0a1,10
> > ./kermit/c/all.tar.Z
> > ./kermit/c/amiga.tar.Z
> > ./kermit/c/dg.tar.Z
> > ./kermit/c/mac.tar.Z
> > ./kermit/c/os2.tar.Z
> > ./kermit/c/os9.tar.Z
> > ./kermit/c/unix.tar.Z
> > ./kermit/c/vms.tar.Z
> > ./xcmalt.tar.Z
> > ./xcomm.tar.Z
>
> ---
>
>
> ===============
>
>     * Test G: Transfers all of the matching files, but bad file checksum
> [out of 1]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> exit line said: ""
> ---
> diff between expected tree and your tree:
> 0a1,10
> > ./kermit/c/all.tar.Z 2606058765 72935
> > ./kermit/c/amiga.tar.Z 1017293024 135815
> > ./kermit/c/dg.tar.Z 2560791635 406317
> > ./kermit/c/mac.tar.Z 2459747744 496551
> > ./kermit/c/os2.tar.Z 2771080586 163808
> > ./kermit/c/os9.tar.Z 370278796 49438
> > ./kermit/c/unix.tar.Z 308104872 349221
> > ./kermit/c/vms.tar.Z 53741681 64755
> > ./xcmalt.tar.Z 536156530 96993
> > ./xcomm.tar.Z 2692844690 64389
>
> ---
>
> ===============
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga21.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: TEST script problem
Date: Tue, 22 Jan 2002 21:14:07 -0800
Lines: 135
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2698


I didn't get any TIMEOUT.
I had this problem in one of my earlier submissions with the same server,
but after that all my
submissions passed this test, until ofcourse my last submission. I wonder
what is happening.

abhishek

On Tue, 22 Jan 2002, Henry Fu wrote:

> I'm having a similar problem. The test script says I am failing the test,
> but when I run my program manually, I did get all of the tree with each file
> having the correct file size. I'm not sure what happened!
>
> -Henry
>
>
>
> "Abhishek Das"  wrote in message
> 
> > hi
> >
> > This is the second time I got this problem out of all my prior
> > submissions. Please take a look. I had earlier mailed Antonios regarding
> > this.
> >
> > I have checked that the tree made my me and as reported below should be
> > the expected tree (by connecting to the machine). Also the checksum part
> > is correct since I have compared the actual file sizes and the bytes I
> > copied to create the files in my machine. BTW, i am working from
> > saga22.stanford.edu.
> >
> > thanks
> > abhishek
> >
> > THE PROBLEM:-
> >
> > ===============
> >
> >     * Test D: Connects to server but does not transfer any matching files
> > in the specified
> >       levels [out of 3.5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> > exit line said: ""
> > ---
> > Program output:
> >
> > ---
> >
> > ===============
> >
> >     * Test F.1: Connects, but does not transfer all the matching files in
> > the top level [out of 1]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> > exit line said: ""
> > ---
> > diff between expected tree and your tree:
> > 0a1,2
> > > ./xcmalt.tar.Z
> > > ./xcomm.tar.Z
> >
> > ---
> >
> >
> > ===============
> >
> >     * Test F.2: Does not transfer all of the matching files in the first
> > level down to the
> >       last specified level [out of 1]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> > exit line said: ""
> > ---
> > diff between expected tree and your tree:
> > 0a1,10
> > > ./kermit/c/all.tar.Z
> > > ./kermit/c/amiga.tar.Z
> > > ./kermit/c/dg.tar.Z
> > > ./kermit/c/mac.tar.Z
> > > ./kermit/c/os2.tar.Z
> > > ./kermit/c/os9.tar.Z
> > > ./kermit/c/unix.tar.Z
> > > ./kermit/c/vms.tar.Z
> > > ./xcmalt.tar.Z
> > > ./xcomm.tar.Z
> >
> > ---
> >
> >
> > ===============
> >
> >     * Test G: Transfers all of the matching files, but bad file checksum
> > [out of 1]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy Z 2 204.123.2.2 localdir /pub/comm
> > exit line said: ""
> > ---
> > diff between expected tree and your tree:
> > 0a1,10
> > > ./kermit/c/all.tar.Z 2606058765 72935
> > > ./kermit/c/amiga.tar.Z 1017293024 135815
> > > ./kermit/c/dg.tar.Z 2560791635 406317
> > > ./kermit/c/mac.tar.Z 2459747744 496551
> > > ./kermit/c/os2.tar.Z 2771080586 163808
> > > ./kermit/c/os9.tar.Z 370278796 49438
> > > ./kermit/c/unix.tar.Z 308104872 349221
> > > ./kermit/c/vms.tar.Z 53741681 64755
> > > ./xcmalt.tar.Z 536156530 96993
> > > ./xcomm.tar.Z 2692844690 64389
> >
> > ---
> >
> > ===============
> >
> >
>
>
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: how to find name of file a link is pointing to
Date: Tue, 22 Jan 2002 21:29:30 -0800
Lines: 58
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2699

if a link l1.lnk points to a file a.txt how do we detect the name of the
file? the response of LIST contains this, but ftpparse just returns the
name of the link.
how then can we figure out that since we are interested in .txt files, we
should also download a file pointed to by a .lnk link?

please clarify! i am really confused!
thanks,
ashmi


On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:

> Pete,
> a symbolic link to a file of interest (because of the file's extention) should
> always be followed and treated as a local file, i.e. should be copied. The idea
> of the ftpclient
> is to make local copies of particular files of interest. Symbolic links aren't
> particularly interesting files since copying them locally may create dangling
> references.
>
> So, the answer is yes to the first, no to the second!
>
> Regards,
> Antonios.
>
>
> Pete Belknap wrote:
>
> > What should we do when the symbolic link has a diff extension than the file
> > it points to?
> >
> > So, for example, if we want to download .txt files and we have the link
> >
> > link.lnk  ->  readme.txt
> >
> > Should we download it?  Of course, the opposite question is, if we are
> > interested in files of extension .lnk, should we download the link?  It
> > seems to me that the answer to my first question should be no and the second
> > should be yes...
> >
> > thanks,
> > Pete
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: Re: Duplicate Files and Circularities
Date: Tue, 22 Jan 2002 21:31:45 -0800
Lines: 56
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2700


Detecting that two symbolic links in the server file system point to the
same file and transfering it only once, or detecting that server has a
loop in its directory structure (symlink to parent) are not required by
the assignment handout. I would consider this answer with a grain of
doubt.

Alex

On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:

>
> Hi Jian,
>
> Yes, your code should be written in such a way that no
> file is downloaded more than once. Cases with symlinks
> to the current directory or to parent directories that you've
> already processed are troublesome cases that you need
> to take care of, in order to avoid downloading twice a file.
>
> Regards,
> Antonios.
>
> Jian Deng wrote:
>
> > I have question on the following test case
> > special cases: Symbolic-link-to-directory [out of 0.5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  Symbolic link to directory in same directory did transfer (wrong!)
> > exit line said: ""
> > ---
> > Your output to stdout was:
> > OK: 14 bytes copied
> >
> > Is this test case testing a transfer should not
> > happen to a symbolic link to the same dir that has already
> > been transfered?
> >
> > I am wondering how can we prevent that with out parsing the
> > actual output of the list which varies with ftp servers..
> > and PWD will not give you the actual path.
> >
> > Does this test case also imply we should handle cycles and not transfer
> > dirs already been transfered?
> >
> > Your clarification is appreciated.
> >
> >
>
>

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: Re: how to find name of file a link is pointing to
Date: Tue, 22 Jan 2002 21:34:44 -0800
Lines: 75
Distribution: su
Message-ID: 
References:  
 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2701


Here is my humble opinion: the name of the file to which the link is
pointing does not matter. For all practical purposes the name of the link
is the name of the file you RETRieve and write locally. The .lnk example
is confusing since on UNIX symlinks are not required to have a specific
extension.

Alex

On Tue, 22 Jan 2002, Ashmi wrote:

> if a link l1.lnk points to a file a.txt how do we detect the name of the
> file? the response of LIST contains this, but ftpparse just returns the
> name of the link.
> how then can we figure out that since we are interested in .txt files, we
> should also download a file pointed to by a .lnk link?
>
> please clarify! i am really confused!
> thanks,
> ashmi
>
>
> On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
>
> > Pete,
> > a symbolic link to a file of interest (because of the file's extention) should
> > always be followed and treated as a local file, i.e. should be copied. The idea
> > of the ftpclient
> > is to make local copies of particular files of interest. Symbolic links aren't
> > particularly interesting files since copying them locally may create dangling
> > references.
> >
> > So, the answer is yes to the first, no to the second!
> >
> > Regards,
> > Antonios.
> >
> >
> > Pete Belknap wrote:
> >
> > > What should we do when the symbolic link has a diff extension than the file
> > > it points to?
> > >
> > > So, for example, if we want to download .txt files and we have the link
> > >
> > > link.lnk  ->  readme.txt
> > >
> > > Should we download it?  Of course, the opposite question is, if we are
> > > interested in files of extension .lnk, should we download the link?  It
> > > seems to me that the answer to my first question should be no and the second
> > > should be yes...
> > >
> > > thanks,
> > > Pete
> >
> >
>
> ###############################################################################
>
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
>
> ###############################################################################
>
>
>
>

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: how to find name of file a link is pointing to
Date: Tue, 22 Jan 2002 21:50:55 -0800
Lines: 112
Distribution: su
Message-ID: 
References:  
 
 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2702

my idea exactly.
but the earlier mail exchange between pete and antonios confuses me.
pete's questiion:
> So, for example, if we want to download .txt files and we have the
>link   link.lnk  ->  readme.txt
> Should we download it?  Of course, the opposite question is, if we
>are interested in files of extension .lnk, should we download the
>link?  It
> seems to me that the answer to my first question should be no and
>the second
> should be yes...

antonios replied:
>So, the answer is yes to the first, no to the second!

ideas anyone? i cant figure out how to find the name of the file the link
points to
thanks,
ashmi


On Tue, 22 Jan 2002, Alex Khomenko wrote:

>
> Here is my humble opinion: the name of the file to which the link is
> pointing does not matter. For all practical purposes the name of the link
> is the name of the file you RETRieve and write locally. The .lnk example
> is confusing since on UNIX symlinks are not required to have a specific
> extension.
>
> Alex
>
> On Tue, 22 Jan 2002, Ashmi wrote:
>
> > if a link l1.lnk points to a file a.txt how do we detect the name of the
> > file? the response of LIST contains this, but ftpparse just returns the
> > name of the link.
> > how then can we figure out that since we are interested in .txt files, we
> > should also download a file pointed to by a .lnk link?
> >
> > please clarify! i am really confused!
> > thanks,
> > ashmi
> >
> >
> > On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
> >
> > > Pete,
> > > a symbolic link to a file of interest (because of the file's extention) should
> > > always be followed and treated as a local file, i.e. should be copied. The idea
> > > of the ftpclient
> > > is to make local copies of particular files of interest. Symbolic links aren't
> > > particularly interesting files since copying them locally may create dangling
> > > references.
> > >
> > > So, the answer is yes to the first, no to the second!
> > >
> > > Regards,
> > > Antonios.
> > >
> > >
> > > Pete Belknap wrote:
> > >
> > > > What should we do when the symbolic link has a diff extension than the file
> > > > it points to?
> > > >
> > > > So, for example, if we want to download .txt files and we have the link
> > > >
> > > > link.lnk  ->  readme.txt
> > > >
> > > > Should we download it?  Of course, the opposite question is, if we are
> > > > interested in files of extension .lnk, should we download the link?  It
> > > > seems to me that the answer to my first question should be no and the second
> > > > should be yes...
> > > >
> > > > thanks,
> > > > Pete
> > >
> > >
> >
> > ###############################################################################
> >
> > 				ASHMI CHOKSHI
> > Graduate Student				    141L Escondido Village
> > Dept. of Computer Science			    Stanford University
> > Stanford University				    Stanford Ca 94305
> > Stanford Ca 94305				    (650)498-1103
> >
> > ###############################################################################
> >
> >
> >
> >
>
> --
> Knowledge is a deadly friend when no one sets the rules.
> The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Removing directory recursively.
Date: Tue, 22 Jan 2002 22:07:16 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Hasan Taufiq Imam 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2703

You can use system() call. Xin


On 23 Jan 2002, Hasan Taufiq Imam wrote:

> If a directory with the same name as a remote directory exists in your 
> local dir you are supposed to recursively remove it and then create the 
> directory. After investigating quite a bit, I could not find any graceful 
> way to remove directory recursively. Other than recursive call to rmdir,
> remove,chdir. So I opted for system() call. 
> Hopefully this is an acceptable solution.
> 
> Hasan
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: Problem Set #2 Question1
Date: Tue, 22 Jan 2002 22:13:04 -0800
Organization: Stanford University
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1011766301 17261 171.66.211.98 (23 Jan 2002 06:11:41 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2704

All good questions.

One thing you'll notice is that there are several correct
answers. Any correct answer is OK. Accounting for
future growth would be good, but is not required.

- Nick



Martin Casado wrote:
> 
>  Hi,
> 
>    I was wondering if, in allocating subnets for University A, do we account
> for future growth?  Such as, "we may add another router off of R2" or can
> we assume the number of routers and interfaces will remain static? As
> always, any comments, tips, thoughs, suggestions etc. are greatly appreciated.
> :-)
> 
>                     ~~martin
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Vasco Chatalbashev" 
Newsgroups: su.class.cs244a
Subject: Re: Removing directory recursively.
Date: Tue, 22 Jan 2002 22:21:28 -0800
Lines: 39
Distribution: su
Message-ID: 
References:  
Reply-To: "Vasco Chatalbashev" 
NNTP-Posting-Host: vassil.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.3018.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.3018.1300
Xref: nntp.stanford.edu su.class.cs244a:2705

The assignment handout's only reference to existing files/directories
is "You should overwrite any files which already exist without prompting."
There is no mention of recursively removing the existing directory.
I think that a more reasonable and less destructive behavior is to just
overwrite existing files,
leaving alone the ones that don't interfere with the transfer of the
remote tree (even though they may reside in the  local directory to which we
are copying).
Am I wrong?


"Xin Wang"  wrote in message

> You can use system() call. Xin
>
>
> On 23 Jan 2002, Hasan Taufiq Imam wrote:
>
> > If a directory with the same name as a remote directory exists in your
> > local dir you are supposed to recursively remove it and then create the
> > directory. After investigating quite a bit, I could not find any
graceful
> > way to remove directory recursively. Other than recursive call to rmdir,
> > remove,chdir. So I opted for system() call.
> > Hopefully this is an acceptable solution.
> >
> > Hasan
> >
>
> --
> ---------------------------------
> Xin Wang
>
> Department of Computer Science
> Stanford University
>
> 


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: Re: Other files in directory.
Date: Tue, 22 Jan 2002 22:34:11 -0800
Lines: 13
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: myth8.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2706

> Your
> code should make sure no extra files exist in the new, local tree you're
> trying to create.

Does that mean we delete all files in the local directory - essentially
doing a rm -rf localdir/* every time we run ftpcopy? My interpretation
of "overwrite any files which already exists" was that, if we want to
download hello.gz, and we already have a copy of it in localdir, then we
overwrite our local copy (leaving the "extra" files alone). Is my
interpretation incorrect?

Thanks, 
Tim
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga21.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: how to find name of file a link is pointing to
Date: Tue, 22 Jan 2002 22:34:38 -0800
Lines: 135
Distribution: su
Message-ID: 
References:  
 
 
 
NNTP-Posting-Host: saga21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2707

hi
Well, what I suggest is when you get the listing (LIST), you get the
path to the actual file, eg. link.lnk -> garbage/text/link.txt
so, you can find out if the actual file is with txt extension. However,
this will land you in problem when the link.txt itself turns out to be a
link to say some link.gz
I am not sure how to handle that.

abhishek

On Tue, 22 Jan 2002, Ashmi wrote:

> my idea exactly.
> but the earlier mail exchange between pete and antonios confuses me.
> pete's questiion:
> > So, for example, if we want to download .txt files and we have the
> >link   link.lnk  ->  readme.txt
> > Should we download it?  Of course, the opposite question is, if we
> >are interested in files of extension .lnk, should we download the
> >link?  It
> > seems to me that the answer to my first question should be no and
> >the second
> > should be yes...
>
> antonios replied:
> >So, the answer is yes to the first, no to the second!
>
> ideas anyone? i cant figure out how to find the name of the file the link
> points to
> thanks,
> ashmi
>
>
> On Tue, 22 Jan 2002, Alex Khomenko wrote:
>
> >
> > Here is my humble opinion: the name of the file to which the link is
> > pointing does not matter. For all practical purposes the name of the link
> > is the name of the file you RETRieve and write locally. The .lnk example
> > is confusing since on UNIX symlinks are not required to have a specific
> > extension.
> >
> > Alex
> >
> > On Tue, 22 Jan 2002, Ashmi wrote:
> >
> > > if a link l1.lnk points to a file a.txt how do we detect the name of the
> > > file? the response of LIST contains this, but ftpparse just returns the
> > > name of the link.
> > > how then can we figure out that since we are interested in .txt files, we
> > > should also download a file pointed to by a .lnk link?
> > >
> > > please clarify! i am really confused!
> > > thanks,
> > > ashmi
> > >
> > >
> > > On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
> > >
> > > > Pete,
> > > > a symbolic link to a file of interest (because of the file's extention) should
> > > > always be followed and treated as a local file, i.e. should be copied. The idea
> > > > of the ftpclient
> > > > is to make local copies of particular files of interest. Symbolic links aren't
> > > > particularly interesting files since copying them locally may create dangling
> > > > references.
> > > >
> > > > So, the answer is yes to the first, no to the second!
> > > >
> > > > Regards,
> > > > Antonios.
> > > >
> > > >
> > > > Pete Belknap wrote:
> > > >
> > > > > What should we do when the symbolic link has a diff extension than the file
> > > > > it points to?
> > > > >
> > > > > So, for example, if we want to download .txt files and we have the link
> > > > >
> > > > > link.lnk  ->  readme.txt
> > > > >
> > > > > Should we download it?  Of course, the opposite question is, if we are
> > > > > interested in files of extension .lnk, should we download the link?  It
> > > > > seems to me that the answer to my first question should be no and the second
> > > > > should be yes...
> > > > >
> > > > > thanks,
> > > > > Pete
> > > >
> > > >
> > >
> > > ###############################################################################
> > >
> > > 				ASHMI CHOKSHI
> > > Graduate Student				    141L Escondido Village
> > > Dept. of Computer Science			    Stanford University
> > > Stanford University				    Stanford Ca 94305
> > > Stanford Ca 94305				    (650)498-1103
> > >
> > > ###############################################################################
> > >
> > >
> > >
> > >
> >
> > --
> > Knowledge is a deadly friend when no one sets the rules.
> > The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
> >
> >
>
> ###############################################################################
>
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
>
> ###############################################################################
>
>
>
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Removing directory recursively.
Date: Tue, 22 Jan 2002 22:44:16 -0800
Lines: 55
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Vasco Chatalbashev 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2708

Either way is fine. Actually we won't be testing the case that local
directory and remote directory have name conflict. Xin


On Tue, 22 Jan 2002, Vasco Chatalbashev wrote:

> The assignment handout's only reference to existing files/directories
> is "You should overwrite any files which already exist without prompting."
> There is no mention of recursively removing the existing directory.
> I think that a more reasonable and less destructive behavior is to just
> overwrite existing files,
> leaving alone the ones that don't interfere with the transfer of the
> remote tree (even though they may reside in the  local directory to which we
> are copying).
> Am I wrong?
> 
> 
> "Xin Wang"  wrote in message
> 
> > You can use system() call. Xin
> >
> >
> > On 23 Jan 2002, Hasan Taufiq Imam wrote:
> >
> > > If a directory with the same name as a remote directory exists in your
> > > local dir you are supposed to recursively remove it and then create the
> > > directory. After investigating quite a bit, I could not find any
> graceful
> > > way to remove directory recursively. Other than recursive call to rmdir,
> > > remove,chdir. So I opted for system() call.
> > > Hopefully this is an acceptable solution.
> > >
> > > Hasan
> > >
> >
> > --
> > ---------------------------------
> > Xin Wang
> >
> > Department of Computer Science
> > Stanford University
> >
> > 
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jing Jiang" 
Newsgroups: su.class.cs244a
Subject: Re: proper RETR and LIST behavior
Date: Tue, 22 Jan 2002 23:16:57 -0800
Lines: 53
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: jjiang.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2314.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
Xref: nntp.stanford.edu su.class.cs244a:2709

So according to RFC, a pathname (possibly including the path and the file
name, to my understanding) is what follows RETR. Then what about
ftp.stanford.edu server which doesn't allow this, as Joong pointed out? Can
we treat it as non-standard and ignore it?

Jing

SUL\\AIR Laneroom Cluster  wrote in message

>
> Joong,
> here's the part of the RFC that explains RETR:
>
>        RETRIEVE (RETR)
>
>             This command causes the server-DTP to transfer a copy of the
>             file, specified in the pathname, to the server- or user-DTP
>             at the other end of the data connection.  The status and
>             contents of the file at the server site shall be unaffected.
>
>
> Please keep in mind that a pathname and a filename are indeed different!
> Regards,
> Antonios.
>
> Joong Ho Won wrote:
>
> > Hello,
> >
> > I wonder if RETR command can take an absoulte/relative path to a file as
an
> > argument. i.e.,
> >
> > RETR /a/path/to/a/file
> >
> > rather than having just a filename, i.e.,
> >
> > RETR file
> >
> > RFC doesn't seem to specify this case.
> >
> > Some ftp(ex. ftp.stanford.edu ) seems not to allow the former case, but
> > others(ex. transfer.stanford.edu ) seems to.
> >
> > I'd like to make it clear for LIST for the same case as well, though it
> > seems to be allowed seeing RFC.
> >
> > Thanks.
> >
> > Joong Ho Won
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: how to find name of file a link is pointing to
Date: Tue, 22 Jan 2002 23:14:58 -0800
Lines: 176
Distribution: su
Message-ID: 
References:      
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2710


Hi...

I'm also one of them who think that only the name of link matters, not the
name of the file/dir to which
the link points. And, isn't it the way  most of unix commands work?
And then, that way of operation is already taken care of by the OS of the
FTP server,
 so I don't see why we should bother...
link.lnk -> link.txt  being processed for 'txt' and not for 'lnk' ; sounds
really strange to me..


"Abhishek Das"  wrote in message

> hi
> Well, what I suggest is when you get the listing (LIST), you get the
> path to the actual file, eg. link.lnk -> garbage/text/link.txt
> so, you can find out if the actual file is with txt extension. However,
> this will land you in problem when the link.txt itself turns out to be a
> link to say some link.gz
> I am not sure how to handle that.
>
> abhishek
>
> On Tue, 22 Jan 2002, Ashmi wrote:
>
> > my idea exactly.
> > but the earlier mail exchange between pete and antonios confuses me.
> > pete's questiion:
> > > So, for example, if we want to download .txt files and we have the
> > >link   link.lnk  ->  readme.txt
> > > Should we download it?  Of course, the opposite question is, if we
> > >are interested in files of extension .lnk, should we download the
> > >link?  It
> > > seems to me that the answer to my first question should be no and
> > >the second
> > > should be yes...
> >
> > antonios replied:
> > >So, the answer is yes to the first, no to the second!
> >
> > ideas anyone? i cant figure out how to find the name of the file the
link
> > points to
> > thanks,
> > ashmi
> >
> >
> > On Tue, 22 Jan 2002, Alex Khomenko wrote:
> >
> > >
> > > Here is my humble opinion: the name of the file to which the link is
> > > pointing does not matter. For all practical purposes the name of the
link
> > > is the name of the file you RETRieve and write locally. The .lnk
example
> > > is confusing since on UNIX symlinks are not required to have a
specific
> > > extension.
> > >
> > > Alex
> > >
> > > On Tue, 22 Jan 2002, Ashmi wrote:
> > >
> > > > if a link l1.lnk points to a file a.txt how do we detect the name of
the
> > > > file? the response of LIST contains this, but ftpparse just returns
the
> > > > name of the link.
> > > > how then can we figure out that since we are interested in .txt
files, we
> > > > should also download a file pointed to by a .lnk link?
> > > >
> > > > please clarify! i am really confused!
> > > > thanks,
> > > > ashmi
> > > >
> > > >
> > > > On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
> > > >
> > > > > Pete,
> > > > > a symbolic link to a file of interest (because of the file's
extention) should
> > > > > always be followed and treated as a local file, i.e. should be
copied. The idea
> > > > > of the ftpclient
> > > > > is to make local copies of particular files of interest. Symbolic
links aren't
> > > > > particularly interesting files since copying them locally may
create dangling
> > > > > references.
> > > > >
> > > > > So, the answer is yes to the first, no to the second!
> > > > >
> > > > > Regards,
> > > > > Antonios.
> > > > >
> > > > >
> > > > > Pete Belknap wrote:
> > > > >
> > > > > > What should we do when the symbolic link has a diff extension
than the file
> > > > > > it points to?
> > > > > >
> > > > > > So, for example, if we want to download .txt files and we have
the link
> > > > > >
> > > > > > link.lnk  ->  readme.txt
> > > > > >
> > > > > > Should we download it?  Of course, the opposite question is, if
we are
> > > > > > interested in files of extension .lnk, should we download the
link?  It
> > > > > > seems to me that the answer to my first question should be no
and the second
> > > > > > should be yes...
> > > > > >
> > > > > > thanks,
> > > > > > Pete
> > > > >
> > > > >
> > > >
> > > >
############################################################################
###
> > > >
> > > > ASHMI CHOKSHI
> > > > Graduate Student     141L Escondido Village
> > > > Dept. of Computer Science     Stanford University
> > > > Stanford University     Stanford Ca 94305
> > > > Stanford Ca 94305     (650)498-1103
> > > >
> > > >
############################################################################
###
> > > >
> > > >
> > > >
> > > >
> > >
> > > --
> > > Knowledge is a deadly friend when no one sets the rules.
> > > The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
> > >
> > >
> >
> >
############################################################################
###
> >
> > ASHMI CHOKSHI
> > Graduate Student     141L Escondido Village
> > Dept. of Computer Science     Stanford University
> > Stanford University     Stanford Ca 94305
> > Stanford Ca 94305     (650)498-1103
> >
> >
############################################################################
###
> >
> >
> >
> >
>
> Abhishek Das
> Graduate Research Assistant
> Computer Systems Lab
> Stanford University
>
> Address:-
> Escondido Village 33B
> Stanford CA 94305
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Re: IMPORTANT CLARIFICATION: failure on ftp.slac.stanford.edu tests
Date: Tue, 22 Jan 2002 23:35:06 -0800
Lines: 43
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2711

I'm failling all my test using the submit_test script. I wonder what happen?
Is there anything I need to take notice?

For the ftp.slac.stanford.edu issue, when I try to run the test on my own,
I can finish the test and got the correct
tree structure and got the correct file size for each file. And I finish
the whole thing in less than 5 seconds, so TIMEOUT shouldn't be a
problem to me. So, I"m not sure what's going on.

Henry


"Arun Upadhyaya Kishan"  wrote in message

> Hello all,
>
> A lot of people have been sending in emails to the staff stating that the
> submit test script is reporting failures on various ftp.slac.stanford.edu,
> such as TIMEOUT or exit line incorrect, or tree incorrect, etc.
>
> Note that the script determines these errors by examining the output of a
> single program run. In particular, if this test does not terminate in
> TIMEOUT time, it is highly likely that all the remaining tests will fail.
> Since this is a relatively deep recursion, the time limit is imposed to
> penalize programs with poor performance. Alternatively, programs that hang
> or recurse will force the program to be abnormally terminated as well.
>
> If you see a TIMEOUT error for a given test, please look into what could
> be causing this. Two things to consider:
>
> a) Do you read data from the socket when there is no data available (i.e.,
> expecting a reply when there should be none)? This will hang your process
>
> b) During directory recursion, do you ignore the "." and ".." entries?
> Some, not all, ftp servers return these as directory names as part of the
> LIST response. Among these is ftp.slac.stanford.edu.
>
> Good luck,
>
> Arun
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!fable8.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: TEST scripts seem to be not working!!
Date: Tue, 22 Jan 2002 23:39:03 -0800
Lines: 65
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: fable8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2712


I am also failing all the tests on test_script when I ran it half an hour
back. They were working fine with the test_script 4 hours back. I then
tried testing them myself and they seemed to work just fine. Are the
test_scripts broken.

-gaurav


On Tue, 22 Jan 2002, Henry Fu wrote:

> I'm failling all my test using the submit_test script. I wonder what happen?
> Is there anything I need to take notice?
>
> For the ftp.slac.stanford.edu issue, when I try to run the test on my own,
> I can finish the test and got the correct
> tree structure and got the correct file size for each file. And I finish
> the whole thing in less than 5 seconds, so TIMEOUT shouldn't be a
> problem to me. So, I"m not sure what's going on.
>
> Henry
>
>
> "Arun Upadhyaya Kishan"  wrote in message
> 
> > Hello all,
> >
> > A lot of people have been sending in emails to the staff stating that the
> > submit test script is reporting failures on various ftp.slac.stanford.edu,
> > such as TIMEOUT or exit line incorrect, or tree incorrect, etc.
> >
> > Note that the script determines these errors by examining the output of a
> > single program run. In particular, if this test does not terminate in
> > TIMEOUT time, it is highly likely that all the remaining tests will fail.
> > Since this is a relatively deep recursion, the time limit is imposed to
> > penalize programs with poor performance. Alternatively, programs that hang
> > or recurse will force the program to be abnormally terminated as well.
> >
> > If you see a TIMEOUT error for a given test, please look into what could
> > be causing this. Two things to consider:
> >
> > a) Do you read data from the socket when there is no data available (i.e.,
> > expecting a reply when there should be none)? This will hang your process
> >
> > b) During directory recursion, do you ignore the "." and ".." entries?
> > Some, not all, ftp servers return these as directory names as part of the
> > LIST response. Among these is ftp.slac.stanford.edu.
> >
> > Good luck,
> >
> > Arun
> >
>
>
>

*************************************************************************
Gaurav Garg                          * Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             *       Escondido Village, Stanford
Department of Electrical Engineering *       CA-94305
Stanford University                  * Ph:650-498-1208
*************************************************************************

-Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Eric Watkins 
Newsgroups: su.class.cs244a
Subject: Re: Regarding Post of 1/16/02, 8:56: Script File Sizes
Date: Tue, 22 Jan 2002 23:40:36 -0800
Lines: 25
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: e-man.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2713

> exit line: ""

That always happens to me too, and I don't know why...


> I'm also having problems with tests I and L. I think the FTP server on
> epic is messed up, because I get:
> 
> Your output to stdout was:
> TIMEOUT! killed program: /afs/ir/class/cs244a/submissions/restricted/hw1/hondroul/oceans//compile.sun/ftpcopy ext 1  /afs/ir/class/cs244a/submissions/restricted/hw1/hondroul/oceans//compile.sun/download/perms .   21715


I think I know why this is.  I grep'd through the output for CWD and
noticed that at the very beginning my program was doing "CWD ." and "CWD
..." an awful lot.  I think the program spent all its time cwd'ing around
in circles and not transfering any files.  If you have your program
ignores the directories "." and ".." it should fix that right up.

> : -When the test lists diffs, do >'s mean that your tree had certain files and
> : that the correct one didn't (ie you shouldn't have had them)?

< means what you had
> means what you should have had

-Eric...
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Empty submit directory?
Date: 23 Jan 2002 08:31:47 GMT
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2714

I've submitted my program (twice), but I noticed that my folder directory
in submissions/hw1/ta/username/ is empty.

Is this normal?
.

Path: shelby.stanford.edu!nntp.stanford.edu!spevack.Stanford.EDU!spevack
From: Max Spevack 
Newsgroups: su.class.cs244a
Subject: End of Data transmission?
Date: Wed, 23 Jan 2002 01:17:31 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: spevack.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2715

Hi,

I'm been going in circles for hours on this, and I would really appreciate
some help.

How do I know when I'm done with a data transmission?  For example, what
indicates to me that LIST is done sending the contents of the directory?

I'm reading from the socket byte by byte, but if I check to see if read
returns a result less than 0, that doesn't work and I end up in an
infinite loop in which it prints out the dir contents and then prints the
very first directory entry over and over again.

Thanks for any input.

--Max

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lin Lu 
Newsgroups: su.class.cs244a
Subject: fix the permission of ftpd
Date: Wed, 23 Jan 2002 01:21:18 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2716

Hi,

I can not run ftpd server since permission denied.  Could any TA fix
this.  I need using it to debug.

thanks,
Lin Lu
.

Path: shelby.stanford.edu!nntp.stanford.edu!fable5.Stanford.EDU!pabbeel
From: Pieter Abbeel 
Newsgroups: su.class.cs244a
Subject: Re: End of Data transmission?
Date: Wed, 23 Jan 2002 01:31:48 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: fable5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2717

check for ==0, not <0
--pieter

On Wed, 23 Jan 2002, Max Spevack wrote:

> Hi,
>
> I'm been going in circles for hours on this, and I would really appreciate
> some help.
>
> How do I know when I'm done with a data transmission?  For example, what
> indicates to me that LIST is done sending the contents of the directory?
>
> I'm reading from the socket byte by byte, but if I check to see if read
> returns a result less than 0, that doesn't work and I end up in an
> infinite loop in which it prints out the dir contents and then prints the
> very first directory entry over and over again.
>
> Thanks for any input.
>
> --Max
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga12.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: End of Data transmission?
Date: Wed, 23 Jan 2002 01:34:08 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2718

In case of LIST command there are two connections:
(1) Control Connection: A response of 226 indicates success. Read control
socket until you either get an error code 4xy or 5xy or 226 after you have
read the data
(2) Data Connection: The data transmission is over when server closes the
socket. You get a 0 sized packet (read returns 0) to indicate this.

Dhawal Kumar

On Wed, 23 Jan 2002, Max Spevack wrote:

> Hi,
>
> I'm been going in circles for hours on this, and I would really appreciate
> some help.
>
> How do I know when I'm done with a data transmission?  For example, what
> indicates to me that LIST is done sending the contents of the directory?
>
> I'm reading from the socket byte by byte, but if I check to see if read
> returns a result less than 0, that doesn't work and I end up in an
> infinite loop in which it prints out the dir contents and then prints the
> very first directory entry over and over again.
>
> Thanks for any input.
>
> --Max
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic2.Stanford.EDU!ambu
From: Ambika Balakrishnan 
Newsgroups: su.class.cs244a
Subject: STRU, MODE and QUIT ?
Date: Wed, 23 Jan 2002 02:40:11 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Arun Upadhyaya Kishan 
Xref: nntp.stanford.edu su.class.cs244a:2719

Hi!

The RFC seems to say that file STRUcture and stream transmission MODE are
the defaults for all ftp servers, even those that do not implement the
STRU and MODE commands. So is it all right if I do not send these commands
to the server?

Also, do we need to send the QUIT command to the server? What purpose does
this serve?

Thank you!
--
Ambika


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Mudassir I Sheikha" 
Newsgroups: su.class.cs244a
Subject: Re: Regarding Post of 1/16/02, 8:56: Script File Sizes
Date: Wed, 23 Jan 2002 03:56:55 -0800
Lines: 53
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: pakistan.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4807.1700
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4807.1700
Xref: nntp.stanford.edu su.class.cs244a:2720

i see the same messages.

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy ch 1 ftp.cs.stanford.edu localdir /pub/cweb
exit line said: ""
---

also, are we supposed to treat 'localdir' as the current directory? my
current implementation creates a directory 'localdir' under the current
directory. test results, however, seem to suggest that localdir points to
the current directory.

mudassir

--
Mudassir I Sheikha
650-219-8800 (Cell)


"Eric Watkins"  wrote in message

> > exit line: ""
>
> That always happens to me too, and I don't know why...
>
>
> > I'm also having problems with tests I and L. I think the FTP server on
> > epic is messed up, because I get:
> >
> > Your output to stdout was:
> > TIMEOUT! killed program:
/afs/ir/class/cs244a/submissions/restricted/hw1/hondroul/oceans//compile.sun
/ftpcopy ext 1 
/afs/ir/class/cs244a/submissions/restricted/hw1/hondroul/oceans//compile.sun
/download/perms .   21715
>
>
> I think I know why this is.  I grep'd through the output for CWD and
> noticed that at the very beginning my program was doing "CWD ." and "CWD
> .." an awful lot.  I think the program spent all its time cwd'ing around
> in circles and not transfering any files.  If you have your program
> ignores the directories "." and ".." it should fix that right up.
>
> > : -When the test lists diffs, do >'s mean that your tree had certain
files and
> > : that the correct one didn't (ie you shouldn't have had them)?
>
> < means what you had
> > means what you should have had
>
> -Eric...


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: left over from previous read
Date: Wed, 23 Jan 2002 07:01:47 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2721


Hi,

Even though I am making sure that each command that I write and then read,
I read till th every end, until I get the last line which has a space
after the response.

Somehow when I do the next read, the stuff from the previoud read is still
there, even though I clear up my reply buffer, it seems the server still
sens the stuff from the previous read. Is there any way to flush the
server contenets after every read.

Any help would be gretaly appreciated.

Thanks,
Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: Re: left over from previous read
Date: Wed, 23 Jan 2002 07:06:41 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2722


Also, this might seem bizzare but if I step thru gdb to check where the
buffer might be giving a problem...it goes thru fine as if the server gets
more time to clear out stuff! Don't know if this makes any sense but thats
whats happening.

Any help?

 On Wed, 23 Jan 2002, Priyanka Gupta wrote:

>
> Hi,
>
> Even though I am making sure that each command that I write and then read,
> I read till th every end, until I get the last line which has a space
> after the response.
>
> Somehow when I do the next read, the stuff from the previoud read is still
> there, even though I clear up my reply buffer, it seems the server still
> sens the stuff from the previous read. Is there any way to flush the
> server contenets after every read.
>
> Any help would be gretaly appreciated.
>
> Thanks,
> Priyanka
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine5.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Homework 2.
Date: Wed, 23 Jan 2002 10:07:37 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2723

Hi,
I have few quesitons regarding the howework 2.
a) First of all in the question 6 it is said that traceroute to
www.berkeley.com and www.cisco.com takes separate separate path out of
campus but i don't see that happening. Is there any site in particular
that i should try rather then doing hit and trial.
b) while trying to get the host name for some of the routers in the
question 5 i got
elaine5:~> host C2-QANH-GSR-QSV-GSR.ATM.calren2.net
C2-QANH-GSR-QSV-GSR.ATM.calren2.net does not exist, try again .
 Is this ok or is there any other way to find out the ip address of
the routers.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: fix the permission of ftpd
Date: Wed, 23 Jan 2002 10:10:30 -0800
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
To: Lin Lu 
Xref: nntp.stanford.edu su.class.cs244a:2724

I have just changed the directory permissions,

pablo

Lin Lu wrote:

> Hi,
> 
> I can not run ftpd server since permission denied.  Could any TA fix
> this.  I need using it to debug.
> 
> thanks,
> Lin Lu
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Duplicate Files and Circularities
Date: Wed, 23 Jan 2002 10:13:20 -0800
Lines: 63
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
To: Alex Khomenko 
Xref: nntp.stanford.edu su.class.cs244a:2725

Hi,

In the grading only very simple symbolic links will be tested, like the 
one in the testing web page. Do not worry about complicated recursive 
links or links to different directory levels.

Pablo

Alex Khomenko wrote:

> Detecting that two symbolic links in the server file system point to the
> same file and transfering it only once, or detecting that server has a
> loop in its directory structure (symlink to parent) are not required by
> the assignment handout. I would consider this answer with a grain of
> doubt.
> 
> Alex
> 
> On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
> 
> 
>>Hi Jian,
>>
>>Yes, your code should be written in such a way that no
>>file is downloaded more than once. Cases with symlinks
>>to the current directory or to parent directories that you've
>>already processed are troublesome cases that you need
>>to take care of, in order to avoid downloading twice a file.
>>
>>Regards,
>>Antonios.
>>
>>Jian Deng wrote:
>>
>>
>>>I have question on the following test case
>>>special cases: Symbolic-link-to-directory [out of 0.5]
>>>
>>>[SUMMARY] NOT OK
>>>[RESULT]  Symbolic link to directory in same directory did transfer (wrong!)
>>>exit line said: ""
>>>---
>>>Your output to stdout was:
>>>OK: 14 bytes copied
>>>
>>>Is this test case testing a transfer should not
>>>happen to a symbolic link to the same dir that has already
>>>been transfered?
>>>
>>>I am wondering how can we prevent that with out parsing the
>>>actual output of the list which varies with ftp servers..
>>>and PWD will not give you the actual path.
>>>
>>>Does this test case also imply we should handle cycles and not transfer
>>>dirs already been transfered?
>>>
>>>Your clarification is appreciated.
>>>
>>>
>>>
>>
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!priyanka
From: Priyanka Gupta 
Newsgroups: su.class.cs244a
Subject: ftp.slac.stanford.edu
Date: Wed, 23 Jan 2002 10:28:57 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2726


If I connect to ftp.slac and go to /doc and do an ls, I do not see
anything. I only see stuff if I see a dir. Also my program when returns
LIST gives only

dr-xr-xr-x   2 ftp      ftp          2048 Jun 19  1995 .

However, the test case fails because it expects us to go inside forms.

Could someone tell me hos do I avoid this?

Thanks,
Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Got the following error message.
Date: Wed, 23 Jan 2002 10:37:00 -0800
Lines: 78
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:2727

Hi,

Contrary to what I wrote some days ago, the minor error in the submit 
script concerning not being able to open the log file has not been 
solved. As I mentioned before this is a minor error, and it does not 
prevent you from submitting your files.

Pablo

Arvind Bhat wrote:

> Hi,
> I jsut submitted and ran into this error message.
> Hope i am ok and does not need to resubmit. Just
> wanted to convey because earlier mail said this
> problem was fixed ?
> 
> Thanks, Arvind
> 
> -------------------------
> Submission succeeded.
> Error: log: open: Permission denied
> 
> A problem statement has been e-mailed to the system
> administrator 
> 
> 
> 
> On Tue, 22 Jan 2002, Russell Greene wrote:
> 
> 
>>(This was taken from an earlier posting)
>>
>>Hi,
>>
>>If you saw an error like this, the error is minor and it did not prevent
>>you from submitting your code, in other words, we received your files
>>correctly. This problem has been solved, so you will not see it again,
>>if you submit now.
>>
>>Pablo
>>
>>Russell Greene wrote:
>>
>>
>>>Has anyone received the error when running the submit script?
>>>
>>>Submission succeeded
>>>Error: log: open: Permission denied
>>>
>>>
>>>--Russ
>>>
>>>
>>>
>>>
>>
>>"Shankar Agarwal"  wrote in message

>>
>>>Hi,
>>>I got the follwoing error message when submitting the hw1. Does this mean
>>>that submisssion was successful or i need to submit again.
>>>Submission succeeded.
>>>Error: log: open: Permission denied
>>>
>>>A problem statement has been e-mailed to the system
>>>administrator 
>>>
>>>Thanks
>>>Regards
>>>Shankar
>>>
>>>
>>
>>
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links and File Extensions
Date: Wed, 23 Jan 2002 11:05:10 -0800
Lines: 79
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2728

Any reasonable way to handle symbolic links is acceptable. As Pablo said
earlier, only very simple symbolic links will be tested, like the one in
the testing web page. Best, Xin



On Tue, 22 Jan 2002, Dhawal Kumar wrote:

> Excerpt from a previous posting:
> 
> <snip>
> Correct.
> 
> 
> On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:
> 
> > Well, we download 'regular files', and recurse into subdirectories.  And
> > you seem to be saying to do both here.  So just to confirm:
> >
> > * If a symlink is pointing to a regular file, download the file, and
> > name it with the name of the symlink.
> 
> </snip>
> 
> This means if we are interested in txt files and we download link.lnk
> (as you suggest) the name on the local dir should be link.lnk whose
> extension is not txt. Is this acceptable? I thought that all files on
> local dir should have the ext specified.
> 
> Dhawal Kumar
> 
> 
> 
> On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
> 
> > Pete,
> > a symbolic link to a file of interest (because of the file's extention) should
> > always be followed and treated as a local file, i.e. should be copied. The idea
> > of the ftpclient
> > is to make local copies of particular files of interest. Symbolic links aren't
> > particularly interesting files since copying them locally may create dangling
> > references.
> >
> > So, the answer is yes to the first, no to the second!
> >
> > Regards,
> > Antonios.
> >
> >
> > Pete Belknap wrote:
> >
> > > What should we do when the symbolic link has a diff extension than the file
> > > it points to?
> > >
> > > So, for example, if we want to download .txt files and we have the link
> > >
> > > link.lnk  ->  readme.txt
> > >
> > > Should we download it?  Of course, the opposite question is, if we are
> > > interested in files of extension .lnk, should we download the link?  It
> > > seems to me that the answer to my first question should be no and the second
> > > should be yes...
> > >
> > > thanks,
> > > Pete
> >
> >
> 
> 


-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Other files in directory.
Date: Wed, 23 Jan 2002 11:31:19 -0800
Lines: 29
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2729

You overwrite any files which already exists in the local directory, and
leave other files in the local directory alone. Xin


On Tue, 22 Jan 2002, Timothy Tay Chao wrote:

> > Your
> > code should make sure no extra files exist in the new, local tree you're
> > trying to create.
> 
> Does that mean we delete all files in the local directory - essentially
> doing a rm -rf localdir/* every time we run ftpcopy? My interpretation
> of "overwrite any files which already exists" was that, if we want to
> download hello.gz, and we already have a copy of it in localdir, then we
> overwrite our local copy (leaving the "extra" files alone). Is my
> interpretation incorrect?
> 
> Thanks, 
> Tim
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Jonathan White Keljo 
Newsgroups: su.class.cs244a
Subject: Re: Empty submit directory?
Date: 23 Jan 2002 20:05:46 GMT
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth1.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2730

Yes. The script actually hides your files under 
cs244a/submissions/grading/<ta name>/hw1

Jonathan

Derrick Wen-Shiuan Tong  wrote:
: I've submitted my program (twice), but I noticed that my folder directory
: in submissions/hw1/ta/username/ is empty.

: Is this normal?
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Homework 2.
Date: Wed, 23 Jan 2002 12:05:21 -0800
Lines: 34
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2731

Hi Shankar,

> I have few quesitons regarding the howework 2.
> a) First of all in the question 6 it is said that traceroute to
> www.berkeley.com and www.cisco.com takes separate separate path out of
> campus but i don't see that happening. Is there any site in particular
> that i should try rather then doing hit and trial.

I just tested it and for me it clearly takes two different paths. On which 
host in Stanford did you run traceroute? Also what do you mean with site? 
The hosts are www.cisco.com and www.berkeley.edu themselves.

> b) while trying to get the host name for some of the routers in the
> question 5 i got
> elaine5:~> host C2-QANH-GSR-QSV-GSR.ATM.calren2.net
> C2-QANH-GSR-QSV-GSR.ATM.calren2.net does not exist, try again .
>  Is this ok or is there any other way to find out the ip address of
> the routers.

I guess you got this router from the Stanford to UCLA traceorute. 
Traceroute directly tells you the IP addresses of the routers in its 
ourput, e.g.

traceroute to www.ucla.edu (169.232.33.129): 1-30 hops, 38 byte packets
 1  leland-gateway.Stanford.EDU (172.24.12.1)  1.17 ms  3.18 ms  0.729 ms
 2  smrb-rtr.Stanford.EDU (171.64.1.197)  1.34 ms (ttl=63!)  1.24 ms 
(ttl=63!)  1.21 ms (ttl=63!)
 3  i2-gateway.Stanford.EDU (171.64.1.33)  0.514 ms  0.407 ms  0.447 ms
(...)

You don't have to run host seperately.

  Guido

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: STRU, MODE and QUIT ?
Date: Wed, 23 Jan 2002 12:11:46 -0800
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2732

Ambika,

> The RFC seems to say that file STRUcture and stream transmission MODE are
> the defaults for all ftp servers, even those that do not implement the
> STRU and MODE commands. So is it all right if I do not send these commands
> to the server?

The best solution is actually that you try to send these commands and parse 
the servers response. However as far as I know it is defautl for all 
servers and I think it is ok if you don't send them.

> Also, do we need to send the QUIT command to the server? What purpose does
> this serve?

QUIT is the preferred way of closing down the connection to the ftp server 
once you have downloaded all files.

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: left over from previous read
Date: Wed, 23 Jan 2002 12:21:15 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2733

Priyanka,

from what you write this is pretty hard to debug. Did you check:
- That read() really returns something (e.g. the return value which is the 
number of bytes read is > 0)
- You read into the right buffer (e.g. don't get pointers mixed up)
The server definitely won't send the same data again, this is almost 
definitely a problem of buffering on the client side. I also don't quite 
understand your strategy for terminating (the space after the response).

If all fails, send me the relevant part of your source code by email (don't 
post it :-) and I will have a look at it at 2pm today.

  Guido

> Even though I am making sure that each command that I write and then read,
> I read till th every end, until I get the last line which has a space
> after the response.
> 
> Somehow when I do the next read, the stuff from the previoud read is still
> there, even though I clear up my reply buffer, it seems the server still
> sens the stuff from the previous read. Is there any way to flush the
> server contenets after every read.
> 
> Any help would be gretaly appreciated.
> 
> Thanks,
> Priyanka

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links and File Extensions
Date: Wed, 23 Jan 2002 13:25:01 -0800
Lines: 89
Distribution: su
Message-ID: 
References:  
 
 
NNTP-Posting-Host: elaine16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2734


So is it reasonable to download link.lnk -> readme.txt when we are
supposed to download txt files and name it link.lnk

Dhawal Kumar

On Wed, 23 Jan 2002, Xin Wang wrote:

> Any reasonable way to handle symbolic links is acceptable. As Pablo said
> earlier, only very simple symbolic links will be tested, like the one in
> the testing web page. Best, Xin
>
>
>
> On Tue, 22 Jan 2002, Dhawal Kumar wrote:
>
> > Excerpt from a previous posting:
> >
> > <snip>
> > Correct.
> >
> >
> > On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:
> >
> > > Well, we download 'regular files', and recurse into subdirectories.  And
> > > you seem to be saying to do both here.  So just to confirm:
> > >
> > > * If a symlink is pointing to a regular file, download the file, and
> > > name it with the name of the symlink.
> >
> > </snip>
> >
> > This means if we are interested in txt files and we download link.lnk
> > (as you suggest) the name on the local dir should be link.lnk whose
> > extension is not txt. Is this acceptable? I thought that all files on
> > local dir should have the ext specified.
> >
> > Dhawal Kumar
> >
> >
> >
> > On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
> >
> > > Pete,
> > > a symbolic link to a file of interest (because of the file's extention) should
> > > always be followed and treated as a local file, i.e. should be copied. The idea
> > > of the ftpclient
> > > is to make local copies of particular files of interest. Symbolic links aren't
> > > particularly interesting files since copying them locally may create dangling
> > > references.
> > >
> > > So, the answer is yes to the first, no to the second!
> > >
> > > Regards,
> > > Antonios.
> > >
> > >
> > > Pete Belknap wrote:
> > >
> > > > What should we do when the symbolic link has a diff extension than the file
> > > > it points to?
> > > >
> > > > So, for example, if we want to download .txt files and we have the link
> > > >
> > > > link.lnk  ->  readme.txt
> > > >
> > > > Should we download it?  Of course, the opposite question is, if we are
> > > > interested in files of extension .lnk, should we download the link?  It
> > > > seems to me that the answer to my first question should be no and the second
> > > > should be yes...
> > > >
> > > > thanks,
> > > > Pete
> > >
> > >
> >
> >
>
>
> --
> ---------------------------------
> Xin Wang
>
> Department of Computer Science
> Stanford University
>
> 
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Homework 2 Question 3
Date: Wed, 23 Jan 2002 13:28:05 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2735

Hi all,

     For Question #3, does Figure 2(c) add any value to the question?  I
didn't see it mentioned and just wanted to make sure that I wasn't missing
anything.  Thanks.

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: ftp.slac.stanford.edu
Date: Wed, 23 Jan 2002 13:33:19 -0800
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2736

Priyanka,
 
> If I connect to ftp.slac and go to /doc and do an ls, I do not see
> anything. I only see stuff if I see a dir. Also my program when returns
> LIST gives only
> 
> dr-xr-xr-x   2 ftp      ftp          2048 Jun 19  1995 .

I just tried connecting and I definitely see a direcroty called forms (see 
the details below). What you see is just the first of three lines of the 
reply (there are two additional ones for ".." and "forms").

> However, the test case fails because it expects us to go inside forms.

Make sure that you really parse three lines and that your client knows how 
to deal with the "." and ".." entries that denote the current and parent 
directory.

Does that help?

  Guido

-----

tree1:~/cs244a> ncftp ftp.slac.stanford.edu
NcFTP 3.0.3 (April 15, 2001) by Mike Gleason 
Connecting to 134.79.18.30...
[... tons of text deleted...]
ncftp / > cd /doc
ncftp /doc > dir
dr-xr-xr-x   2 ftp      ftp         2048   Jun 19  1995   .
dr-xr-xr-x   4 ftp      ftp         2048   Dec 14 17:39   ..
dr-xr-xr-x   7 ftp      ftp         2048   Mar 13  1999   forms
ncftp /doc > ls
../      ../     forms/
ncftp /doc >
.

Path: shelby.stanford.edu!nntp.stanford.edu!fable12.Stanford.EDU!pabbeel
From: Pieter Abbeel 
Newsgroups: su.class.cs244a
Subject: when localdir doesnot exist
Date: Wed, 23 Jan 2002 13:55:40 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2737

hey,

the webpage says if the specified localdir doesnot exist, we should exit
with message saying localdir doesnot exist (and not down anything)

i was wondering whether the testscript follows this prescription?

(this could explain why it complains about me not downloading anything in
some cases ...)

thanks,
pieter

(obviously i'd love to test this on the server, but that is taking
forever it seems ... (why doesn't that #trials<5 is implemented as
advertised and as would have guaranteed us some more decent
turnaround ????)


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine42.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Wed, 23 Jan 2002 14:13:09 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine42.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2738


I submitted my assignment at 2:10 pm (Wednesday), I got the same error.

TAs, I hope this is still a false alarm.

-anurag



On Tue, 22 Jan 2002, Pablo Molinero wrote:

> As mentioned in the assignment handout, you have to have two
> purify-related files, .purify and a purify.output, in your homework
> directory before you can submit your code.
>
> Pablo
>
>
> Shankar Agarwal wrote:
>
> > Hi,
> > I am trying to submit hw1 and it has ftpcopy.purify in that directory but
> > it still says that no .purify is present.
> > Thanks
> > Shankar
> >
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic9.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: microsoft server timeout! ??
Date: Wed, 23 Jan 2002 14:22:15 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2739

Hi there,

I just sutmitted a test script. The microsoft server gave NOT OK, and the
output is timeout!... All other servers are ok and this server used to
work fine too.

Any hint?

Thanks!!

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: when localdir doesnot exist
Date: Wed, 23 Jan 2002 14:36:42 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2740

Pieter,

> the webpage says if the specified localdir doesnot exist, we should exit
> with message saying localdir doesnot exist (and not down anything)
> 
> i was wondering whether the testscript follows this prescription?

The current test scripts that you can access do not test this case. To 
receive full credit your program should test whether the local directory 
was created successfully (this is actually not much work, we are talking 
about a two-line if statement here.)

> (this could explain why it complains about me not downloading anything in
> some cases ...)

I would guess this is probably not the reason.

> (obviously i'd love to test this on the server, but that is taking
> forever it seems ... (why doesn't that #trials<5 is implemented as
> advertised and as would have guaranteed us some more decent
> turnaround ????)

What do you mean #trials<5 is not implemented? As far as I know it is (if 
not, tell us).

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: microsoft server timeout! ??
Date: Wed, 23 Jan 2002 14:38:07 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2741

Huan,

I guess I wood need more information to even make an educated guess. What 
exactly is the testing script telling you is wrong?

  Guido

> Hi there,
> 
> I just sutmitted a test script. The microsoft server gave NOT OK, and the
> output is timeout!... All other servers are ok and this server used to
> work fine too.
> 
> Any hint?
> 
> Thanks!!

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!pabbeel
From: Pieter Abbeel 
Newsgroups: su.class.cs244a
Subject: Re: microsoft server timeout! ??
Date: Wed, 23 Jan 2002 15:00:51 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Guido Appenzeller 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2742

i just manually logged into microsoft (at least tried) and it said: max
number of connections reached (...) try again later

pieter


On Wed, 23 Jan 2002, Guido Appenzeller wrote:

> Huan,
>
> I guess I wood need more information to even make an educated guess. What
> exactly is the testing script telling you is wrong?
>
>   Guido
>
> > Hi there,
> >
> > I just sutmitted a test script. The microsoft server gave NOT OK, and the
> > output is timeout!... All other servers are ok and this server used to
> > work fine too.
> >
> > Any hint?
> >
> > Thanks!!
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic13.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: Re: microsoft server timeout! ??
Date: Wed, 23 Jan 2002 15:04:40 -0800
Lines: 43
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Guido Appenzeller 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2743

Thanks Guido!

Attache is the error message.  I just submitted another test and it was
ok.
Thanks!
---------------------------------------------
    * Test D: Connects to server but does not transfer any matching files
in the specified levels [out of 3.5]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie
exit line said: ""
---
Program output was:
TIMEOUT! killed program:
/afs/ir/class/cs244a/submissions/restricted/hw1/xwang00/wjiang//compile.sun/ftpc
opy txt 5 ftp.microsoft.com
/afs/ir/class/cs244a/submissions/restricted/hw1/xwang00/wjiang//compile.sun/down
load/ftp.microsoft.com /peropsys/ie   2199
--------------------------------------------------


On Wed, 23 Jan 2002, Guido Appenzeller wrote:

> Huan,
>
> I guess I wood need more information to even make an educated guess. What
> exactly is the testing script telling you is wrong?
>
>   Guido
>
> > Hi there,
> >
> > I just sutmitted a test script. The microsoft server gave NOT OK, and the
> > output is timeout!... All other servers are ok and this server used to
> > work fine too.
> >
> > Any hint?
> >
> > Thanks!!
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine25.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Error in actual submission
Date: Wed, 23 Jan 2002 15:48:32 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine25.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2744

I get the following while submitting.

....
a ./.purify 0K
a ./defines.h 1K
a ./fndefs.h 1K
a ./ftpparse.h 2K
a ./globals.h 1K

Submission succeeded.
Error: log: open: Permission denied

A problem statement has been e-mailed to the system
administrator 

How do I know if my assignment been submitted? Should I try another time?
I would appreciate a quick response - otherwise I would miss the deadline

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!hershd
From: Hersh Satyanarayan Dangayach 
Newsgroups: su.class.cs244a
Subject: Re: Error in actual submission
Date: Wed, 23 Jan 2002 16:53:43 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2745

I got the same response.

- Hersh



On Wed, 23 Jan 2002, Dhawal Kumar wrote:

> I get the following while submitting.
>
> ...
> a ./.purify 0K
> a ./defines.h 1K
> a ./fndefs.h 1K
> a ./ftpparse.h 2K
> a ./globals.h 1K
>
> Submission succeeded.
> Error: log: open: Permission denied
>
> A problem statement has been e-mailed to the system
> administrator 
>
> How do I know if my assignment been submitted? Should I try another time?
> I would appreciate a quick response - otherwise I would miss the deadline
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Error in actual submission
Date: Wed, 23 Jan 2002 16:56:59 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2746

Dhawal,

it looks like it is only a log file problem so you should be 
fine.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 23 Jan 2002, Dhawal Kumar wrote:

> I get the following while submitting.
> 
> ...
> a ./.purify 0K
> a ./defines.h 1K
> a ./fndefs.h 1K
> a ./ftpparse.h 2K
> a ./globals.h 1K
> 
> Submission succeeded.
> Error: log: open: Permission denied
> 
> A problem statement has been e-mailed to the system
> administrator 
> 
> How do I know if my assignment been submitted? Should I try another time?
> I would appreciate a quick response - otherwise I would miss the deadline
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Submit Fails
Date: Wed, 23 Jan 2002 17:04:34 -0800
Lines: 59
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Anurag Gupta 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2747

Anurag,

your submission from 14:04 today is ok (see below).

  Guido

---

somewhere> tar -tzf */hw1/anuragg/*
../Makefile
../README
../ftpparse.c
../list.c
../file.c
../ftpcopy.c
../purify.output
../.purify
../ftpparse.h
../ftpcopy.h

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 23 Jan 2002, Anurag Gupta wrote:

> 
> I submitted my assignment at 2:10 pm (Wednesday), I got the same error.
> 
> TAs, I hope this is still a false alarm.
> 
> -anurag
> 
> 
> 
> On Tue, 22 Jan 2002, Pablo Molinero wrote:
> 
> > As mentioned in the assignment handout, you have to have two
> > purify-related files, .purify and a purify.output, in your homework
> > directory before you can submit your code.
> >
> > Pablo
> >
> >
> > Shankar Agarwal wrote:
> >
> > > Hi,
> > > I am trying to submit hw1 and it has ftpcopy.purify in that directory but
> > > it still says that no .purify is present.
> > > Thanks
> > > Shankar
> > >
> > >
> >
> >
> >
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Symbolic Links and File Extensions
Date: Wed, 23 Jan 2002 17:06:32 -0800
Lines: 107
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2748

Dhawal,

well, I'd argue it is not very reasonable but for this assignment we'll
accept it as correct as file name handling is not the main focus of the
exercise.

Cheers,

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 23 Jan 2002, Dhawal Kumar wrote:

> 
> So is it reasonable to download link.lnk -> readme.txt when we are
> supposed to download txt files and name it link.lnk
> 
> Dhawal Kumar
> 
> On Wed, 23 Jan 2002, Xin Wang wrote:
> 
> > Any reasonable way to handle symbolic links is acceptable. As Pablo said
> > earlier, only very simple symbolic links will be tested, like the one in
> > the testing web page. Best, Xin
> >
> >
> >
> > On Tue, 22 Jan 2002, Dhawal Kumar wrote:
> >
> > > Excerpt from a previous posting:
> > >
> > > <snip>
> > > Correct.
> > >
> > >
> > > On Tue, 15 Jan 2002, Sandeep Tamhankar wrote:
> > >
> > > > Well, we download 'regular files', and recurse into subdirectories.  And
> > > > you seem to be saying to do both here.  So just to confirm:
> > > >
> > > > * If a symlink is pointing to a regular file, download the file, and
> > > > name it with the name of the symlink.
> > >
> > > </snip>
> > >
> > > This means if we are interested in txt files and we download link.lnk
> > > (as you suggest) the name on the local dir should be link.lnk whose
> > > extension is not txt. Is this acceptable? I thought that all files on
> > > local dir should have the ext specified.
> > >
> > > Dhawal Kumar
> > >
> > >
> > >
> > > On Tue, 22 Jan 2002, SUL\AIR Laneroom Cluster wrote:
> > >
> > > > Pete,
> > > > a symbolic link to a file of interest (because of the file's extention) should
> > > > always be followed and treated as a local file, i.e. should be copied. The idea
> > > > of the ftpclient
> > > > is to make local copies of particular files of interest. Symbolic links aren't
> > > > particularly interesting files since copying them locally may create dangling
> > > > references.
> > > >
> > > > So, the answer is yes to the first, no to the second!
> > > >
> > > > Regards,
> > > > Antonios.
> > > >
> > > >
> > > > Pete Belknap wrote:
> > > >
> > > > > What should we do when the symbolic link has a diff extension than the file
> > > > > it points to?
> > > > >
> > > > > So, for example, if we want to download .txt files and we have the link
> > > > >
> > > > > link.lnk  ->  readme.txt
> > > > >
> > > > > Should we download it?  Of course, the opposite question is, if we are
> > > > > interested in files of extension .lnk, should we download the link?  It
> > > > > seems to me that the answer to my first question should be no and the second
> > > > > should be yes...
> > > > >
> > > > > thanks,
> > > > > Pete
> > > >
> > > >
> > >
> > >
> >
> >
> > --
> > ---------------------------------
> > Xin Wang
> >
> > Department of Computer Science
> > Stanford University
> >
> > 
> >
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: PA2: STCP packet size vs. data size?
Date: Wed, 23 Jan 2002 17:10:04 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1011834605 26668 128.12.133.48 (24 Jan 2002 01:10:05 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2749

The handout on page 4 says the STCP packet size should be 536 bytes, but the
table on page 5 implies that this is the MSS (size of the data, not
including the header).

Are STCP packets 536 bytes of data + the header, or 536 bytes total?

Thanks,

Jonathan


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Problem Set 2, Problem 3
Date: Wed, 23 Jan 2002 19:23:35 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2750


Hello,

  I am wondering if problem 3 is missing a section, or if
figure 2(c) is extraneous. Thanks.

                ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Project #2
Date: Wed, 23 Jan 2002 20:02:01 -0800
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2751

Hi,

  I think I am missing something in trying to understand project #2.
So if I am correct, we are writing a transport layer, over the given
network layer, so that client and server compile and run (assuming reliability
and in order delivery for the first part).
  I assume then that client/server will be using the tansport layer
which will use the network layer under the covers.  So we will have to provide
our own versions of connect, read, write, etc. which hides all the newfangled
work of syns/acks/fins/ etc.
    It seems that we then, have to write our own transport versions of
myconnect(..), myaccept() etc. so that the client will work without changing
the source, however, mysocket.c defines the symbol myconnect(..) with global
linkage so we are going to have a problem if we define a duplicate symbol.
    - Are we supposed to create new method names?
    - Are the new versions of myconnect(..) myaccept(..) etc. supposed to live in
      transport.c?
If some could please clear this up, I would forever be indebted. Thanks.

                            ~~m




.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Jonathan White Keljo 
Newsgroups: su.class.cs244a
Subject: Re: Project #2
Date: 24 Jan 2002 04:08:01 GMT
Lines: 31
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine11.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2752

myconnect(), myaccept(), etc. are already written for us and are in 
mysocket.c. We're supposed to do all of the work to make transport.c 
actually work.

Jonathan

Martin Casado  wrote:
: Hi,

:   I think I am missing something in trying to understand project #2.
: So if I am correct, we are writing a transport layer, over the given
: network layer, so that client and server compile and run (assuming reliability
: and in order delivery for the first part).
:   I assume then that client/server will be using the tansport layer
: which will use the network layer under the covers.  So we will have to provide
: our own versions of connect, read, write, etc. which hides all the newfangled
: work of syns/acks/fins/ etc.
:     It seems that we then, have to write our own transport versions of
: myconnect(..), myaccept() etc. so that the client will work without changing
: the source, however, mysocket.c defines the symbol myconnect(..) with global
: linkage so we are going to have a problem if we define a duplicate symbol.
:     - Are we supposed to create new method names?
:     - Are the new versions of myconnect(..) myaccept(..) etc. supposed to live in
:       transport.c?
: If some could please clear this up, I would forever be indebted. Thanks.

:                             ~~m




.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Project #2
Date: Wed, 23 Jan 2002 20:44:27 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2753


Ahh I see, mysock.c uses the transport layer which is where our code
is supposed to go. Got it :-) Thanks a bunch!

                ~~m


> myconnect(), myaccept(), etc. are already written for us and are in
> mysocket.c. We're supposed to do all of the work to make transport.c
> actually work.
>
> Jonathan
>
> Martin Casado  wrote:
> : Hi,
>
> :   I think I am missing something in trying to understand project #2.
> : So if I am correct, we are writing a transport layer, over the given
> : network layer, so that client and server compile and run (assuming reliability
> : and in order delivery for the first part).
> :   I assume then that client/server will be using the tansport layer
> : which will use the network layer under the covers.  So we will have to provide
> : our own versions of connect, read, write, etc. which hides all the newfangled
> : work of syns/acks/fins/ etc.
> :     It seems that we then, have to write our own transport versions of
> : myconnect(..), myaccept() etc. so that the client will work without changing
> : the source, however, mysocket.c defines the symbol myconnect(..) with global
> : linkage so we are going to have a problem if we define a duplicate symbol.
> :     - Are we supposed to create new method names?
> :     - Are the new versions of myconnect(..) myaccept(..) etc. supposed to live in
> :       transport.c?
> : If some could please clear this up, I would forever be indebted. Thanks.
>
> :                             ~~m
>
>
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: microsoft server timeout! ??
Date: Wed, 23 Jan 2002 23:46:42 -0800
Lines: 62
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1011858402 29645 128.12.133.48 (24 Jan 2002 07:46:42 GMT)
X-Complaints-To: 
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2754

This appears to be a result of the testing script deciding that your program
has been running for too long and killing your program. (At least that's my
guess since nowhere in my source code do you see the words "TIMEOUT! killed
program" and the ftp.microsoft.com test runs fine when I do it myself.) This
error popped up all of a sudden for me earlier this week...seems like the
TAs tweaked the timeout on the grading script a little too low.

Jonathan

"Huan Wanda Jiang"  wrote in message

> Thanks Guido!
>
> Attache is the error message.  I just submitted another test and it was
> ok.
> Thanks!
> ---------------------------------------------
>     * Test D: Connects to server but does not transfer any matching files
> in the specified levels [out of 3.5]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie
> exit line said: ""
> ---
> Program output was:
> TIMEOUT! killed program:
>
/afs/ir/class/cs244a/submissions/restricted/hw1/xwang00/wjiang//compile.sun/
ftpc
> opy txt 5 ftp.microsoft.com
>
/afs/ir/class/cs244a/submissions/restricted/hw1/xwang00/wjiang//compile.sun/
down
> load/ftp.microsoft.com /peropsys/ie   2199
> --------------------------------------------------
>
>
> On Wed, 23 Jan 2002, Guido Appenzeller wrote:
>
> > Huan,
> >
> > I guess I wood need more information to even make an educated guess.
What
> > exactly is the testing script telling you is wrong?
> >
> >   Guido
> >
> > > Hi there,
> > >
> > > I just sutmitted a test script. The microsoft server gave NOT OK, and
the
> > > output is timeout!... All other servers are ok and this server used to
> > > work fine too.
> > >
> > > Any hint?
> > >
> > > Thanks!!
> >
> >
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Grading inormation - Problem Set #1
Date: Thu, 24 Jan 2002 11:36:12 -0800
Lines: 44
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2755

Hi everyone,

below you find some more detailed information how the first home work
assignement is graded. It also contains information on what to do
if you have questions about the grading.

  Guido

------------------------------------------------------------------

** Grading System

For each problem or sometimes part of a problem you should find one
of four different marks next to the problem:

Check+ - Everything correct, no significant mistakes
Check  - Small mistakes but overall on the right path
Minus  - Major mistakes, overall direction still seems right
Zero   - Wrong approach or no answer given

These marks correspond to 100%, 66%, 33% and 0% of the points you
can get for the problem (e.g. a check in a problem worth 4.5 points
adds 3 points to your total). For some problems but not all of them
points where rounded up after adding the points of the subproblems.

** Who to contact with questions

If you have questions about the grading please contact the TA who
graded the particular problem directly. The best way to do this is
during office hours, however email works as well.

 Problem 1,6,8  Arun
 Problem 4,5    Antonios
 Problem 2,3    Xin
 Problem 7      Guido
 Problem 9      Matt

** Picking up your Home Work

If you did not pick up you homework in class you can find it in the
drawers on the 3d floor of Gates Building (turn left from the
elevators and look for white filing cabinets).


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: microsoft server timeout! ??
Date: Thu, 24 Jan 2002 11:40:21 -0800
Lines: 82
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2756

Jonathan,

the Microsoft web server was actually pretty swamped yesterday
afternoon with connections sometimes failing. For final grading
we will make sure this is not the case.

If your client is extremely inefficient and thus slow that 
would be a differet matter. But again, yesterday afternoon
is not a good test case.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 23 Jan 2002, Jonathan Keljo wrote:

> This appears to be a result of the testing script deciding that your program
> has been running for too long and killing your program. (At least that's my
> guess since nowhere in my source code do you see the words "TIMEOUT! killed
> program" and the ftp.microsoft.com test runs fine when I do it myself.) This
> error popped up all of a sudden for me earlier this week...seems like the
> TAs tweaked the timeout on the grading script a little too low.
> 
> Jonathan
> 
> "Huan Wanda Jiang"  wrote in message
> 
> > Thanks Guido!
> >
> > Attache is the error message.  I just submitted another test and it was
> > ok.
> > Thanks!
> > ---------------------------------------------
> >     * Test D: Connects to server but does not transfer any matching files
> > in the specified levels [out of 3.5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy txt 5 ftp.microsoft.com localdir /peropsys/ie
> > exit line said: ""
> > ---
> > Program output was:
> > TIMEOUT! killed program:
> >
> /afs/ir/class/cs244a/submissions/restricted/hw1/xwang00/wjiang//compile.sun/
> ftpc
> > opy txt 5 ftp.microsoft.com
> >
> /afs/ir/class/cs244a/submissions/restricted/hw1/xwang00/wjiang//compile.sun/
> down
> > load/ftp.microsoft.com /peropsys/ie   2199
> > --------------------------------------------------
> >
> >
> > On Wed, 23 Jan 2002, Guido Appenzeller wrote:
> >
> > > Huan,
> > >
> > > I guess I wood need more information to even make an educated guess.
> What
> > > exactly is the testing script telling you is wrong?
> > >
> > >   Guido
> > >
> > > > Hi there,
> > > >
> > > > I just sutmitted a test script. The microsoft server gave NOT OK, and
> the
> > > > output is timeout!... All other servers are ok and this server used to
> > > > work fine too.
> > > >
> > > > Any hint?
> > > >
> > > > Thanks!!
> > >
> > >
> >
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: hw2 question 5g)
Date: Thu, 24 Jan 2002 18:50:15 -0800
Lines: 35
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2757

My traceroutes on question 5g) seem to be timing out when I tell it to start
from the router at UT.  It works fine when I do a traceroute from elaine5,
though.  Anybody know what's up?  The same thing was happening to me last
night.

Here is my output, with the special parts of traceroute that are part of the
answer left out:

elaine5:~> traceroute <options excluded> eecs.berkeley.edu
traceroute to eecs.berkeley.edu (169.229.60.28) from 128.83.37.21: 1-30
hops, 38 byte packets
 1  *  *  *
 2  *  *  *
 3  *  *  *
 4  *  *  *
 5  *  *  *
 6  *  *  *
 7  *  *  *
 8  *  *  *
 9  *  *  *
10  *  *  *
11  *  *  *
12  *  *  *
13  *  *  *
14  *  *  *
15  *  *  *
16  *  *  *
17  *  *  *
18  *  *  *
19  *  *  *

Thanks!
Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Re: hw2 question 5g)
Date: Thu, 24 Jan 2002 21:48:34 -0800
Organization: Stanford University
Lines: 42
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1011937660 9168 171.66.211.98 (25 Jan 2002 05:47:40 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2758

I just tried it (with the correct options)
and it worked fine for me.

- Nick



Peter Belknap wrote:
> 
> My traceroutes on question 5g) seem to be timing out when I tell it to start
> from the router at UT.  It works fine when I do a traceroute from elaine5,
> though.  Anybody know what's up?  The same thing was happening to me last
> night.
> 
> Here is my output, with the special parts of traceroute that are part of the
> answer left out:
> 
> elaine5:~> traceroute <options excluded> eecs.berkeley.edu
> traceroute to eecs.berkeley.edu (169.229.60.28) from 128.83.37.21: 1-30
> hops, 38 byte packets
>  1  *  *  *
>  2  *  *  *
>  3  *  *  *
>  4  *  *  *
>  5  *  *  *
>  6  *  *  *
>  7  *  *  *
>  8  *  *  *
>  9  *  *  *
> 10  *  *  *
> 11  *  *  *
> 12  *  *  *
> 13  *  *  *
> 14  *  *  *
> 15  *  *  *
> 16  *  *  *
> 17  *  *  *
> 18  *  *  *
> 19  *  *  *
> 
> Thanks!
> Pete
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Nick McKeown 
Newsgroups: su.class.cs244a
Subject: Problem Set 2: Missing Question 2(c)
Date: Thu, 24 Jan 2002 21:57:05 -0800
Organization: Stanford University
Lines: 25
Distribution: su
Message-ID: 
NNTP-Posting-Host: mckeown-pbdsl1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1011938172 9237 171.66.211.98 (25 Jan 2002 05:56:12 GMT)
X-Complaints-To: 
To: 
X-Mailer: Mozilla 4.61 [en] (Win98; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2759


Someone pointed out that part (c) is missing
for Question #2. Part (c) asked what happens
when the link breaks (as shown in the figure).
Not sure where it went, but it seems to have
fallen off the problem set! 

My error, and so you don't have to do part (c). 
Just ignore the redundant Figure 2(c).

--------

While I have your attention, let me make a couple
of announcements that I forgot to make in class today:

1. Tomorrow (Friday) Arun will lead the discussion
   section. The topic will be Programming Assignment #2.
   I strongly encourage you to attend or watch on video.

2. You should read Sections 5.1, 5.2, 5.4 of Peterson 
   and Davie before class on Tuesday.

Have a good weekend,

- Nick
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: PA2: STCP packet size vs. data size?
Date: 25 Jan 2002 06:48:14 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine24.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2760

The payload is 536 bytes; the header is excluded from this byte count.

Arun

Jonathan Keljo  wrote:
: The handout on page 4 says the STCP packet size should be 536 bytes, but the
: table on page 5 implies that this is the MSS (size of the data, not
: including the header).

: Are STCP packets 536 bytes of data + the header, or 536 bytes total?

: Thanks,

: Jonathan


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: Re: hw2 question 5g)
Date: Thu, 24 Jan 2002 23:03:29 -0800
Lines: 54
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2761

Nick is right.  My options were wrong.

Pete

"Nick McKeown"  wrote in message

> I just tried it (with the correct options)
> and it worked fine for me.
>
> - Nick
>
>
>
> Peter Belknap wrote:
> >
> > My traceroutes on question 5g) seem to be timing out when I tell it to
start
> > from the router at UT.  It works fine when I do a traceroute from
elaine5,
> > though.  Anybody know what's up?  The same thing was happening to me
last
> > night.
> >
> > Here is my output, with the special parts of traceroute that are part of
the
> > answer left out:
> >
> > elaine5:~> traceroute <options excluded> eecs.berkeley.edu
> > traceroute to eecs.berkeley.edu (169.229.60.28) from 128.83.37.21: 1-30
> > hops, 38 byte packets
> >  1  *  *  *
> >  2  *  *  *
> >  3  *  *  *
> >  4  *  *  *
> >  5  *  *  *
> >  6  *  *  *
> >  7  *  *  *
> >  8  *  *  *
> >  9  *  *  *
> > 10  *  *  *
> > 11  *  *  *
> > 12  *  *  *
> > 13  *  *  *
> > 14  *  *  *
> > 15  *  *  *
> > 16  *  *  *
> > 17  *  *  *
> > 18  *  *  *
> > 19  *  *  *
> >
> > Thanks!
> > Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Romain Jean Thibaux 
Newsgroups: su.class.cs244a
Subject: register a callback
Date: Fri, 25 Jan 2002 18:12:16 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic9.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2762

What does a "callback" mean ?
What does it mean to have it "registered" ?

	Romain
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: register a callback
Date: 26 Jan 2002 06:17:06 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga6.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2763

In the current model, this really does not mean much. This is just a flag 
(so you "register" it by setting it equal to 1) that you can use to decide 
if you want to invoke transport_sock_io or transport_appl_io.

Arun

 Romain Jean Thibaux  wrote:
: What does a "callback" mean ?
: What does it mean to have it "registered" ?

: 	Romain
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine2.Stanford.EDU!pabbeel
From: Pieter Abbeel 
Newsgroups: su.class.cs244a
Subject: mapping path between 2 hosts
Date: Sat, 26 Jan 2002 11:56:43 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2764

hi,

how does one map the path between 2 hosts like done in handout #4, slide
25?

thanks,
pieter
(i mean more detail than just play around with
ifconfig,traceroute,host,..)

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: EFAULT errno 14.
Date: Sat, 26 Jan 2002 13:48:43 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2765

Hi,
While sending packet to the server. After the 2 way handshake is completed
i get error when i do network_send for the first data packet. It says
errno 14 which is EFAULT. I am not sure why i am getting this problem. As
i was successful in sending the first packet on the udp socket and i have
not closed the socket in my child. So is there something i am missing. If
someone can suggest something then it will be of great help to me.Right
now its almost like black hole to me....
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: mapping path between 2 hosts
Date: 26 Jan 2002 23:35:21 GMT
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine7.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2766

Pieter Abbeel  writes:

>hi,

>how does one map the path between 2 hosts like done in handout #4, slide
>25?

>thanks,
>pieter
>(i mean more detail than just play around with
>ifconfig,traceroute,host,..)


It might be helpful to rewatch the lecture (I think it was from a week
last Thursday) and work through an example from whichever system you're on.
The basic idea is to use traceroute to find the routers along the path,
and host to determine which interfaces on each router could have been used
for your path.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: EFAULT errno 14.
Date: 26 Jan 2002 23:41:29 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine7.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2767

Shankar Agarwal  writes:

>Hi,
>While sending packet to the server. After the 2 way handshake is completed
>i get error when i do network_send for the first data packet. It says
>errno 14 which is EFAULT. I am not sure why i am getting this problem. As
>i was successful in sending the first packet on the udp socket and i have
>not closed the socket in my child. So is there something i am missing. If
>someone can suggest something then it will be of great help to me.Right
>now its almost like black hole to me....

This could happen if you pass in a bad address (e.g. a null pointer)
or an invalid buffer length.  Without any more information, it's difficult
to say what's going wrong.  I'd suggest that you run in gdb to debug a
problem like this.

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!araik
From: Araik Grigoryan 
Newsgroups: su.class.cs244a
Subject: attaching to process in ddd
Date: Sat, 26 Jan 2002 15:42:17 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2768


Hi,

Does anyone know how to correctly attach to the child process in DDD? I've
tried it a few times and DDD hangs every time. Thanks.

araik

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!htimam
From:  (Hasan Taufiq Imam)
Newsgroups: su.class.cs244a
Subject: Callback functions
Date: 26 Jan 2002 23:42:56 GMT
Organization: Stanford University, CA 94305, USA
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine8.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2769

Does callback feature have any use for us for the project? I am really 
confused. One place it says that appropriate function will get called 
depending on IO on peer socket or internal socket. On the other hand, 
we can check which sd is busy and call the functions. So are these two options
and we are supposed to choose one or the either. Or do we need to use both?
If we need to use both, why? Variables in CTX are set to one to register 
callback feature. Is that enough or we not do anything extra to activate 
callback? What else do we need to do? If we just need to use one procedure then
does one have any advantage over other. 
  
Hasan
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Callback functions
Date: 26 Jan 2002 23:59:31 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine7.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2770

 (Hasan Taufiq Imam) writes:

>Does callback feature have any use for us for the project? I am really 
>confused. One place it says that appropriate function will get called 
>depending on IO on peer socket or internal socket. On the other hand, 
>we can check which sd is busy and call the functions. So are these two options
>and we are supposed to choose one or the either. Or do we need to use both?
>If we need to use both, why? Variables in CTX are set to one to register 
>callback feature. Is that enough or we not do anything extra to activate 
>callback? What else do we need to do? If we just need to use one procedure then
>does one have any advantage over other. 
>  
>Hasan

These options are one and the same; i.e. you just have to check which socket
is busy, and call the appropriate callback function.

You needn't get hung up on the term "callback"; all this means here is that
you should have functions to process data from the peer and to process data
from the application, and that you should put the code to invoke the
appropriate one of these in control_loop().  If you look through transport.c,
you'll see that it's much simpler than it perhaps sounds at first.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Callback functions
Date: 27 Jan 2002 00:07:27 GMT
Lines: 28
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine7.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2771

 (Matthew Jonathan Holliman) writes:

 (Hasan Taufiq Imam) writes:

>>Does callback feature have any use for us for the project? I am really 
>>confused. One place it says that appropriate function will get called 
>>depending on IO on peer socket or internal socket. On the other hand, 
>>we can check which sd is busy and call the functions. So are these two options
>>and we are supposed to choose one or the either. Or do we need to use both?
>>If we need to use both, why? Variables in CTX are set to one to register 
>>callback feature. Is that enough or we not do anything extra to activate 
>>callback? What else do we need to do? If we just need to use one procedure then
>>does one have any advantage over other. 
>>  
>>Hasan

>These options are one and the same; i.e. you just have to check which socket
>is busy, and call the appropriate callback function.

>You needn't get hung up on the term "callback"; all this means here is that
>you should have functions to process data from the peer and to process data
>from the application, and that you should put the code to invoke the
>appropriate one of these in control_loop().  If you look through transport.c,
>you'll see that it's much simpler than it perhaps sounds at first.

BTW, just to follow on--there are a couple of FAQ entries about callbacks,
and about which parts of transport.c you can change or need to use.  It might
be worthwhile to look through it to clarify any questions.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine5.Stanford.EDU!htimam
From:  (Hasan Taufiq Imam)
Newsgroups: su.class.cs244a
Subject: Peer address.
Date: 27 Jan 2002 03:53:09 GMT
Organization: Stanford University, CA 94305, USA
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine5.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2772

Who initializes peer address for network.c? Do I need to specify ports in 
tcp header or it does not matter? I could not find how fromaddr in global_net
_ctx got initialized? Would I be write if I said only variables in tcp_header
we need to worry about are th_seq, th_ack, th_flags. 
Hasan.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Romain Thibaux 
Newsgroups: su.class.cs244a
Subject: receiving and sending
Date: Sat, 26 Jan 2002 21:00:36 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic9.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2773

(what a broad title...)

The presence of callback functions only for incoming data seems to mean
that each time we receive data we do all we can to send it right away.
But we might block on write, which is bad because some data might still
be coming, isn't that right ? I assume I don't know if writes or reads
are fast or how long they block.

Also, if the writes are only done when we read something, ie inside the
callback function, then we'll have to handle packet retransmission
separately, which means the windows cannot be local to the callback
functions. I don't like global variables, but if you tell me that's the
way...

	Romain
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: receiving and sending
Date: Sat, 26 Jan 2002 22:48:36 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2774

On Sat, 26 Jan 2002, Romain Thibaux wrote:

> (what a broad title...)
>
> The presence of callback functions only for incoming data seems to mean
> that each time we receive data we do all we can to send it right away.
> But we might block on write, which is bad because some data might still
> be coming, isn't that right ? I assume I don't know if writes or reads
> are fast or how long they block.

You could always poll the local data file descriptor for writing, and
if it is busy store the packet data in a temporary buffer.
If the temp buffer gets too full, stop acking the sender until
it frees up.

> Also, if the writes are only done when we read something, ie inside the
> callback function, then we'll have to handle packet retransmission
> separately, which means the windows cannot be local to the callback
> functions. I don't like global variables, but if you tell me that's the
> way...

I don't see why this necessarily means you will need globals, you could,
for example, stick the window in the context. Either way, globals aren't
horrendous in this framework since each socket gets a process for the transport
layer to work in.

                        ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Darren Lewis" 
Newsgroups: su.class.cs244a
Subject: Problem Set 2: Question 3a
Date: Sat, 26 Jan 2002 23:36:23 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: darren.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2775

For question 3a, are we only supposed to draw the contents of the routing
tables after they have initially stabilized, or do we have to show the
process of them stabilizing from their very first state?

-Darren


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic5.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: EFAULT errno 14.
Date: Sat, 26 Jan 2002 23:43:53 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Matthew Jonathan Holliman 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2776

The socket is already in the connected state as its the active open from
the client side. The data len is also not zero. I am passing the name of
the file to be downloaded. As we are using sendmsg in the network.c i
don't know have the facility to check the send address. But the socket
descriptor is correct. Also the debugger does not take me below sendmsg
and as the return errno for the sendmsg is 14 i really don't understand
why it is so. Also i was able to exchange the SYN packet using that
socket.


On 26 Jan 2002, Matthew Jonathan Holliman wrote:

> Shankar Agarwal  writes:
>
> >Hi,
> >While sending packet to the server. After the 2 way handshake is completed
> >i get error when i do network_send for the first data packet. It says
> >errno 14 which is EFAULT. I am not sure why i am getting this problem. As
> >i was successful in sending the first packet on the udp socket and i have
> >not closed the socket in my child. So is there something i am missing. If
> >someone can suggest something then it will be of great help to me.Right
> >now its almost like black hole to me....
>
> This could happen if you pass in a bad address (e.g. a null pointer)
> or an invalid buffer length.  Without any more information, it's difficult
> to say what's going wrong.  I'd suggest that you run in gdb to debug a
> problem like this.
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: SYN then FIN
Date: Sat, 26 Jan 2002 23:55:58 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2777


Hi,

 What is the expected behavior if the active socket sends
a SYN and gets a FIN, a FIN-ACK or a data packet in reply.
Should this packet be dropped or should the entire connection
be dropped? (i.e. should the active side close down the socket)
Thanks :-)

            ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic22.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Getting the data on the data_sd[1] in the child.
Date: Sun, 27 Jan 2002 12:23:16 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2778

Hi,
My child process (i.e. tcp layer) is getting the same data back on
data_sd[1] which it is seding to the parent process i.e. the client. As a
result of this the program is in infinite loop.
Can someone please tell me what could be wrong with it.
Regards
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Peer address.
Date: 27 Jan 2002 22:46:20 GMT
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga11.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2779

Those fields get initialized upon the first call to network_recv. They are then available to transmission 
to the parent process over the syn_ds socket.

Arun

Hasan Taufiq Imam  wrote:
: Who initializes peer address for network.c? Do I need to specify ports in 
: tcp header or it does not matter? I could not find how fromaddr in global_net
: _ctx got initialized? Would I be write if I said only variables in tcp_header
: we need to worry about are th_seq, th_ack, th_flags. 
: Hasan.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Problem Set 2: Question 3a
Date: 27 Jan 2002 23:13:13 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga14.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2780

For this question, we only care about the final tables.

Good luck,

Arun

Darren Lewis  wrote:
: For question 3a, are we only supposed to draw the contents of the routing
: tables after they have initially stabilized, or do we have to show the
: process of them stabilizing from their very first state?

: -Darren


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: SYN then FIN
Date: 28 Jan 2002 00:53:33 GMT
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2781


> What is the expected behavior if the active socket sends
>a SYN and gets a FIN, a FIN-ACK or a data packet in reply.
>Should this packet be dropped or should the entire connection
>be dropped? (i.e. should the active side close down the socket)
>Thanks :-)

A reasonable (and simple) approach would be to just ignore that segment,
and wait for the SYN-ACK to turn up.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Getting the data on the data_sd[1] in the child.
Date: 28 Jan 2002 01:03:47 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2782


>My child process (i.e. tcp layer) is getting the same data back on
>data_sd[1] which it is seding to the parent process i.e. the client. As a
>result of this the program is in infinite loop.
>Can someone please tell me what could be wrong with it.

Could you describe what you observed when you stepped through in gdb?

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: PS2, 3(b)
Date: Sun, 27 Jan 2002 21:13:30 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2783

So when the link is added, the tables are not reset?  If this is the case,
those routers connected to the new link update their tables and the new
costs are iteratively sent out as in the normal Bellman-Ford algo?

Thanks for your help,
Bret



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: callbacks
Date: Sun, 27 Jan 2002 21:34:45 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2784

Hey folks,

Where does the whole "callback" thing mentioned
on the FAQ and the previous posts fit in?

Does it simply refer to using select to wait for events,
and then using a switch statement to handle the
event appropriately?

Thanks,
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic17.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Getting the data on the data_sd[1] in the child.
Date: Sun, 27 Jan 2002 21:39:10 -0800
Lines: 20
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic17.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Matthew Jonathan Holliman 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2785

Hi Jonathan,
I was able to figure out both the problems. I was not using nework_send
properly. We should pass a NULL in the end which was leading to the
failure of my send call. Also i was not checking for the return len of 0
for the local_data_sock which was creating the problem of loopback. Which
was not exactly loopback. Thanks a lot.
Shankar
 On 28 Jan 2002, Matthew Jonathan
Holliman wrote:

>
> >My child process (i.e. tcp layer) is getting the same data back on
> >data_sd[1] which it is seding to the parent process i.e. the client. As a
> >result of this the program is in infinite loop.
> >Can someone please tell me what could be wrong with it.
>
> Could you describe what you observed when you stepped through in gdb?
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic17.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding the compilation.
Date: Sun, 27 Jan 2002 22:06:13 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic17.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2786

Hi,
I am not sure whats wrong with the environment where the code submitted
for test is compile. But i am able to compile in my local directory but
when i submitted the code for test it gave me compilation error.
The errors that i got were for the retxQueue. I have define this structure
in transport.h which i submitted with the code. But still it is not able
to see it. Can you please have a look at it.
Thanks
Shankar

 transport.c:109: field
`retxQueue' has
incomplete type transport.c:110: field `recvQueue' has incomplete type
transport.c: In function `transport_appl_io':
transport.c:641: sizeof applied to an incomplete type
transport.c:647: dereferencing pointer to incomplete type
transport.c:648: dereferencing pointer to incomplete type
transport.c:649: dereferencing pointer to incomplete typeA

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Xiaoyan Cheng 
Newsgroups: su.class.cs244a
Subject: two-way handshake?
Date: Sun, 27 Jan 2002 22:21:14 -0800
Lines: 2
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga9.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2787

Is it okay if we just implement 2-way handshake(instead of 3-way
handshake)?
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: on testing
Date: Sun, 27 Jan 2002 22:48:46 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2788

Hi,

  I'm passing all of the tests except for the ones which use
the TA's receiver... and I'm really not sure why. Is it against
the rules/honor code to modify my transport.c to open a file
in a shared directory of mine, and write out a bunch of
debug code during testing? (like all packets sent by
the TA's receiver)
  I'm assuming that this is not allowed.... but it never
hurts to ask (I hope) :-)

                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the compilation.
Date: Mon, 28 Jan 2002 01:16:38 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2789

I guess the submit script doesnt take anything else than transport.c from
your directory. Look at the deliverables for HW2.A

-anurag



On Sun, 27 Jan 2002, Shankar Agarwal wrote:

> Hi,
> I am not sure whats wrong with the environment where the code submitted
> for test is compile. But i am able to compile in my local directory but
> when i submitted the code for test it gave me compilation error.
> The errors that i got were for the retxQueue. I have define this structure
> in transport.h which i submitted with the code. But still it is not able
> to see it. Can you please have a look at it.
> Thanks
> Shankar
>
>  transport.c:109: field
> `retxQueue' has
> incomplete type transport.c:110: field `recvQueue' has incomplete type
> transport.c: In function `transport_appl_io':
> transport.c:641: sizeof applied to an incomplete type
> transport.c:647: dereferencing pointer to incomplete type
> transport.c:648: dereferencing pointer to incomplete type
> transport.c:649: dereferencing pointer to incomplete typeA
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: on testing
Date: Mon, 28 Jan 2002 10:16:37 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2790

I too am passing all of the tests except for the parts which use the TA's
receiver (both parts of Test 1.B).  I have no idea why, as in my own tests
the transmitter appears to work fine.

-Russ

"Martin Casado"  wrote in message

> Hi,
>
>   I'm passing all of the tests except for the ones which use
> the TA's receiver... and I'm really not sure why. Is it against
> the rules/honor code to modify my transport.c to open a file
> in a shared directory of mine, and write out a bunch of
> debug code during testing? (like all packets sent by
> the TA's receiver)
>   I'm assuming that this is not allowed.... but it never
> hurts to ask (I hope) :-)
>
>                     ~~m
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: SIGINT handling?
Date: Mon, 28 Jan 2002 11:14:56 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2791


Child process seems to be calling sigint_handler() on SIGINT, which just
exits with exit(1).
But I wonder why we implement this way of behavior for SIGINT signal...
What kind of interrupts are we expecting here?
( This may be unimpotant for this assignment.. but I'm curioius.. )



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS2, 3(b)
Date: 28 Jan 2002 19:26:53 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine3.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2792

"Bret Taylor"  writes:

>So when the link is added, the tables are not reset?  If this is the case,
>those routers connected to the new link update their tables and the new
>costs are iteratively sent out as in the normal Bellman-Ford algo?

Yep, that's correct.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: callbacks
Date: 28 Jan 2002 19:32:28 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine3.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2793

Luis Robles  writes:

>Hey folks,

>Where does the whole "callback" thing mentioned
>on the FAQ and the previous posts fit in?

>Does it simply refer to using select to wait for events,
>and then using a switch statement to handle the
>event appropriately?

Yes, that's about it.  (Basically, if you ignore the whole callback name
and description, and just think through how you would most naturally
implement the solution--the chances are very good that it will be exactly
what the "callback" was intending!)

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: two-way handshake?
Date: 28 Jan 2002 19:48:14 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine3.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2794

Xiaoyan Cheng  writes:

>Is it okay if we just implement 2-way handshake(instead of 3-way
>handshake)?

I'm not quite sure what you're referring to; the STCP specification in
the assignment explicitly requires two-way SYN handshake rather than
the three-way handshake that TCP uses.  Could you clarify?

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: SIGINT handling?
Date: 28 Jan 2002 19:51:47 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine3.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2795

"BAEHOPIL"  writes:


>Child process seems to be calling sigint_handler() on SIGINT, which just
>exits with exit(1).
>But I wonder why we implement this way of behavior for SIGINT signal...
>What kind of interrupts are we expecting here?
>( This may be unimpotant for this assignment.. but I'm curioius.. )

SIGINT in this case is Ctrl-C.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Regarding the compilation.
Date: 28 Jan 2002 20:22:55 GMT
Lines: 19
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine3.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2796

Anurag Gupta  writes:

>I guess the submit script doesnt take anything else than transport.c from
>your directory. Look at the deliverables for HW2.A

>-anurag

>> I am not sure whats wrong with the environment where the code submitted
>> for test is compile. But i am able to compile in my local directory but
>> when i submitted the code for test it gave me compilation error.
>> The errors that i got were for the retxQueue. I have define this structure
>> in transport.h which i submitted with the code. But still it is not able
>> to see it. Can you please have a look at it.

Keep in mind that transport.h is a symbolic link to the original
version (assuming you used the gethw2 script); you should not need
(and in fact are not permitted) to make modifications to this file.
You should only need to change transport.c for the first two parts
of the assignment.
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding the test scprit.
Date: Mon, 28 Jan 2002 12:44:40 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2797

Hi,
I am seeing a strange problem with the test scpirt. My tests are failing
for the cases when TA's program is supposed to act as a receiver. I don't
see this problem in my testing. Is it possible to get hold of the binary
which does this testing. As the result does not give much information
regarding what went wrong while testing my program. Is it also possible to
get the files which the TA's script is trying to transfer. So that we can
use them to test to figure out what went wrong.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: on testing
Date: Mon, 28 Jan 2002 12:48:22 -0800
Lines: 34
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012250903 14833 128.12.133.48 (28 Jan 2002 20:48:23 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2798

You guys must be violating the standard somehow. Read the handout very
carefully to make sure your behavior is correct. You might also try running
your client against another student's server and vice versa, if you could
manage to do so without ever looking at their code. ;-)

Jonathan

"Russell Greene"  wrote in message

> I too am passing all of the tests except for the parts which use the TA's
> receiver (both parts of Test 1.B).  I have no idea why, as in my own tests
> the transmitter appears to work fine.
>
> -Russ
>
> "Martin Casado"  wrote in message
> 
> > Hi,
> >
> >   I'm passing all of the tests except for the ones which use
> > the TA's receiver... and I'm really not sure why. Is it against
> > the rules/honor code to modify my transport.c to open a file
> > in a shared directory of mine, and write out a bunch of
> > debug code during testing? (like all packets sent by
> > the TA's receiver)
> >   I'm assuming that this is not allowed.... but it never
> > hurts to ask (I hope) :-)
> >
> >                     ~~m
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Xiaoyan Cheng 
Newsgroups: su.class.cs244a
Subject: Re: two-way handshake?
Date: Mon, 28 Jan 2002 14:07:26 -0800
Lines: 12
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: fable2.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2799

Matthew Jonathan Holliman wrote:
> 
> Xiaoyan Cheng  writes:
> 
> >Is it okay if we just implement 2-way handshake(instead of 3-way
> >handshake)?
> 
> I'm not quite sure what you're referring to; the STCP specification in
> the assignment explicitly requires two-way SYN handshake rather than
> the three-way handshake that TCP uses.  Could you clarify?

That's all that I wanted to know, thanks.
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic26.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Can you please clarify what this error means.
Date: Mon, 28 Jan 2002 18:38:21 -0800
Lines: 50
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic26.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2800

Hi,
I am getting the following errors when i submit the script for the part
1B. I am not able to figure out what exactly is it saying. Is it that the
receiver gets more data or is it receiving less data.
Moreover there are 3 figures for each TESTDATA. what each of them means.
Thanks
Regards
Shankar
===============

    * Test 1.B: Transmitter fails to work in reliable mode with our
Receiver
        [out of 10]

[SUMMARY] NOT OK
[RESULT]
===========================
===(T1.Ba) Byte count differences found (INCORRECT!)

---
Difference between expected result and your result:
1,3c1,3
<    2641   15865  110230 TESTDATA.0
<    1786    9963  299328 TESTDATA.1
<     125     536    5098 TESTDATA.2
---
>    2666   15720  110230 TESTDATA.0
>    1812   10049  299328 TESTDATA.1
>     131     501    5098 TESTDATA.2

---

===========================
===(T1.Bb) Checksum differences found (INCORRECT!)

---
Difference between expected result and your result:
1,3c1,3
< 3108646830 110230 TESTDATA.0
< 4231670217 299328 TESTDATA.1
< 2946179251 5098 TESTDATA.2
---
> 1705403052 110230 TESTDATA.0
> 4265638025 299328 TESTDATA.1
> 2181702833 5098 TESTDATA.2

---

===============

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!araik
From: Araik Grigoryan 
Newsgroups: su.class.cs244a
Subject: attaching to process in ddd - instructions
Date: Mon, 28 Jan 2002 20:18:32 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Abhishek Das 
Xref: nntp.stanford.edu su.class.cs244a:2801


Here you go ...

1. Put sleep(30) somewhere in the child process. Compile.
2. Open DDD and open the client executable via the File menu.
3. By clicking on function names, make your way down to the location where
you placed the sleep and insert a break point right after the sleep
statement.
4. Now run the client with all it's arguments on the command line
(outside DDD, of course).
5. Go back to DDD and choose Attach Process from the File menu. Choose the
process of the child and wait until sleep is done.
6. If you see that nothing is happening for a long time, just click Next
or Step once and you will see the pointer.
7. You can do the same for the server.

araik



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the test scprit.
Date: Mon, 28 Jan 2002 20:27:09 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma13.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2802


Shankar,
the testing platform should be familiar to you by
now:) I'm afraid, everyone in the class has to work with
the test scripts given. Last year people didn't even
have this flexibility!

Regards,
Antonios.

Shankar Agarwal wrote:

> Hi,
> I am seeing a strange problem with the test scpirt. My tests are failing
> for the cases when TA's program is supposed to act as a receiver. I don't
> see this problem in my testing. Is it possible to get hold of the binary
> which does this testing. As the result does not give much information
> regarding what went wrong while testing my program. Is it also possible to
> get the files which the TA's script is trying to transfer. So that we can
> use them to test to figure out what went wrong.
> Thanks
> Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: SUL\\AIR Laneroom Cluster 
Newsgroups: su.class.cs244a
Subject: Re: Can you please clarify what this error means.
Date: Mon, 28 Jan 2002 20:32:17 -0800
Lines: 68
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: sul-lr-ozma13.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.76 [en]C-CCK-MCD   (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2803

Shankar,

the numbers in the first test your code fails are
BYTE counts. the top ones are the expected
and the bottom ones are yours. As far as I can
tell, in some cases your code transmits fewer bytes
and in some more.

The other numbers are just checksum numbers
with no particular meaning to you, other than
that your data isn't correctly transmitted (it gets
corrupted somewhere, overwritten, contains garbage etc.)

Regards,
Antonios.

Shankar Agarwal wrote:

> Hi,
> I am getting the following errors when i submit the script for the part
> 1B. I am not able to figure out what exactly is it saying. Is it that the
> receiver gets more data or is it receiving less data.
> Moreover there are 3 figures for each TESTDATA. what each of them means.
> Thanks
> Regards
> Shankar
> ===============
>
>     * Test 1.B: Transmitter fails to work in reliable mode with our
> Receiver
>         [out of 10]
>
> [SUMMARY] NOT OK
> [RESULT]
> ===========================
> ===(T1.Ba) Byte count differences found (INCORRECT!)
>
> ---
> Difference between expected result and your result:
> 1,3c1,3
> <    2641   15865  110230 TESTDATA.0
> <    1786    9963  299328 TESTDATA.1
> <     125     536    5098 TESTDATA.2
> ---
> >    2666   15720  110230 TESTDATA.0
> >    1812   10049  299328 TESTDATA.1
> >     131     501    5098 TESTDATA.2
>
> ---
>
> ===========================
> ===(T1.Bb) Checksum differences found (INCORRECT!)
>
> ---
> Difference between expected result and your result:
> 1,3c1,3
> < 3108646830 110230 TESTDATA.0
> < 4231670217 299328 TESTDATA.1
> < 2946179251 5098 TESTDATA.2
> ---
> > 1705403052 110230 TESTDATA.0
> > 4265638025 299328 TESTDATA.1
> > 2181702833 5098 TESTDATA.2
>
> ---
>
> ===============

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: finserver compile error
Date: Mon, 28 Jan 2002 21:01:52 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2804

Is anyone else getting this compile error?

saga7:~/cs244a/hw2> make all
umask 022
gcc -g -ansi -DSOLARIS  -c client.c
gcc -g -ansi -DSOLARIS  -c transport.c
gcc -g -ansi -DSOLARIS  -c network.c
gcc -g -ansi -DSOLARIS  -c mysock.c
gcc -g -ansi -DSOLARIS  -o client client.o transport.o network.o
mysock.o -lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c server.c
gcc -o server server.o transport.o network.o mysock.o -lsocket -lnsl -lm 
make: *** No rule to make target `finserver.o', needed by `finserver'. 
Stop.
saga7:~/cs244a/hw2> 

I used the gethw2 script, and have not modified
the makefile,

Thanks....
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Sequence number clarifications
Date: Mon, 28 Jan 2002 21:17:03 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2805

Hello,

  I would greatly appreciate it if someone could help me clear up
a few questions I have about sequence numbering...

Assume the active end starts the connection with seqx, and the passive
replies with seqy (acking with seqx+1)

1. The assignment sheet states that the active end records seqy so that it
   will expect the next byte of the passive end to start at seqy+1, can
   we also assume that the passive end expects the active end to start at
   seqx+1? (this isn't stated either way)

2. For acknowledgements, the project sheet says to set the "ack_seq" field to
   the sequence number of the next byte of data the receiver expects.  tcp_hdr
   has no ack_seq field, do we set th_ack or th_seq to this value, or both?

3. Do we have to place/handle a value in the th_seq field of data packets?

Thanks :-)

                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga11.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: finserver compile error
Date: Mon, 28 Jan 2002 21:54:10 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2806

I got the same error.

On Mon, 28 Jan 2002, Luis Robles wrote:

> Is anyone else getting this compile error?
>
> saga7:~/cs244a/hw2> make all
> umask 022
> gcc -g -ansi -DSOLARIS  -c client.c
> gcc -g -ansi -DSOLARIS  -c transport.c
> gcc -g -ansi -DSOLARIS  -c network.c
> gcc -g -ansi -DSOLARIS  -c mysock.c
> gcc -g -ansi -DSOLARIS  -o client client.o transport.o network.o
> mysock.o -lsocket -lnsl -lm
> gcc -g -ansi -DSOLARIS  -c server.c
> gcc -o server server.o transport.o network.o mysock.o -lsocket -lnsl -lm
> make: *** No rule to make target `finserver.o', needed by `finserver'.
> Stop.
> saga7:~/cs244a/hw2>
>
> I used the gethw2 script, and have not modified
> the makefile,
>
> Thanks....
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga11.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: bind and gethostname
Date: Mon, 28 Jan 2002 22:02:12 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2807

hi

I remember from last assignment that bind() followed by getsockname(),
 before calling accept(), didn't give the correctlocal IP address. How
come in this assignment, we call mybind() followed by myaccept(), which
uses getsockname() before calling transport_init()? Don't we need to
have `sin' reflect the correct IP address by calling gethostname()?

thanks
Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: finserver compile error
Date: Mon, 28 Jan 2002 22:07:06 -0800
Lines: 44
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2808

I've also tried redownloading the hw2 files
into another directory using the gethw2 script
and I still get the error...

What is finserver anyway? Should I just
comment it out of the makefile to see
if it compiles then?

Abhishek Das wrote:
> 
> I got the same error.
> 
> On Mon, 28 Jan 2002, Luis Robles wrote:
> 
> > Is anyone else getting this compile error?
> >
> > saga7:~/cs244a/hw2> make all
> > umask 022
> > gcc -g -ansi -DSOLARIS  -c client.c
> > gcc -g -ansi -DSOLARIS  -c transport.c
> > gcc -g -ansi -DSOLARIS  -c network.c
> > gcc -g -ansi -DSOLARIS  -c mysock.c
> > gcc -g -ansi -DSOLARIS  -o client client.o transport.o network.o
> > mysock.o -lsocket -lnsl -lm
> > gcc -g -ansi -DSOLARIS  -c server.c
> > gcc -o server server.o transport.o network.o mysock.o -lsocket -lnsl -lm
> > make: *** No rule to make target `finserver.o', needed by `finserver'.
> > Stop.
> > saga7:~/cs244a/hw2>
> >
> > I used the gethw2 script, and have not modified
> > the makefile,
> >
> > Thanks....
> >
> 
> Abhishek Das
> Graduate Research Assistant
> Computer Systems Lab
> Stanford University
> 
> Address:-
> Escondido Village 33B
> Stanford CA 94305
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding Wrap arong (UINT_MAX)
Date: Mon, 28 Jan 2002 22:10:39 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2809

Hi,
Can we assume that the files transfered will be less than 4 GB so that we
don't have to bother about the wrap aroung. Also i think the data_len and
seq_num should be send in network order. Is that right.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Test Script Issues
Date: Mon, 28 Jan 2002 22:38:58 -0800
Lines: 57
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2810

Hi All,

     I tested my client and server against Martin's respective server and
client and we were able to transfer multiple large files back and fourth
without any problems.  However both of us have been getting test results of
the type that appear below.  This leads me to believe that either both of us
have misunderstood the assignment (in very similar ways), or there is a
problem with the test script.  I ask the TAs to please double check this
part of the test script and verify its functionalty as we aren't the only
two students who are running into this problem.  Thanks!

--Russ

   * Test 1.B: Transmitter fails to work in reliable mode with our Receiver
        [out of 10]


[SUMMARY] NOT OK
[RESULT]
===========================
===(T1.Ba) Byte count differences found (INCORRECT!)


---
Difference between expected result and your result:
1,2c1,2
<    2641   15865  110230 TESTDATA.0
<     600    3170   92696 TESTDATA.1
---
>    2666   15720  110230 TESTDATA.0
>    1812   10049  299328 TESTDATA.1


---


===========================
===(T1.Bb) Checksum differences found (INCORRECT!)


---
Difference between expected result and your result:
1,2c1,2
< 3108646830 110230 TESTDATA.0
< 3014550704 92696 TESTDATA.1
---
> 1705403052 110230 TESTDATA.0
> 4265638025 299328 TESTDATA.1


---



===============


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine41.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: unable to network_send stcpHeader
Date: Mon, 28 Jan 2002 22:49:24 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine41.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2811


I am unable to send stcpHeader using network_send. I am able though to
send an arbitrary string.

I know this is trivial, thats why I am asking :-)

any ideas?
-anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine24.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Processing ACKS etc.
Date: Mon, 28 Jan 2002 23:42:52 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2812

Since we can assume in first part of assignment 2 that packets are not
lost, do we need to worry about what an ACK packet has to tell? And do we
need to worry about send and recv windows?

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: When there are several child processes..
Date: Mon, 28 Jan 2002 23:52:15 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2813


This seems to be very basic... so I'm sorry to ask, but I need some
clarification...

Suppose 4 clients connect to the same server.
Then the server has 4 child processes for each connection,
  but the connections are all on the same UDP receiving port..
Server-side child processes will have to determine whether a received packet
is
for them or not, by looking at the source address of the packet.

My question is,
 if a packet arrives to the UDP port and one of the children reads it first,
 is the packet still available for reading by the rest of the children?

And, is the assignment going to be tested under multi-client situation?

Any corrections or comments please...



.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: finserver compile error
Date: Tue, 29 Jan 2002 00:08:42 -0800
Lines: 56
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2814


An incorrect Makefile was included in the HW2 directory. You can get the 
correct Makefile either by copying the file 
/afs/ir/class/cs244a/WWW/homeworks/hw2/src/Makefile or by calling gethw2 
in a *NEW* directory, and then copy your files over.

Pablo

On Mon, 28 Jan 2002, Luis Robles wrote:

> I've also tried redownloading the hw2 files
> into another directory using the gethw2 script
> and I still get the error...
> 
> What is finserver anyway? Should I just
> comment it out of the makefile to see
> if it compiles then?
> 
> Abhishek Das wrote:
> > 
> > I got the same error.
> > 
> > On Mon, 28 Jan 2002, Luis Robles wrote:
> > 
> > > Is anyone else getting this compile error?
> > >
> > > saga7:~/cs244a/hw2> make all
> > > umask 022
> > > gcc -g -ansi -DSOLARIS  -c client.c
> > > gcc -g -ansi -DSOLARIS  -c transport.c
> > > gcc -g -ansi -DSOLARIS  -c network.c
> > > gcc -g -ansi -DSOLARIS  -c mysock.c
> > > gcc -g -ansi -DSOLARIS  -o client client.o transport.o network.o
> > > mysock.o -lsocket -lnsl -lm
> > > gcc -g -ansi -DSOLARIS  -c server.c
> > > gcc -o server server.o transport.o network.o mysock.o -lsocket -lnsl -lm
> > > make: *** No rule to make target `finserver.o', needed by `finserver'.
> > > Stop.
> > > saga7:~/cs244a/hw2>
> > >
> > > I used the gethw2 script, and have not modified
> > > the makefile,
> > >
> > > Thanks....
> > >
> > 
> > Abhishek Das
> > Graduate Research Assistant
> > Computer Systems Lab
> > Stanford University
> > 
> > Address:-
> > Escondido Village 33B
> > Stanford CA 94305
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Can you please clarify what this error means.
Date: Tue, 29 Jan 2002 00:39:39 -0800
Lines: 88
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2815


Hi,

After revising the code for the grading script for part 1.B, there was a 
mislabeling of the test. In fact the test that was being performed was 1.C 
"Our Tx against student's Rx (reliable mode)". So up until now, test 1.B 
was labeled 1.C, and test 1.C was labeled 1.B in the report that you got.
I appologize for the confusion this typo may have caused.

In your case, Shankar, your receiver is writing the right amount of data
(file sizes are OK), but the contents are not, the checksum and the number
of \n [new lines] are different than in the original file.

Pablo


On Mon, 28 Jan 2002, SUL\AIR Laneroom Cluster wrote:

> Shankar,
> 
> the numbers in the first test your code fails are
> BYTE counts. the top ones are the expected
> and the bottom ones are yours. As far as I can
> tell, in some cases your code transmits fewer bytes
> and in some more.
> 
> The other numbers are just checksum numbers
> with no particular meaning to you, other than
> that your data isn't correctly transmitted (it gets
> corrupted somewhere, overwritten, contains garbage etc.)
> 
> Regards,
> Antonios.
> 
> Shankar Agarwal wrote:
> 
> > Hi,
> > I am getting the following errors when i submit the script for the part
> > 1B. I am not able to figure out what exactly is it saying. Is it that the
> > receiver gets more data or is it receiving less data.
> > Moreover there are 3 figures for each TESTDATA. what each of them means.
> > Thanks
> > Regards
> > Shankar
> > ===============
> >
> >     * Test 1.B: Transmitter fails to work in reliable mode with our
> > Receiver
> >         [out of 10]
> >
> > [SUMMARY] NOT OK
> > [RESULT]
> > ===========================
> > ===(T1.Ba) Byte count differences found (INCORRECT!)
> >
> > ---
> > Difference between expected result and your result:
> > 1,3c1,3
> > <    2641   15865  110230 TESTDATA.0
> > <    1786    9963  299328 TESTDATA.1
> > <     125     536    5098 TESTDATA.2
> > ---
> > >    2666   15720  110230 TESTDATA.0
> > >    1812   10049  299328 TESTDATA.1
> > >     131     501    5098 TESTDATA.2
> >
> > ---
> >
> > ===========================
> > ===(T1.Bb) Checksum differences found (INCORRECT!)
> >
> > ---
> > Difference between expected result and your result:
> > 1,3c1,3
> > < 3108646830 110230 TESTDATA.0
> > < 4231670217 299328 TESTDATA.1
> > < 2946179251 5098 TESTDATA.2
> > ---
> > > 1705403052 110230 TESTDATA.0
> > > 4265638025 299328 TESTDATA.1
> > > 2181702833 5098 TESTDATA.2
> >
> > ---
> >
> > ===============
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth9.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Regarding Wrap arong (UINT_MAX)
Date: Tue, 29 Jan 2002 01:24:51 -0800
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2816


Hi Shankar,

from the faq, we dont need to handle the wrap around for
sequence nos.

Thanks, Arvind

On Mon, 28 Jan 2002, Shankar Agarwal wrote:

> Hi,
> Can we assume that the files transfered will be less than 4 GB so that we
> don't have to bother about the wrap aroung. Also i think the data_len and
> seq_num should be send in network order. Is that right.
> Thanks
> Shankar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: attaching to process in ddd - instructions
Date: Tue, 29 Jan 2002 02:36:31 -0800
Lines: 39
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012300562 22459 128.12.133.48 (29 Jan 2002 10:36:02 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2817

Even quicker:

1. put sleep(30)
2. run client
3. from another prompt, run
    ddd client <pid> &
4. set a breakpoint and hit continue (hitting next will not work this way)

Also, for the server you don't even have to put the sleep(30), as long as
you attach ddd before you start the client. Once you get
connection-initiation working, you can drop the sleep(30) for both sides,
because after they connect they will block as the client waits for input.

Jonathan

"Araik Grigoryan"  wrote in message

>
> Here you go ...
>
> 1. Put sleep(30) somewhere in the child process. Compile.
> 2. Open DDD and open the client executable via the File menu.
> 3. By clicking on function names, make your way down to the location where
> you placed the sleep and insert a break point right after the sleep
> statement.
> 4. Now run the client with all it's arguments on the command line
> (outside DDD, of course).
> 5. Go back to DDD and choose Attach Process from the File menu. Choose the
> process of the child and wait until sleep is done.
> 6. If you see that nothing is happening for a long time, just click Next
> or Step once and you will see the pointer.
> 7. You can do the same for the server.
>
> araik
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Sequence number clarifications
Date: Tue, 29 Jan 2002 02:40:23 -0800
Lines: 45
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012300793 22461 128.12.133.48 (29 Jan 2002 10:39:53 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2818


"Martin Casado"  wrote in message

> Hello,
>
>   I would greatly appreciate it if someone could help me clear up
> a few questions I have about sequence numbering...
>
> Assume the active end starts the connection with seqx, and the passive
> replies with seqy (acking with seqx+1)
>
> 1. The assignment sheet states that the active end records seqy so that it
>    will expect the next byte of the passive end to start at seqy+1, can
>    we also assume that the passive end expects the active end to start at
>    seqx+1? (this isn't stated either way)

Yes, by acking with seqx+1, the passive end is telling the active end that
it expects seqx+1.

>
> 2. For acknowledgements, the project sheet says to set the "ack_seq" field
to
>    the sequence number of the next byte of data the receiver expects.
tcp_hdr
>    has no ack_seq field, do we set th_ack or th_seq to this value, or
both?

th_ack

>
> 3. Do we have to place/handle a value in the th_seq field of data packets?

Yes, absolutely! th_seq tells what the first sequence number in the packet
is, so it's absolutely essential for placing the data in the right place in
your window (this is a part B thing). If you meant ACK packets, no I don't
think you have to place or handle it (except for SYN ACK of course). The
TA's would know for sure.

>
> Thanks :-)
>
>                     ~~m
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Test Script Issues
Date: Tue, 29 Jan 2002 02:51:21 -0800
Lines: 86
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012301450 22596 128.12.133.48 (29 Jan 2002 10:50:50 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2819


"Russell Greene"  wrote in message

> Hi All,
>
>      I tested my client and server against Martin's respective server and
> client and we were able to transfer multiple large files back and fourth
> without any problems.  However both of us have been getting test results
of
> the type that appear below.  This leads me to believe that either both of
us
> have misunderstood the assignment (in very similar ways), or there is a
> problem with the test script.  I ask the TAs to please double check this
> part of the test script and verify its functionalty as we aren't the only
> two students who are running into this problem.  Thanks!

Wow. The thing that I find really interesting is that the script actually
transfers THREE files. One you appear to have transferred correctly (the
really small one), one you have the right size but the wrong data, and the
other you have way fewer bytes than you should.

My program transfers fine with other students and with all parts of the test
script, and I know of others who have no problems, so I expect you've got
some more debugging to do. :-(

Are you sure that the local file named "rcvd" (the file that client puts
what it downloads in) has the same size and checksum as whatever file is
being sent from the other side? It seems that if you're somehow writing the
data from the TAs script, you'd probably be writing the wrong data from your
own server in some cases as well. Reexamine your sliding-window code, as
frequently problems like this come from incorrect handling of special cases
in your implementation. Also, maybe you're somehow dropping data on the way
to the parent process.

Jonathan

>
> --Russ
>
>    * Test 1.B: Transmitter fails to work in reliable mode with our
Receiver
>         [out of 10]
>
>
> [SUMMARY] NOT OK
> [RESULT]
> ===========================
> ===(T1.Ba) Byte count differences found (INCORRECT!)
>
>
> ---
> Difference between expected result and your result:
> 1,2c1,2
> <    2641   15865  110230 TESTDATA.0
> <     600    3170   92696 TESTDATA.1
> ---
> >    2666   15720  110230 TESTDATA.0
> >    1812   10049  299328 TESTDATA.1
>
>
> ---
>
>
> ===========================
> ===(T1.Bb) Checksum differences found (INCORRECT!)
>
>
> ---
> Difference between expected result and your result:
> 1,2c1,2
> < 3108646830 110230 TESTDATA.0
> < 3014550704 92696 TESTDATA.1
> ---
> > 1705403052 110230 TESTDATA.0
> > 4265638025 299328 TESTDATA.1
>
>
> ---
>
>
>
> ===============
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: Tue, 29 Jan 2002 02:56:10 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012301738 22608 128.12.133.48 (29 Jan 2002 10:55:38 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2820


"Dhawal Kumar"  wrote in message

> Since we can assume in first part of assignment 2 that packets are not
> lost, do we need to worry about what an ACK packet has to tell? And do we
> need to worry about send and recv windows?

The test script checks that you enforce the window sizes, so you definitely
do have to worry about that. I suspect it also checks that your ACK packets
are well-formed. I think the point of the assumption is that you don't have
to do anything about an ACK packet that you were expecting to receive but
never did. If you hack together some one-off thing that meets the
requirements of 2.A rather than designing something that's better suited to
the overall assignment, you're gonna really be hurting when you try to do
2.B.

Jonathan

>
> Dhawal Kumar
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: When there are several child processes..
Date: Tue, 29 Jan 2002 03:32:18 -0800
Lines: 57
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012303937 22881 128.12.133.48 (29 Jan 2002 11:32:17 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2821


"BAEHOPIL"  wrote in message

>
> This seems to be very basic... so I'm sorry to ask, but I need some
> clarification...
>
> Suppose 4 clients connect to the same server.
> Then the server has 4 child processes for each connection,
>   but the connections are all on the same UDP receiving port..
> Server-side child processes will have to determine whether a received
packet
> is
> for them or not, by looking at the source address of the packet.
>
> My question is,
>  if a packet arrives to the UDP port and one of the children reads it
first,
>  is the packet still available for reading by the rest of the children?

It's actually not all that basic. In this assignment we're writing a
simplified variant of TCP. TCP is usually run directly on top of IP, but in
order for us to actually do that, we'd pretty much have to be in the kernel,
and you know they're not going to let us into the kernel on a machine in
Sweet Hall. (Not that we'd really want to be there; debugging is a pain in
the ass...sucks when you segfault in the kernel....)

So, we have this network.c thing, which (cooperating with mysock.c a little
bit) uses UDP to simulate IP. Thus, it's best to forget the UDP socket
exists at all, and consider network.c to be a way of reading raw IP.

IP has no facility for ports; anything you send over IP is addressed to the
machine, period. It is then the job of higher-level protocols like UDP and
TCP to do the (de)multiplexing required to implement ports. STCP, being
simplified TCP, cannot do ports--it's effectively a single-ported TCP. So,
multi-client servers _can't_ be implemented using just STCP. (The server
would have to have its own multiplexer on top of STCP.) You can see that our
server code expects only one client at a time, and it's as a result of this.

>
> And, is the assignment going to be tested under multi-client situation?

There are definitely some tricky cases to be thought about here. For
instance, in normal TCP, the sending party's address and port are used to
demux multiple connections on one port. In STCP, the first computer to SYN
to a port gets it, and everyone else should be SOL. Whether that will be
tested, well, who knows?

Jonathan

>
> Any corrections or comments please...
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic10.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Disk FUll error which submitting test.
Date: Tue, 29 Jan 2002 08:01:00 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2822

Hi,
I got the following error while submitting the code for test.
Please look into it.
tar (child): Cannot open archive
/afs/ir/class/cs244a/submissions/hw2.A/akishan/shankara/files.tar.gz: No
space left on device
tar (child): Error is not recoverable: exiting now
======================================================================
Your code could not be sumitted because of some error.
If you think there is a mistake, please send an email to your TA
(akishan;  with the relevant information.
======================================================================
Thanks
Regards
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic10.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Just a trivial clarification.
Date: Tue, 29 Jan 2002 08:02:12 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2823

Hi,
I just want a clarification in case i am missing something. Is it ok to
assume that the ack packet that we get does not contain data i.e. there is
no piggybacking.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: Just a trivial clarification.
Date: Tue, 29 Jan 2002 08:48:36 -0800
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2824

From the assignment:

"Although the packet structure is defined so that an acknowledgment can
accompany data in the same packet, STCP does not support this. All ACK
packets should have no payload (i.e. data_len field in the packet header set
to zero)."

Bret

"Shankar Agarwal"  wrote in message

> Hi,
> I just want a clarification in case i am missing something. Is it ok to
> assume that the ack packet that we get does not contain data i.e. there is
> no piggybacking.
> Thanks
> Shankar
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Just a trivial clarification.
Date: Tue, 29 Jan 2002 08:51:39 -0800
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2825


The second bullet in the assignment sheet explicitly says that STCP does
not support piggy backing and ACK packets should haev no payload.

                ~~m

> Hi,
> I just want a clarification in case i am missing something. Is it ok to
> assume that the ack packet that we get does not contain data i.e. there is
> no piggybacking.
> Thanks
> Shankar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Disk FUll error which submitting test.
Date: Tue, 29 Jan 2002 10:31:21 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2826

Hi,

The disk space got exhausted last night because one student's code started
outputing debugging information while going into a infinite loop. We are
working on a way of eliminating this problem in the future. Until then, I
recomend that you do not submit code that produces too much debugging
information, otherwise you will hurt yourself (the test will fail) and
hurt your fellow students who cannot test their code.

Pablo

On Tue, 29 Jan 2002, Shankar Agarwal wrote:

> Hi,
> I got the following error while submitting the code for test.
> Please look into it.
> tar (child): Cannot open archive
> /afs/ir/class/cs244a/submissions/hw2.A/akishan/shankara/files.tar.gz: No
> space left on device
> tar (child): Error is not recoverable: exiting now
> ======================================================================
> Your code could not be sumitted because of some error.
> If you think there is a mistake, please send an email to your TA
> (akishan;  with the relevant information.
> ======================================================================
> Thanks
> Regards
> Shankar
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: th_seq and th_ack associated with FIN, FIN_ACK
Date: Tue, 29 Jan 2002 11:23:31 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2827


 I wonder how I should set the th_seq and th_ack fields for FIN and
FIN_ACK..
 I couldn't find a detailed description about it, unlike SYN and SYN_ACK...

 I have been setting them to just zero,
 but I'm thinking that it may be causing my test failure with TA's rx and
tx..



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Can you please clarify what this error means.
Date: Tue, 29 Jan 2002 13:12:59 -0800
Lines: 115
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2828

Wait a sec.  So you're telling me that the expected result in my test 
below is "0 0 0 TESTDATA.0" and the actual result my program generated 
is 3 lines long?

Please clarify.  Thanks!

-Sandeep

  * Test 1.B: Transmitter fails to work in reliable mode with our Receiver
	[out of 10]

[SUMMARY] NOT OK
[RESULT]
===========================
===(T1.Ba) Byte count differences found (INCORRECT!)
Student's Tx against our Rx (reliable mode)

---

---
Difference between expected result and your result:
1c1,3
< 0       0       0 TESTDATA.0
---

 > 2666   15720  110230 TESTDATA.0
 > 1812   10049  299328 TESTDATA.1
 > 131     501    5098 TESTDATA.2



SUL\\AIR Laneroom Cluster wrote:

> Shankar,
> 
> the numbers in the first test your code fails are
> BYTE counts. the top ones are the expected
> and the bottom ones are yours. As far as I can
> tell, in some cases your code transmits fewer bytes
> and in some more.
> 
> The other numbers are just checksum numbers
> with no particular meaning to you, other than
> that your data isn't correctly transmitted (it gets
> corrupted somewhere, overwritten, contains garbage etc.)
> 
> Regards,
> Antonios.
> 
> Shankar Agarwal wrote:
> 
> 
>>Hi,
>>I am getting the following errors when i submit the script for the part
>>1B. I am not able to figure out what exactly is it saying. Is it that the
>>receiver gets more data or is it receiving less data.
>>Moreover there are 3 figures for each TESTDATA. what each of them means.
>>Thanks
>>Regards
>>Shankar
>>===============
>>
>>    * Test 1.B: Transmitter fails to work in reliable mode with our
>>Receiver
>>        [out of 10]
>>
>>[SUMMARY] NOT OK
>>[RESULT]
>>===========================
>>===(T1.Ba) Byte count differences found (INCORRECT!)
>>
>>---
>>Difference between expected result and your result:
>>1,3c1,3
>><    2641   15865  110230 TESTDATA.0
>><    1786    9963  299328 TESTDATA.1
>><     125     536    5098 TESTDATA.2
>>---
>>
>>>   2666   15720  110230 TESTDATA.0
>>>   1812   10049  299328 TESTDATA.1
>>>    131     501    5098 TESTDATA.2
>>>
>>---
>>
>>===========================
>>===(T1.Bb) Checksum differences found (INCORRECT!)
>>
>>---
>>Difference between expected result and your result:
>>1,3c1,3
>>< 3108646830 110230 TESTDATA.0
>>< 4231670217 299328 TESTDATA.1
>>< 2946179251 5098 TESTDATA.2
>>---
>>
>>>1705403052 110230 TESTDATA.0
>>>4265638025 299328 TESTDATA.1
>>>2181702833 5098 TESTDATA.2
>>>
>>---
>>
>>===============
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: Tue, 29 Jan 2002 13:56:07 -0800
Lines: 66
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2829

Why do you have to worry about window sizes for Part A?  More 
importantly, why does the test script test this case?

The way I see it, you don't need a sending or receiving window for Part 
A.  When the application wants to transmit data, it does a mywrite, then 
the child process reads upto 536 bytes of that data, packages it, and 
sends it.  If the transmitter is doing a mywrite with a ton of data, the 
control loop in the child process will keep triggering until all the 536 
byte data blocks (and the remainder) are complete.  So the child ends up 
sending a bunch of packets (via network_send) when the caller asked to 
do a large mywrite.

On the receiving end, the child is using network_recv to get data from 
the "outside world" socket.  It receives datagrams of maximum size 536 
bytes + sizeof(STCPHeader).  It extracts the data and sends it to the 
parent.

Everything should work fine, right? There's no chance of exceeding the 
3072 byte buffer size since you'll only get 536+hdrSize bytes at a time.

If I'm missing something, please correct me.  My program's got lots of 
problems right now, and if I'm conceptually wrong about the above, then 
knowing how this is supposed to work correctly will probably help me 
solve my problems.

-Sandeep



Jonathan Keljo wrote:

> "Dhawal Kumar"  wrote in message
> 
> 
>>Since we can assume in first part of assignment 2 that packets are not
>>lost, do we need to worry about what an ACK packet has to tell? And do we
>>need to worry about send and recv windows?
>>
> 
> The test script checks that you enforce the window sizes, so you definitely
> do have to worry about that. I suspect it also checks that your ACK packets
> are well-formed. I think the point of the assumption is that you don't have
> to do anything about an ACK packet that you were expecting to receive but
> never did. If you hack together some one-off thing that meets the
> requirements of 2.A rather than designing something that's better suited to
> the overall assignment, you're gonna really be hurting when you try to do
> 2.B.
> 
> Jonathan
> 
> 
>>Dhawal Kumar
>>
>>
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Network Initiation and reliability
Date: 29 Jan 2002 21:59:56 GMT
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine11.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2830


The first bullet point under "Network Initiation" states that the
active end sends an SYN and then waits for a SYN_ACK.

If we send the SYN using network_send, which is unreliable,
how do we know that the SYN packet was not dropped? Or is it
the case that we don't use network_send for the SYN handshake?
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Sending/receiving windows for part A
Date: Tue, 29 Jan 2002 14:18:27 -0800
Lines: 25
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2831

Hi all,
My question is, since we do not have to do retransmissions yet, where do I
fit in the use of sending/receiving windows? Except of course to ensure
that the arriving packets fall in the windows? I mean, do we need to
buffer the packets before sending them on the network / sending them to
parent application? Or do we just send  packets as we (STCP) get them,
since they will be in order?

I guess I am confused about the functionality required. Any help would be
great!
Thanks,
Ashmi

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Found it
Date: Tue, 29 Jan 2002 16:52:27 -0800
Lines: 67
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2832

I ran into this problem, too, but I had a few other problems, which 
ironically, made it easier to figure out (at least for my program; 
hopefully you guys have the same problem and thus the same solution will 
work):

Despite what the TAs say, we really can't assume a reliable delivery 
mechanism in this assignment; we are after all, sending UDP datagrams 
everywhere.

One of the problems I was seeing in my project is that the client would 
occasionally block forever (against my server).  At first I thought that 
my code was dropping a packet it had received (like if I got one type of 
packet when I really wanted another in some state).  But I verified that 
that wasn't the problem.  So I added debug code to verify that all my 
reads/writes/network_send's/network_recv's were indeed transferring the 
right number of bytes.  All of that seemed fine.

Yet from my debugging output I could see that some message was sent from 
one end and simply not received on the other (and it varied from run to 
run...often I'd get perfectly clean runs on the 50K file I was using for 
testing).  The only conclusion I could draw was that I was losing 
packets at the UDP level.

So I ended up adding a tiny delay (1ms) before every network_send.  That 
way the network would have enough time to deliver my packet to the 
receiving end.  It's a hack, but it's the best I can do right now since 
I don't have time to do Part B.  And it worked.

-Sandeep

Russell Greene wrote:

> I too am passing all of the tests except for the parts which use the TA's
> receiver (both parts of Test 1.B).  I have no idea why, as in my own tests
> the transmitter appears to work fine.
> 
> -Russ
> 
> "Martin Casado"  wrote in message
> 
> 
>>Hi,
>>
>>  I'm passing all of the tests except for the ones which use
>>the TA's receiver... and I'm really not sure why. Is it against
>>the rules/honor code to modify my transport.c to open a file
>>in a shared directory of mine, and write out a bunch of
>>debug code during testing? (like all packets sent by
>>the TA's receiver)
>>  I'm assuming that this is not allowed.... but it never
>>hurts to ask (I hope) :-)
>>
>>                    ~~m
>>
>>
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: th_seq and th_ack associated with FIN, FIN_ACK
Date: Tue, 29 Jan 2002 17:15:23 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: BAEHOPIL 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2833

We don't care the th_seq and the_ack fields for FIN and FIN_ACK. You can
just set them zeros. Xin

On Tue, 29 Jan 2002, BAEHOPIL wrote:

> 
>  I wonder how I should set the th_seq and th_ack fields for FIN and
> FIN_ACK..
>  I couldn't find a detailed description about it, unlike SYN and SYN_ACK...
> 
>  I have been setting them to just zero,
>  but I'm thinking that it may be causing my test failure with TA's rx and
> tx..
> 
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Sending/receiving windows for part A
Date: Tue, 29 Jan 2002 17:33:17 -0800
Lines: 42
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Ashmi 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2834

Milestone #1 only considers the reliable mode. In a reliable mode, no
packets are dropped, timeout or appear out of order. So you may not need
to buffer the packets. Best, Xin


On Tue, 29 Jan 2002, Ashmi wrote:

> Hi all,
> My question is, since we do not have to do retransmissions yet, where do I
> fit in the use of sending/receiving windows? Except of course to ensure
> that the arriving packets fall in the windows? I mean, do we need to
> buffer the packets before sending them on the network / sending them to
> parent application? Or do we just send  packets as we (STCP) get them,
> since they will be in order?
> 
> I guess I am confused about the functionality required. Any help would be
> great!
> Thanks,
> Ashmi
> 
> ###############################################################################
> 
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
> 
> ###############################################################################
> 
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth3.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: Tue, 29 Jan 2002 18:50:19 -0800
Lines: 83
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: myth3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Sandeep Tamhankar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2835


Hi Sandeep,

What prevents your transmitter from sending multiple packets ?
If it waits for the ack from receiver after each packet, the
window effectively reduces to 536 bytes :-)

Does that give you any ideas ? Or may be i did not understand
the question...

Thanks, Arvind


On Tue, 29 Jan 2002, Sandeep Tamhankar wrote:

> Why do you have to worry about window sizes for Part A?  More
> importantly, why does the test script test this case?
>
> The way I see it, you don't need a sending or receiving window for Part
> A.  When the application wants to transmit data, it does a mywrite, then
> the child process reads upto 536 bytes of that data, packages it, and
> sends it.  If the transmitter is doing a mywrite with a ton of data, the
> control loop in the child process will keep triggering until all the 536
> byte data blocks (and the remainder) are complete.  So the child ends up
> sending a bunch of packets (via network_send) when the caller asked to
> do a large mywrite.
>
> On the receiving end, the child is using network_recv to get data from
> the "outside world" socket.  It receives datagrams of maximum size 536
> bytes + sizeof(STCPHeader).  It extracts the data and sends it to the
> parent.
>
> Everything should work fine, right? There's no chance of exceeding the
> 3072 byte buffer size since you'll only get 536+hdrSize bytes at a time.
>
> If I'm missing something, please correct me.  My program's got lots of
> problems right now, and if I'm conceptually wrong about the above, then
> knowing how this is supposed to work correctly will probably help me
> solve my problems.
>
> -Sandeep
>
>
>
> Jonathan Keljo wrote:
>
> > "Dhawal Kumar"  wrote in message
> > 
> >
> >>Since we can assume in first part of assignment 2 that packets are not
> >>lost, do we need to worry about what an ACK packet has to tell? And do we
> >>need to worry about send and recv windows?
> >>
> >
> > The test script checks that you enforce the window sizes, so you definitely
> > do have to worry about that. I suspect it also checks that your ACK packets
> > are well-formed. I think the point of the assumption is that you don't have
> > to do anything about an ACK packet that you were expecting to receive but
> > never did. If you hack together some one-off thing that meets the
> > requirements of 2.A rather than designing something that's better suited to
> > the overall assignment, you're gonna really be hurting when you try to do
> > 2.B.
> >
> > Jonathan
> >
> >
> >>Dhawal Kumar
> >>
> >>
> >
> >
>
>
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: Problem attaching child process in gdb
Date: Tue, 29 Jan 2002 19:35:21 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2836



  Hi everybody,

  I have problem in attaching child process in gdb. I open another window
and run gdb then try to attach to the child process by typing : "attach
1549" (1549 is my child process id). Here is the answer I get from gdb:

Attaching to program
`/afs/ir.stanford.edu/users/a/m/aminf13/cs244a/assignment2/parta/client',
process 1549
procfs:3643 -- /proc/1549: No such file or directory.
do_attach: couldn't open /proc file for process 1549

  I don't know if I'm doing anything wrong. Any help will be greatly
appreciated.

  Yours,
  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine19.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Sequence number clarifications
Date: Tue, 29 Jan 2002 19:42:03 -0800
Lines: 57
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine19.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2837

Does the answer in 2 mean that we don't need to set th_seq field for ACKs?
Or do we need to treat an ACK as a 1 byte packet as far as sequence number
space is concerned i.e. sender of ACK would advance his sequence number by
1 and receiver would slide his receive window by 1 (of course assuming
that the receiver was waiting for this seq number).

Dhawal Kumar

On Tue, 29 Jan 2002, Jonathan Keljo wrote:

>
> "Martin Casado"  wrote in message
> 
> > Hello,
> >
> >   I would greatly appreciate it if someone could help me clear up
> > a few questions I have about sequence numbering...
> >
> > Assume the active end starts the connection with seqx, and the passive
> > replies with seqy (acking with seqx+1)
> >
> > 1. The assignment sheet states that the active end records seqy so that it
> >    will expect the next byte of the passive end to start at seqy+1, can
> >    we also assume that the passive end expects the active end to start at
> >    seqx+1? (this isn't stated either way)
>
> Yes, by acking with seqx+1, the passive end is telling the active end that
> it expects seqx+1.
>
> >
> > 2. For acknowledgements, the project sheet says to set the "ack_seq" field
> to
> >    the sequence number of the next byte of data the receiver expects.
> tcp_hdr
> >    has no ack_seq field, do we set th_ack or th_seq to this value, or
> both?
>
> th_ack
>
> >
> > 3. Do we have to place/handle a value in the th_seq field of data packets?
>
> Yes, absolutely! th_seq tells what the first sequence number in the packet
> is, so it's absolutely essential for placing the data in the right place in
> your window (this is a part B thing). If you meant ACK packets, no I don't
> think you have to place or handle it (except for SYN ACK of course). The
> TA's would know for sure.
>
> >
> > Thanks :-)
> >
> >                     ~~m
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: Another strange problem with gdb
Date: Tue, 29 Jan 2002 19:43:10 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2838



  I noticed another problem, with gdb, when I run the server program
(original version, no modification is done), it blocks and waits for
connection from client, but when I trace is step by step in gdb, it does
not block and exits with code 1! I am wondering what this means and why
this happens.

  Yours,
  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: Tue, 29 Jan 2002 20:41:51 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2839

>
> The test script checks that you enforce the window sizes, so you definitely
> do have to worry about that. I suspect it also checks that your ACK packets
> are well-formed. I think the point of the assumption is that you don't have
> to do anything about an ACK packet that you were expecting to receive but
> never did. If you hack together some one-off thing that meets the
> requirements of 2.A rather than designing something that's better suited to
> the overall assignment, you're gonna really be hurting when you try to do
> 2.B.
>
> Jonathan
>

Quick question...

Packets outside of the window are easy to deal with.. aka throw them out.
But in the assignment sheet it says that "received data may cross either end of
the current receiver window; in this case, the data is split into two parts
and dealt with accordingly"

Does that mean.. if we get a packet that crosses the top boundry of the window,
we chop it in two, Ack the first hald and forget about the second half?  And
if we get a packet that is below the bottom window we chop it, forget about
the first slice and Ack the second?  Won't we then need retransmission to
handle data of split packets where packets above the window are thrown
away?

Thoughts?

Thanks :-)

                ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: Tue, 29 Jan 2002 22:47:51 -0800
Lines: 104
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2840

No, you're right on the money; after some deep thought I realized what I 
was doing wrong.  I still don't use sliding windows, but I've covered 
this is far as Part A goes and I pass all the tests now.

-Sandeep

Arvind Bhat wrote:

> Hi Sandeep,
> 
> What prevents your transmitter from sending multiple packets ?
> If it waits for the ack from receiver after each packet, the
> window effectively reduces to 536 bytes :-)
> 
> Does that give you any ideas ? Or may be i did not understand
> the question...
> 
> Thanks, Arvind
> 
> 
> On Tue, 29 Jan 2002, Sandeep Tamhankar wrote:
> 
> 
>>Why do you have to worry about window sizes for Part A?  More
>>importantly, why does the test script test this case?
>>
>>The way I see it, you don't need a sending or receiving window for Part
>>A.  When the application wants to transmit data, it does a mywrite, then
>>the child process reads upto 536 bytes of that data, packages it, and
>>sends it.  If the transmitter is doing a mywrite with a ton of data, the
>>control loop in the child process will keep triggering until all the 536
>>byte data blocks (and the remainder) are complete.  So the child ends up
>>sending a bunch of packets (via network_send) when the caller asked to
>>do a large mywrite.
>>
>>On the receiving end, the child is using network_recv to get data from
>>the "outside world" socket.  It receives datagrams of maximum size 536
>>bytes + sizeof(STCPHeader).  It extracts the data and sends it to the
>>parent.
>>
>>Everything should work fine, right? There's no chance of exceeding the
>>3072 byte buffer size since you'll only get 536+hdrSize bytes at a time.
>>
>>If I'm missing something, please correct me.  My program's got lots of
>>problems right now, and if I'm conceptually wrong about the above, then
>>knowing how this is supposed to work correctly will probably help me
>>solve my problems.
>>
>>-Sandeep
>>
>>
>>
>>Jonathan Keljo wrote:
>>
>>
>>>"Dhawal Kumar"  wrote in message

>>>
>>>
>>>>Since we can assume in first part of assignment 2 that packets are not
>>>>lost, do we need to worry about what an ACK packet has to tell? And do we
>>>>need to worry about send and recv windows?
>>>>
>>>>
>>>The test script checks that you enforce the window sizes, so you definitely
>>>do have to worry about that. I suspect it also checks that your ACK packets
>>>are well-formed. I think the point of the assumption is that you don't have
>>>to do anything about an ACK packet that you were expecting to receive but
>>>never did. If you hack together some one-off thing that meets the
>>>requirements of 2.A rather than designing something that's better suited to
>>>the overall assignment, you're gonna really be hurting when you try to do
>>>2.B.
>>>
>>>Jonathan
>>>
>>>
>>>
>>>>Dhawal Kumar
>>>>
>>>>
>>>>
>>>
>>
>>
>>--
>>---------------------------------------------
>>Sandeep V. Tamhankar
>>M.S. Student
>>Computer Science
>>Email: 
>>
>>
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: Tue, 29 Jan 2002 22:50:11 -0800
Lines: 53
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2841

I believe that for Part A, since we're assuming reliable underlying 
transport, the case you describe isn't possible.  The closest thing is 
that you'll get full repeats of packets you've already processed (and 
would thus be outside of your window).

-Sandeep

Martin Casado wrote:

>>The test script checks that you enforce the window sizes, so you definitely
>>do have to worry about that. I suspect it also checks that your ACK packets
>>are well-formed. I think the point of the assumption is that you don't have
>>to do anything about an ACK packet that you were expecting to receive but
>>never did. If you hack together some one-off thing that meets the
>>requirements of 2.A rather than designing something that's better suited to
>>the overall assignment, you're gonna really be hurting when you try to do
>>2.B.
>>
>>Jonathan
>>
>>
> 
> Quick question...
> 
> Packets outside of the window are easy to deal with.. aka throw them out.
> But in the assignment sheet it says that "received data may cross either end of
> the current receiver window; in this case, the data is split into two parts
> and dealt with accordingly"
> 
> Does that mean.. if we get a packet that crosses the top boundry of the window,
> we chop it in two, Ack the first hald and forget about the second half?  And
> if we get a packet that is below the bottom window we chop it, forget about
> the first slice and Ack the second?  Won't we then need retransmission to
> handle data of split packets where packets above the window are thrown
> away?
> 
> Thoughts?
> 
> Thanks :-)
> 
>                 ~~m
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the test scprit.
Date: Tue, 29 Jan 2002 23:05:23 -0800
Lines: 71
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2842

Well, presumably the reason that you provided the tests to students this 
year is that they could learn more from seeing their mistakes and trying 
to figure out the solutions than for the graders to find their mistakes 
and tell them the answer.

 From that perspective, I don't think it's unreasonable to want a little 
more information from the tests.  In the last project test failures 
indicated what was done to produce the failure so that the student could 
debug it on his/her own.  It would've been nice to have the same 
flexibility in this project.

That said, the least you can do is explain what the results of a given 
test mean; for example, which output is the expected output vs. the 
actual output of running our program.  Also, it would be great to know 
what those results mean.  For example see the following output that I 
just made up:

35  85 2195  TESTA.0

40  76 2500  TESTA.0

What do the numbers mean?  After sifting through a number of postings 
earlier today, I figured it out.  But it was not at all obvious.

My point is that these tests weren't made available to us as an act of 
kindness or mercy.  Thus we have no reason to be grateful for their 
existence.  They were provided to help us learn more about the cases we 
need to cover.  And this particular set of tests doesn't do that as well 
as the first project's tests, so people like Shankar (and myself earlier 
today) are stumbling to figure out how to deal with it.

Ok, I'm done with my rampage now.

-Sandeep

SUL\\AIR Laneroom Cluster wrote:

> Shankar,
> the testing platform should be familiar to you by
> now:) I'm afraid, everyone in the class has to work with
> the test scripts given. Last year people didn't even
> have this flexibility!
> 
> Regards,
> Antonios.
> 
> Shankar Agarwal wrote:
> 
> 
>>Hi,
>>I am seeing a strange problem with the test scpirt. My tests are failing
>>for the cases when TA's program is supposed to act as a receiver. I don't
>>see this problem in my testing. Is it possible to get hold of the binary
>>which does this testing. As the result does not give much information
>>regarding what went wrong while testing my program. Is it also possible to
>>get the files which the TA's script is trying to transfer. So that we can
>>use them to test to figure out what went wrong.
>>Thanks
>>Shankar
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!mgulati
From: Madhup Gulati 
Newsgroups: su.class.cs244a
Subject: passing peer address!
Date: Tue, 29 Jan 2002 23:37:22 -0800
Lines: 34
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2843


I have the peer address got from sockfd. But I am in
transport_sock_io()
function which is called from control loop. Here I don't have access to
syn_sd[1] . So how do I send this peer address to the parent.

 I can declare a new address variable in the context structure which gets
passed to all fuctions or add a new argument to all my fuctions so that
peer address is stored and passed. Both are not good techniques...

Ne suggestions??

Thanks,

Madhup


-------------------------------------------------------------------------------
Madhup Gulati
Graduate Student              Home:
Computer Science Dept.        141 K
William Gates Building        Escondido Road
Stanford University           Stanford
CA 94305                      CA 94305

          Ph: 650-497-4701

-------------------------------------------------------------------------------






.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: passing peer address!
Date: Tue, 29 Jan 2002 23:58:54 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2844

I guess it depends on where you finish setting up
your initial connection,

I take care of that in my control loop, so
I added a field to ctx, and that seems to work fine...

Cheers

Madhup Gulati wrote:
> 
> I have the peer address got from sockfd. But I am in
> transport_sock_io()
> function which is called from control loop. Here I don't have access to
> syn_sd[1] . So how do I send this peer address to the parent.
> 
>  I can declare a new address variable in the context structure which gets
> passed to all fuctions or add a new argument to all my fuctions so that
> peer address is stored and passed. Both are not good techniques...
> 
> Ne suggestions??
> 
> Thanks,
> 
> Madhup
> 
> -------------------------------------------------------------------------------
> Madhup Gulati
> Graduate Student              Home:
> Computer Science Dept.        141 K
> William Gates Building        Escondido Road
> Stanford University           Stanford
> CA 94305                      CA 94305
> 
>           Ph: 650-497-4701
> 
> -------------------------------------------------------------------------------

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: getpeername vs. network_peer_name
Date: Wed, 30 Jan 2002 00:04:31 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2845

Are we supposed to always use network_peer_name to
get the peer struct to pass up to the parent?

I was using getpeername for the active connection and that
worked, but I had to switch to network_peer_name on
the passive connection because getpeername was giving
me a the following error:

": Transport endpoint is not connected"

Is this due to the underlying usage of UDP sockets, or
something else?

Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Submission Error
Date: Wed, 30 Jan 2002 00:17:24 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2846

Hi All,

     I cannot submit my project.  When I run the command in the assignment

/afs/ir/class/cs244a/bin/submit.pl hw2.X ta I get a permission denied.

Immediately below (in the handout) it has the line

/afs/ir/class/cs244a/bin/submit ta hw2.X

I run that and it tells me that the ta should go after the assignment.  I
make the fix and run the command and it then tells me that hw2.A is not a
valid assignment for this class.

What am I doing wrong?

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Sequence number clarifications
Date: 30 Jan 2002 08:37:47 GMT
Lines: 14
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2847

Dhawal Kumar  writes:

>Does the answer in 2 mean that we don't need to set th_seq field for ACKs?
>Or do we need to treat an ACK as a 1 byte packet as far as sequence number
>space is concerned i.e. sender of ACK would advance his sequence number by
>1 and receiver would slide his receive window by 1 (of course assuming
>that the receiver was waiting for this seq number).

>Dhawal Kumar

If a packet contains only an ACK, the sequence number does not need to be
set (but it doesn't hurt, either).  Sending an ACK doesn't advance the
sender's sequence number.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: unable to network_send stcpHeader
Date: 30 Jan 2002 08:43:05 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2848

Anurag Gupta  writes:


>I am unable to send stcpHeader using network_send. I am able though to
>send an arbitrary string.

>I know this is trivial, thats why I am asking :-)

>any ideas?
>-anurag


Are you passing a valid pointer and length to the stcpheader?  Have you
passed a final NULL pointer to terminate the varargs?  Is it crashing,
or is errno being set?

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: 30 Jan 2002 08:49:33 GMT
Lines: 29
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2849

"Jonathan Keljo"  writes:


>"Dhawal Kumar"  wrote in message

>> Since we can assume in first part of assignment 2 that packets are not
>> lost, do we need to worry about what an ACK packet has to tell? And do we
>> need to worry about send and recv windows?

>The test script checks that you enforce the window sizes, so you definitely
>do have to worry about that. I suspect it also checks that your ACK packets
>are well-formed. I think the point of the assumption is that you don't have
>to do anything about an ACK packet that you were expecting to receive but
>never did. If you hack together some one-off thing that meets the
>requirements of 2.A rather than designing something that's better suited to
>the overall assignment, you're gonna really be hurting when you try to do
>2.B.


This is a very good point.  I'd *really* encourage people to think of this
as a large assignment, with the real goal being the implementation of a
completely functional TCP-like transport layer.  Treating it as three
separate assignments, and planning to just add on functionality for the
next milestone, will be very difficult.  It's much easier to start out
with a design that seems reasonable and will still be useful for the
next milestone--i.e. even though packet losses and retransmissions aren't
covered in the first milestone, you should still be thinking about these
as you decide how to implement even the basics of the protocol.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: 30 Jan 2002 08:58:40 GMT
Lines: 15
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2850

Sandeep Tamhankar  writes:

>I believe that for Part A, since we're assuming reliable underlying 
>transport, the case you describe isn't possible.  The closest thing is 
>that you'll get full repeats of packets you've already processed (and 
>would thus be outside of your window).


Well, since the underlying transport layer is still UDP, it's possible.
(If you haven't had a chance to implement support for this case yet, it would
probably make sense to throw some asserts or debugging output in to catch
the condition before it causes difficult-to-reproduce bugs to appear).

But as the assignment states, the first milestone will not test this
condition.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: When there are several child processes..
Date: 30 Jan 2002 09:09:41 GMT
Lines: 11
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2851


> And, is the assignment going to be tested under multi-client situation?

Well, it would make sense to me from both a software engineering and didactic
viewpoint to approach the assignment as though you were implementing a
real TCP layer--i.e. maintain a context for a particular socket (e.g.
using the context_t given in transport.c) rather than have global variables
polluting the whole process, etc.

But, in terms of multiple clients, no--that's not really practical with
the current setup.
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: Wed, 30 Jan 2002 01:19:22 -0800
Lines: 49
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2852

If our sender window is full and the application gives more data to
transmit, shoud we buffer it (if yes to what extent - I mean is there an
upper limit) or should we not read the data until the sender window clears
up - in this case select on data_sd will immediately return and we'll not
read again because there is no space in window and this would keep happening -
effectively a busy waiting.

This can happen even in first part of assigment 2 if sender gives more
that 3072 bytes to send.

A couple of more questions:
(1) What should CWND and ERTT be for first part of assignment 2
(2) How do we calculate CWND?

Dhawal Kumar

On 30 Jan 2002, Matthew Jonathan Holliman wrote:

> "Jonathan Keljo"  writes:
>
>
> >"Dhawal Kumar"  wrote in message
> 
> >> Since we can assume in first part of assignment 2 that packets are not
> >> lost, do we need to worry about what an ACK packet has to tell? And do we
> >> need to worry about send and recv windows?
>
> >The test script checks that you enforce the window sizes, so you definitely
> >do have to worry about that. I suspect it also checks that your ACK packets
> >are well-formed. I think the point of the assumption is that you don't have
> >to do anything about an ACK packet that you were expecting to receive but
> >never did. If you hack together some one-off thing that meets the
> >requirements of 2.A rather than designing something that's better suited to
> >the overall assignment, you're gonna really be hurting when you try to do
> >2.B.
>
>
> This is a very good point.  I'd *really* encourage people to think of this
> as a large assignment, with the real goal being the implementation of a
> completely functional TCP-like transport layer.  Treating it as three
> separate assignments, and planning to just add on functionality for the
> next milestone, will be very difficult.  It's much easier to start out
> with a design that seems reasonable and will still be useful for the
> next milestone--i.e. even though packet losses and retransmissions aren't
> covered in the first milestone, you should still be thinking about these
> as you decide how to implement even the basics of the protocol.
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Network Initiation and reliability
Date: 30 Jan 2002 09:25:26 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2853

Derrick Wen-Shiuan Tong  writes:


>The first bullet point under "Network Initiation" states that the
>active end sends an SYN and then waits for a SYN_ACK.

>If we send the SYN using network_send, which is unreliable,
>how do we know that the SYN packet was not dropped? Or is it
>the case that we don't use network_send for the SYN handshake?


For the next milestone, you'll have to handle retransmission of SYNs if
you don't receive a SYN-ACK in time.  (While you don't need to implement
this for the first milestone, keep it in mind as you design your code).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: passing peer address!
Date: 30 Jan 2002 09:31:38 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2854

Madhup Gulati  writes:


>I have the peer address got from sockfd. But I am in
>transport_sock_io()
>function which is called from control loop. Here I don't have access to
>syn_sd[1] . So how do I send this peer address to the parent.

> I can declare a new address variable in the context structure which gets
>passed to all fuctions or add a new argument to all my fuctions so that
>peer address is stored and passed. Both are not good techniques...


Well, including connection context (e.g. the syn_sd descriptor) in the
connection context structure seems a reasonable technique to me...
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Submission Error
Date: 30 Jan 2002 09:38:29 GMT
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2855

"Russell Greene"  writes:

>Hi All,

>     I cannot submit my project.  When I run the command in the assignment

>/afs/ir/class/cs244a/bin/submit.pl hw2.X ta I get a permission denied.

>Immediately below (in the handout) it has the line

>/afs/ir/class/cs244a/bin/submit ta hw2.X

>I run that and it tells me that the ta should go after the assignment.  I
>make the fix and run the command and it then tells me that hw2.A is not a
>valid assignment for this class.

>What am I doing wrong?

>--Russ


Probably nothing--someone had edited the submit configuration, and HW2 is
commented out at the moment.  Just hold off on submitting for the moment;
someone will post on here (hopefully soon) when it's fixed again.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Processing ACKS etc.
Date: 30 Jan 2002 09:49:16 GMT
Lines: 30
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2856

Dhawal Kumar  writes:

>If our sender window is full and the application gives more data to
>transmit, shoud we buffer it (if yes to what extent - I mean is there an
>upper limit) or should we not read the data until the sender window clears
>up - in this case select on data_sd will immediately return and we'll not
>read again because there is no space in window and this would keep happening -
>effectively a busy waiting.

>This can happen even in first part of assigment 2 if sender gives more
>that 3072 bytes to send.

That's your decision--whatever makes sense.

However, if you decide not to read data if the sender window is full, then
busy waiting is *not* appropriate.  (Note that select() or poll() will only
immediately return if you'd asked them to wait on the application data
socket).

>A couple of more questions:
>(1) What should CWND and ERTT be for first part of assignment 2

For the ERTT, see the timeout discussion in the assignment.

>(2) How do we calculate CWND?

You don't need to; see the comments in
print_congestion_window_and_estimatedRTT().
It's a historical artifact from a time when the assignment was even more
interesting :-)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!mgulati
From: Madhup Gulati 
Newsgroups: su.class.cs244a
Subject: How many windows?
Date: Wed, 30 Jan 2002 04:29:12 -0800
Lines: 30
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2857

Hi..
Since the active and passive sides both can send data and acks.....will we
have a sender and receiver window on each of the sides.

If so then we can have a situation where the child is trying to write data
to the  parent but parent still has more data to send. So writing fails.
If parent keeps sending data, the child will never get to write.

Similar problem can occur for the sockfd between passive and active
entities when both try to write.

Thanks,
Madhup
-------------------------------------------------------------------------------
Madhup Gulati
Graduate Student              Home:
Computer Science Dept.        141 K
William Gates Building        Escondido Road
Stanford University           Stanford
CA 94305                      CA 94305

          Ph: 650-497-4701

-------------------------------------------------------------------------------






.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Jian Deng 
Newsgroups: su.class.cs244a
Subject: Test script down
Date: 30 Jan 2002 18:02:38 GMT
Lines: 3
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga2.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2858

Is the test script down?  I've submited tests but never got executed.
The las time I submit was 7 hours ago.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: About windowing, mostly.
Date: Wed, 30 Jan 2002 10:37:51 -0800
Lines: 35
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2859

Some questions for you kind folks out there. Responses to any part(s) would
be greatly appreciated.

1. What happens when the child receives a contiguous block of data (that is,
all holes filled in)?
Do we then write it into the local data socket?
And does this block count as part of our receiving window?
If these things are true, can we and do we try to track how many bytes have
been written into the socket but not read by the parent? It seems it would
be necessary to do that.

2. To make sure I understand ACKing and Go-back-N, please correct me if I am
wrong:

Suppose the receiver has never received a block starting with seq number
500. If it receives a bunch of blocks after that, but still not 500, it just
sends an ACK 500 every time. Now suppose the transmission of block 500
finally times out. Then since the sender never received the ACK for 500, it
will pretend it never sent anything past 500 and start over from there (eg.
500, 550, 600 all over again).

Furthermore, these resent packets must all be divided as they were before.
If we had sent 500-549, 550-599 earlier we must send 500-549, 550-599
again--we cannot repackage and send 500-599.

All correct?

3. Can an unlimited amount of data be written into a socket (a real one, not
mysocket) without the other side's reading it?

4. Where can documentation for select() be found?  What does it return if
both sockets have data that need to be processed? I imagine that there's
most always data that the application is waiting to have sent.


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine38.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: parent: reading the peer address
Date: Wed, 30 Jan 2002 10:52:53 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine38.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2860


I always get this error

"parent: reading the peer address on a syn handshake (read only 0 bytes)"

How do we check that we got the peeraddress on the socket?

any clues what I am not doing?

thanks
-anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth5.Stanford.EDU!cpjones
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: Test script down
Date: Wed, 30 Jan 2002 13:42:25 -0800
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2861

I am also having this problem.

Thanks
Clayton

On 30 Jan 2002, Jian Deng wrote:

> Is the test script down?  I've submited tests but never got executed.
> The las time I submit was 7 hours ago.
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine41.Stanford.EDU!brandon
From: Brandon Badger 
Newsgroups: su.class.cs244a
Subject: Getting Same Error -- parent: reading the peer address
Date: Wed, 30 Jan 2002 13:57:37 -0800
Lines: 31
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine41.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2862

I'm getting the same error when I run the client:
"parent: reading the peer address on syn handshake (read only 0
bytes)recv: Error 0"

Here's where I'm trying to initialize the connection:

if(is_passive == 0){
	/* I'm trying to send the SYN packet for network initialization */
	/* I create an STCP packet. In this case there is no data */

	hdr = malloc(sizeof(STCPHeader));
	hdr->tcp_hdr.th_seq = 1;
	hdr->data_len = 0;

	/* Then I try to send the packet to the passive server end */

	network_send(ctx->sockfd, hdr, sizeof(STCPHeader), NULL);
}


When I run the client, I get the "parent: reading" error mentioned above.
Am I not sending the SYN packet correctly?

Also, should I have an -- if(is_passive == 1) section of code that reads
on sockfd, waiting for the client to send the initial handshake message,
and then in this section of code send the SYN_ACK back to the client?

Thanks for any clarification.

-Brandon

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine32.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Test script down
Date: Wed, 30 Jan 2002 14:22:46 -0800
Lines: 18
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine32.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2863

Count Me also.....

On Wed, 30 Jan 2002, Clayton Pierce Jones wrote:

> I am also having this problem.
>
> Thanks
> Clayton
>
> On 30 Jan 2002, Jian Deng wrote:
>
> > Is the test script down?  I've submited tests but never got executed.
> > The las time I submit was 7 hours ago.
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine29.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: How many windows?
Date: 30 Jan 2002 23:14:48 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine29.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2864

Madhup Gulati  writes:

>Hi..
>Since the active and passive sides both can send data and acks.....will we
>have a sender and receiver window on each of the sides.

>If so then we can have a situation where the child is trying to write data
>to the  parent but parent still has more data to send. So writing fails.
>If parent keeps sending data, the child will never get to write.

Yes, once the connection is established, both sides are symmetric, so both
will maintain sender and receiver windows.

If you receive data from the peer, you can write it to the application
and assume that it will be read; i.e. you don't need to worry about deadlock
situations in the application layer as you describe.

>Similar problem can occur for the sockfd between passive and active
>entities when both try to write.

I'm not sure that I follow you.  Both sides are STCP-compliant, and so will
not try to write if the window is full, so you shouldn't encounter a problem.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: About windowing, mostly.
Date: 30 Jan 2002 23:55:44 GMT
Lines: 59
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2865


>1. What happens when the child receives a contiguous block of data (that is,
>all holes filled in)?
>Do we then write it into the local data socket?
>And does this block count as part of our receiving window?
>If these things are true, can we and do we try to track how many bytes have
>been written into the socket but not read by the parent? It seems it would
>be necessary to do that.

Yes, yes, and no, respectively.  Once you write to the parent's socket,
you can't (as far as I know) tell how much has actually been read.  So
for the purpose of this assignment, you can just assume that all has
been read by the application.


>2. To make sure I understand ACKing and Go-back-N, please correct me if I am
>wrong:

>Suppose the receiver has never received a block starting with seq number
>500. If it receives a bunch of blocks after that, but still not 500, it just
>sends an ACK 500 every time. Now suppose the transmission of block 500
>finally times out. Then since the sender never received the ACK for 500, it
>will pretend it never sent anything past 500 and start over from there (eg.
>500, 550, 600 all over again).

Yep, that's right.


>Furthermore, these resent packets must all be divided as they were before.
>If we had sent 500-549, 550-599 earlier we must send 500-549, 550-599
>again--we cannot repackage and send 500-599.

From my interpretation of the STCP "specification," a compliant receiver
should be able to handle both cases.  For the purposes of calculating
timeouts for segments, the first is more accurate.  (I can't see a
restriction in the assignment specification concerning how you package
retransmissions).


>3. Can an unlimited amount of data be written into a socket (a real one, not
>mysocket) without the other side's reading it?

No, but for practical purposes you can assume this for the current
assignment's communication with the parent.  (See setsockopt in transport.c).


>4. Where can documentation for select() be found?  What does it return if
>both sockets have data that need to be processed? I imagine that there's
>most always data that the application is waiting to have sent.

man -s 2 poll
man -s 3c select

I imagine Stevens discusses these calls.  If/when one or more sockets have
data for processing (either read, write, or error condition, depending
on the flags you've set), the call will return.  (If you don't want the
call to wake up on a particular condition for a particular socket, don't
tag that condition as interesting for that descriptor).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Getting Same Error -- parent: reading the peer address
Date: 31 Jan 2002 00:00:12 GMT
Lines: 34
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2866

Brandon Badger  writes:

>I'm getting the same error when I run the client:
>"parent: reading the peer address on syn handshake (read only 0
>bytes)recv: Error 0"

>Here's where I'm trying to initialize the connection:

>if(is_passive == 0){
>	/* I'm trying to send the SYN packet for network initialization */
>	/* I create an STCP packet. In this case there is no data */

>	hdr = malloc(sizeof(STCPHeader));
>	hdr->tcp_hdr.th_seq = 1;
>	hdr->data_len = 0;

>	/* Then I try to send the packet to the passive server end */

>	network_send(ctx->sockfd, hdr, sizeof(STCPHeader), NULL);
>}


>When I run the client, I get the "parent: reading" error mentioned above.
>Am I not sending the SYN packet correctly?

Where do you set the SYN flag in the header?


>Also, should I have an -- if(is_passive == 1) section of code that reads
>on sockfd, waiting for the client to send the initial handshake message,
>and then in this section of code send the SYN_ACK back to the client?

Yes, that's correct.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: select(....) not working
Date: Wed, 30 Jan 2002 16:05:18 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2867

hi,

For the first time, when my client sends a SYN packet to server,
select(...) works. It works even at the server end.

But after that, when client.c calls write(...) which
writes to local_data_sd, select seems to be blocked. So, I am never able
to get to call transport_application_io(..).

any clues?

thanks
-anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: select(....) not working
Date: Wed, 30 Jan 2002 16:55:27 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Anurag Gupta 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2868

Hi,
When u return from select the filefds get reset to 0 except for the ones
on which there was some event. So i guess that could be ur problem. So try
to look at that part and restore the filefds before going to select.
Regards
Shankar

Wed, 30 Jan 2002, Anurag Gupta wrote:

> hi,
>
> For the first time, when my client sends a SYN packet to server,
> select(...) works. It works even at the server end.
>
> But after that, when client.c calls write(...) which
> writes to local_data_sd, select seems to be blocked. So, I am never able
> to get to call transport_application_io(..).
>
> any clues?
>
> thanks
> -anurag
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth3.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Test script down
Date: Wed, 30 Jan 2002 18:10:28 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: myth3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2869


I submitted earlier today, and still haven't got results...

On Wed, 30 Jan 2002, Shankar Agarwal wrote:

> Count Me also.....
>
> On Wed, 30 Jan 2002, Clayton Pierce Jones wrote:
>
> > I am also having this problem.
> >
> > Thanks
> > Clayton
> >
> > On 30 Jan 2002, Jian Deng wrote:
> >
> > > Is the test script down?  I've submited tests but never got executed.
> > > The las time I submit was 7 hours ago.
> > >
> > >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: mywrite does not block--buffer overflow?
Date: 31 Jan 2002 03:15:26 GMT
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2870

From the implementation of mywrite in mysock.c, it seems that it
does not block while our socket is sending the data. Doesn't this
introduce the possibility that it can overflow our send buffer, 
however large it is?

For instance, if we have a sending buffer of 3072 (or any size)
that's almost full, but we're still waiting for the ACK from the
receiver, what happens if our mysock client requests to send
more bytes that are left in our buffer? Can we get the parent
process to block somehow while we are sending and verifying
the transmission? I assume that this is what a Berkeley SOCK_STREAM
socket does when you perform a write(...) on it.

Thanks,

Derrick

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: void  print_congestion_window_and_estimatedRTT(int fid)   Question
Date: Wed, 30 Jan 2002 19:32:07 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2871

Do we need to implement void  print_congestion_window_and_estimatedRTT(int
fid) ?  I ask because from what I can tell no other parts of the program are
calling it (I did a grep).

If so, does it need to be done for part A?  I wouldn't think so... RTT
doesn't apply until part B.  What exactly is the congestion window?

thanks,
Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Hasan Taufiq Imam 
Newsgroups: su.class.cs244a
Subject: Test script down.
Date: Wed, 30 Jan 2002 20:25:05 -0800
Lines: 5
Distribution: su
Message-ID: 
Reply-To: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2872

Is the test script down or there is just a long queue. I have submitted
mine couple of hours ago and I have not yet received my confirmation
yet. There are few other postings regarding the test script being down.
Thanks,
Hasan
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: Re: Test script down.
Date: Wed, 30 Jan 2002 21:03:30 -0800
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2873

It's down, and Nick is working on it.

Pete

"Hasan Taufiq Imam"  wrote in message

> Is the test script down or there is just a long queue. I have submitted
> mine couple of hours ago and I have not yet received my confirmation
> yet. There are few other postings regarding the test script being down.
> Thanks,
> Hasan


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Ron Luman II" 
Newsgroups: su.class.cs244a
Subject: Control Loop and Multiple Child Processes
Date: Wed, 30 Jan 2002 21:52:45 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
X-Trace: news.Stanford.EDU 1012456383 14976 171.64.66.201 (31 Jan 2002 05:53:03 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2874

Hi all,

I'm slightly confused as to how things should work when there are multiple
entries in the context array (and therefore multiple child processes.)   It
seems to me that a single child process could handle all of the transport
layer processing, yet we seem to fork a child process for each connection.
While this eases the communications, it seems to me that it complicates the
control loop...

Enough rambling; here's my question:

Should we be coding the control loop so that it only 'selects' the file
descriptors associated with this particular (child) process?    (I don't see
another way, but the skeleton code seems to imply otherwise.)

Thanks,

--Ron


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: Initial SYN Send
Date: Wed, 30 Jan 2002 22:27:18 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2875

Interesting behavior:

When my client sends out the SYN packet from ctx->sockfd on the very first
call (22 bytes, successful), select immediately begins to register the
presence of data to be read on ctx->sockfd. [This is true because I skip the
preceding code if I am passive, and select does the blocking/timeout in that
case.] This is weird because I have not implemented the SYN_ACK on the other
side. An attempt to investigate this mysterious data through network_recv
produces a -1.

Anyone encounter anything similar? I am correctly resetting the socket set
before it heads into the select statement.

Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Test script down.
Date: Wed, 30 Jan 2002 22:30:23 -0800
Lines: 16
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2876

Yes we are and we will post here once it is up again.

  Guido

> It's down, and Nick is working on it.
> 
> Pete
> 
> "Hasan Taufiq Imam"  wrote in message
> 
>> Is the test script down or there is just a long queue. I have submitted
>> mine couple of hours ago and I have not yet received my confirmation
>> yet. There are few other postings regarding the test script being down.
>> Thanks,
>> Hasan

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga8.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Network Initiation
Date: Wed, 30 Jan 2002 22:40:15 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2877

Hi,

I just want to make sure that what I will say next is correct:

When the server receives a SYN packet, it will send back two packets,
first has both the SYN and ACK flags and server sequence number. The
second packet is just an Ack to the client for its SYN packet, with the
client_seq_num+1.

This is how I understand the handout. Am I correct?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Debashis Sahoo" 
Newsgroups: su.class.cs244a
Subject: failed to compile ...
Date: Wed, 30 Jan 2002 22:42:33 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: dsahoo.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2878

Hi,
      I got following error message.
I don't know what it means.


    * Test 1.A: Fails to compile
  [out of 20]

[SUMMARY] NOT OK
[RESULT]  ERROR:  No testserver.sample in
/afs/ir/class/cs244a/submissions/restricted/hw2.A/holliman/sahoo//compile.su
n.

Thanks,
Sahoo


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Jian Deng 
Newsgroups: su.class.cs244a
Subject: Re: failed to compile ...
Date: 31 Jan 2002 06:47:03 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic5.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2879

I got the same error too.

Debashis Sahoo  wrote:
> Hi,
>       I got following error message.
> I don't know what it means.


>     * Test 1.A: Fails to compile
>   [out of 20]

> [SUMMARY] NOT OK
> [RESULT]  ERROR:  No testserver.sample in
> /afs/ir/class/cs244a/submissions/restricted/hw2.A/holliman/sahoo//compile.su
> n.

> Thanks,
> Sahoo


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Re: Network Initiation
Date: Wed, 30 Jan 2002 22:51:10 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2880

My impression of the specification is that th_flags would be SYN | ACK,
th_ack would be client_seq_num + 1, and th_seq would be server_seq_num (all
in one packet).  I have not run my code against theirs yet, so I may be
wrong.

Bret


"Moh'd Saleem Saleem Alrawashdeh"  wrote in message

> Hi,
>
> I just want to make sure that what I will say next is correct:
>
> When the server receives a SYN packet, it will send back two packets,
> first has both the SYN and ACK flags and server sequence number. The
> second packet is just an Ack to the client for its SYN packet, with the
> client_seq_num+1.
>
> This is how I understand the handout. Am I correct?
>
> Thanks,
>
> Moh'd
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Network Initiation
Date: Wed, 30 Jan 2002 22:53:38 -0800
Lines: 37
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2882

Hi..
I think Bret is right...
I'm passing that part of the test, so I think I'm sure...

"Bret Taylor"  wrote in message

> My impression of the specification is that th_flags would be SYN | ACK,
> th_ack would be client_seq_num + 1, and th_seq would be server_seq_num
(all
> in one packet).  I have not run my code against theirs yet, so I may be
> wrong.
>
> Bret
>
>
> "Moh'd Saleem Saleem Alrawashdeh"  wrote in message
> 
> > Hi,
> >
> > I just want to make sure that what I will say next is correct:
> >
> > When the server receives a SYN packet, it will send back two packets,
> > first has both the SYN and ACK flags and server sequence number. The
> > second packet is just an Ack to the client for its SYN packet, with the
> > client_seq_num+1.
> >
> > This is how I understand the handout. Am I correct?
> >
> > Thanks,
> >
> > Moh'd
> >
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: getpeername vs. network_peer_name
Date: Wed, 30 Jan 2002 22:54:23 -0800
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2883

Luis,

yes, you will have to always use network_peer. The problem is
not UDP sockets but how the library uses them.

  Guido

> Are we supposed to always use network_peer_name to
> get the peer struct to pass up to the parent?
> 
> I was using getpeername for the active connection and that
> worked, but I had to switch to network_peer_name on
> the passive connection because getpeername was giving
> me a the following error:
> 
> ": Transport endpoint is not connected"
> 
> Is this due to the underlying usage of UDP sockets, or
> something else?
> 
> Thanks

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: failed to compile ...
Date: Wed, 30 Jan 2002 22:58:05 -0800
Lines: 27
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2884

Everyone,

these compile errors are due to us restarting the testing server. Ignore 
the results, this won't count as one of your submission attempts.

  Guido

> I got the same error too.
> 
> Debashis Sahoo  wrote:
>> Hi,
>>       I got following error message.
>> I don't know what it means.
> 
> 
>>     * Test 1.A: Fails to compile
>>   [out of 20]
> 
>> [SUMMARY] NOT OK
>> [RESULT]  ERROR:  No testserver.sample in
>> 
/afs/ir/class/cs244a/submissions/restricted/hw2.A/holliman/sahoo//compile.su
>> n.
> 
>> Thanks,
>> Sahoo

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: script results
Date: Wed, 30 Jan 2002 23:08:04 -0800
Lines: 55
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2885

Hey folks,

Thankfully the script is running again...

I'm getting the following results even though my client
and server seem quite happy to exchange files, even
really large ones when I run them...

===========================
===(T1.Ba) Byte count differences found (INCORRECT!)
Student's Tx against our Rx (reliable mode)

---

---
Difference between expected result and your result:
1c1,3
<       0       0       0 TESTDATA.0
---
>    2666   15720  110230 TESTDATA.0
>    1812   10049  299328 TESTDATA.1
>     131     501    5098 TESTDATA.2

---

===========================

--

and

===========================
===(T1.Bb) Checksum differences found (INCORRECT!)
Student's Tx against our Rx (reliable mode)

---

---
Difference between expected result and your result:
1c1,3
< 4294967295 0 TESTDATA.0
---
> 1705403052 110230 TESTDATA.0
> 4265638025 299328 TESTDATA.1
> 2181702833 5098 TESTDATA.2

---

What is my Tx and their Rx? And what do these printouts mean
(are the numbers file sizes and the names filenames) ?

Also is it that they expected one file called TESTDATA.0 and
my code resulted in three files?

Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Server back up again, Tests today are not counted.
Date: Wed, 30 Jan 2002 23:22:17 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2886

Hi everyone,

as some of you have already noticed, the testing server is back up again. 
Sorry for any inconvenience it might have caused.

As some test this evening returned bogus or no results, we have decided 
that any tests submitted today (We) will not be counted in the testing 
limit.

Happy coding!

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Disk full again
Date: Wed, 30 Jan 2002 23:55:44 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2887


Hi...

I cannot submit_test my code now, because of "no space left on device"..
I'm submitting to  akishan..



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: parent: reading the peer address
Date: Wed, 30 Jan 2002 23:52:34 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2888

Anurag,

this looks strange. The error message looks like transport.c after 
splitting off a new process for the client, the server part of transport.c 
is not able to read the socket infromation from this client process. It 
actually gets woken up on select but only gets zero bytes. Can you provide 
any more hints, code fragments etc. on what you are doing? What does your 
code do after the fork()?

  Guido


> I always get this error
> 
> "parent: reading the peer address on a syn handshake (read only 0 bytes)"
> 
> How do we check that we got the peeraddress on the socket?
> 
> any clues what I am not doing?
> 
> thanks
> -anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!talon2.Stanford.EDU!tortuga
From: 
Newsgroups: su.class.cs244a
Subject: Connection termination
Date: Thu, 31 Jan 2002 00:05:48 -0800
Lines: 8
Sender: 
Distribution: su
Message-ID: 
NNTP-Posting-Host: talon2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2889

If I send a FIN, should I just ignore all packets that are not a FIN or
FIN+ACK. In normal TCP I'd still be required to receive and acknowledge
data packets but that is because both sides need to agree on the
termination.

Thanks,
Josh

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: mywrite does not block--buffer overflow?
Date: Thu, 31 Jan 2002 00:14:59 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2890

Derrick,

I assume you are using the transport_appl_io callback. You don't have to 
read anything from the context in the socket. If you simply return you 
haven't accepted an additional data. The parent process will eventually 
block but you actually don't have to worry about that.

  Guido 

> From the implementation of mywrite in mysock.c, it seems that it
> does not block while our socket is sending the data. Doesn't this
> introduce the possibility that it can overflow our send buffer,
> however large it is?
> 
> For instance, if we have a sending buffer of 3072 (or any size)
> that's almost full, but we're still waiting for the ACK from the
> receiver, what happens if our mysock client requests to send
> more bytes that are left in our buffer? Can we get the parent
> process to block somehow while we are sending and verifying
> the transmission? I assume that this is what a Berkeley SOCK_STREAM
> socket does when you perform a write(...) on it.
> 
> Thanks,
> 
> Derrick

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: void  print_congestion_window_and_estimatedRTT(int fid)   Question
Date: Thu, 31 Jan 2002 00:18:28 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2891

Peter,

there is actually nothing to implement here. The function is complete as 
is. You should only update/use the global variables and that's it.

  Guido

> Do we need to implement void  print_congestion_window_and_estimatedRTT(int
> fid) ?  I ask because from what I can tell no other parts of the program
> are calling it (I did a grep).
> 
> If so, does it need to be done for part A?  I wouldn't think so... RTT
> doesn't apply until part B.  What exactly is the congestion window?
> 
> thanks,
> Pete

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: Random initial sequence number
Date: Thu, 31 Jan 2002 00:28:21 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2892

I noticed that network.c calls srand() with a constant value.  This means
that if we used rand() to produce our random initial sequence number, our
client and server will always produce the same random sequence number, among
other things.  Should we re-seed the RNG or just leave it the way it is?

Bret



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Disk full again
Date: Thu, 31 Jan 2002 00:19:52 -0800
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2893

Ooops, working on it.

> 
> Hi...
> 
> I cannot submit_test my code now, because of "no space left on device"..
> I'm submitting to  akishan..

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Disk full again
Date: Thu, 31 Jan 2002 00:23:52 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2894

It should be ok now, please try again and send me an email (not news) if 
you still have problems.

  Guido

> 
> Hi...
> 
> I cannot submit_test my code now, because of "no space left on device"..
> I'm submitting to  akishan..

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Random initial sequence number
Date: Thu, 31 Jan 2002 00:32:17 -0800
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2895

Bret Taylor wrote:

> I noticed that network.c calls srand() with a constant value.  This means
> that if we used rand() to produce our random initial sequence number, our
> client and server will always produce the same random sequence number,
> among
> other things. 

Correct.

> Should we re-seed the RNG or just leave it the way it is?

You should re-seed it. We don't expect anything cryptographically strong 
though, something like:

 srand(getpid());

in the right place will do the job.

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine38.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Packet size > STCPHeader size + 536
Date: Thu, 31 Jan 2002 00:52:03 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine38.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2896


Should we drop such packets or read off 536 bytes. If we try to read first
536 bytes the data_len field in header will not match the value obtained using
network_recv.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine5.Stanford.EDU!mgulati
From: Madhup Gulati 
Newsgroups: su.class.cs244a
Subject: Race condition!
Date: Thu, 31 Jan 2002 01:01:22 -0800
Lines: 47
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2897


Hi all,

I am stuck!!

As soon as the parent gets the peer address it puts data on the data
socket for child to read.

In my program,in case of the active side the child sends the peer address
to parent after recieving SYN_ACK but for the passive side I don't know when the
child should send the peer address?

When I send it at the start in transport_init in if(is_passive != 0) the
parent at once gives the data.....the child instead of reading SYN from
network goes to read the parent!!

So depending on which of the active of passive sides gets the address
first the data transfer starts from that end....

Where do I put the code for the child to send peer address to parent??

Thanks,

Madhup






-------------------------------------------------------------------------------
Madhup Gulati
Graduate Student              Home:
Computer Science Dept.        141 K
William Gates Building        Escondido Road
Stanford University           Stanford
CA 94305                      CA 94305

          Ph: 650-497-4701

-------------------------------------------------------------------------------






.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: Compling error?
Date: Thu, 31 Jan 2002 01:24:57 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2898


Thanks a lot for help!

I submited to test script several times today, but keep on getting
compling error.

I am quite confident they are complied well in leland cluster.
Is there anything wrong with test script?

This is what I get from test script, but I am sure I define these
in transport.h
*************************************************************************
transport.c:52: `MAX_PAYLOAD_SIZE' undeclared here (not in a function)
transport.c:125: `MAX_NUM_SOCKFDS' undeclared here (not in a function)
transport.c: In function `get_glob_ctx_local_data_pipe':
transport.c:170: `MAX_NUM_SOCKFDS' undeclared (first use in this function)

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine19.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Compling error?
Date: Thu, 31 Jan 2002 01:47:20 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine19.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Jinhui Pan 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2899



According to my understanding,
we are not supposed to modify transport.h

so you should have #defines in transport.c itself.

thanks, Arvind

On Thu, 31 Jan 2002, Jinhui Pan wrote:

>
> Thanks a lot for help!
>
> I submited to test script several times today, but keep on getting
> compling error.
>
> I am quite confident they are complied well in leland cluster.
> Is there anything wrong with test script?
>
> This is what I get from test script, but I am sure I define these
> in transport.h
> *************************************************************************
> transport.c:52: `MAX_PAYLOAD_SIZE' undeclared here (not in a function)
> transport.c:125: `MAX_NUM_SOCKFDS' undeclared here (not in a function)
> transport.c: In function `get_glob_ctx_local_data_pipe':
> transport.c:170: `MAX_NUM_SOCKFDS' undeclared (first use in this function)
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: Re: Compling error?
Date: Thu, 31 Jan 2002 01:56:41 -0800
Lines: 44
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2900


Thanks a lot!
I guess this is the reason. But the script does scratch our transport.h.
It is realy confusing that the script  gets transport.h  but does not
recogniz it...

Thanks!




On Thu, 31 Jan 2002, Arvind Bhat wrote:

>
>
> According to my understanding,
> we are not supposed to modify transport.h
>
> so you should have #defines in transport.c itself.
>
> thanks, Arvind
>
> On Thu, 31 Jan 2002, Jinhui Pan wrote:
>
> >
> > Thanks a lot for help!
> >
> > I submited to test script several times today, but keep on getting
> > compling error.
> >
> > I am quite confident they are complied well in leland cluster.
> > Is there anything wrong with test script?
> >
> > This is what I get from test script, but I am sure I define these
> > in transport.h
> > *************************************************************************
> > transport.c:52: `MAX_PAYLOAD_SIZE' undeclared here (not in a function)
> > transport.c:125: `MAX_NUM_SOCKFDS' undeclared here (not in a function)
> > transport.c: In function `get_glob_ctx_local_data_pipe':
> > transport.c:170: `MAX_NUM_SOCKFDS' undeclared (first use in this function)
> >
> >
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Packet size > STCPHeader size + 536
Date: Thu, 31 Jan 2002 02:27:05 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012472825 17155 128.12.133.48 (31 Jan 2002 10:27:05 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2901

Any transmitter that sends such packets is not compliant with the protocol.
So drop 'em.

Jonathan

"Dhawal Kumar"  wrote in message

>
> Should we drop such packets or read off 536 bytes. If we try to read first
> 536 bytes the data_len field in header will not match the value obtained
using
> network_recv.
>
> Dhawal Kumar
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: script results
Date: Thu, 31 Jan 2002 02:31:47 -0800
Lines: 75
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012473107 17190 128.12.133.48 (31 Jan 2002 10:31:47 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2902


"Luis Robles"  wrote in message

> Hey folks,
>
> Thankfully the script is running again...
>
> I'm getting the following results even though my client
> and server seem quite happy to exchange files, even
> really large ones when I run them...
>
> ===========================
> ===(T1.Ba) Byte count differences found (INCORRECT!)
> Student's Tx against our Rx (reliable mode)
>
> ---
>
> ---
> Difference between expected result and your result:
> 1c1,3
> <       0       0       0 TESTDATA.0
> ---
> >    2666   15720  110230 TESTDATA.0
> >    1812   10049  299328 TESTDATA.1
> >     131     501    5098 TESTDATA.2
>
> ---
>
> ===========================
>
> --
>
> and
>
> ===========================
> ===(T1.Bb) Checksum differences found (INCORRECT!)
> Student's Tx against our Rx (reliable mode)
>
> ---
>
> ---
> Difference between expected result and your result:
> 1c1,3
> < 4294967295 0 TESTDATA.0
> ---
> > 1705403052 110230 TESTDATA.0
> > 4265638025 299328 TESTDATA.1
> > 2181702833 5098 TESTDATA.2
>
> ---
>
> What is my Tx and their Rx? And what do these printouts mean
> (are the numbers file sizes and the names filenames) ?

As far as I know, from running the script a bunch and using my wonderful
powers of deductive reasoning...

Your TX is server.c compiled against your transport layer. Their RX is
client.c compiled against their transport layer. The output here is the
output of the cksum program (see man cksum) which is the checksum followed
by the file size. (The output for the size mismatch is the output of the wc
program, which is lines, words, bytes.) The top is your program's output,
the bottom is expected.

>
> Also is it that they expected one file called TESTDATA.0 and
> my code resulted in three files?

Man, wouldn't that mess with your head!? It'd be a helluva bug if your
program created files out of thin air...

>
> Thanks


.

Path: shelby.stanford.edu!nntp.stanford.edu!cardinal5.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: Question about connection termination
Date: Thu, 31 Jan 2002 02:40:16 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: cardinal5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2903

In the FAQ, the explaination about termination is:
"When you call myclose, it calls transport_close, which closes the
parent's end of application socket and waits for the child process to terminate.
When the child is doing a read on the localpipe, it should return a zero.
That is an indication to initiate the FIN exchange. "
But I am confused. Since we use "select" to know if network socket or the
localpipe has pending data, and then do the read. If the parent has
closed the localpipe, there must not have pending data in localpipe, then
select can't tell us to read the localpipe.
So what makes the child process to read on the localpipe and return a
zero, and then initiate the FIN? (the zero also indicates there is no
pending data in the localpipe)


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: Re: Question about connection termination
Date: Thu, 31 Jan 2002 03:30:34 -0800
Lines: 6
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2904

This is something I wanted to know as well.

-Steve



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bret Taylor" 
Newsgroups: su.class.cs244a
Subject: ACK packet fields
Date: Thu, 31 Jan 2002 04:04:35 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: roble-01-352a.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2905

From the assignment:

"Data is acknowledged by setting the ACK bit in the flags field in the
packet header. If this bit is set, then the ack_seq field contains the
sequence number of the next byte of data the receiver expects..."

There is no "ack_seq" field in the packet: is this the th_seq or th_ack
field?  It does not matter when both clients use my transport layer, but it
will when they interact with other implemenations.

Thanks,
Bret



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: purify UMRs
Date: Thu, 31 Jan 2002 05:26:31 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine26.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2906

When I run purify, I get Uninitialized Mem Reads on my STCP header
struct whenever i send use network_send(). My program runs fine, but
this troubles me. I don't see why I would be reading uninitialized
memory -- when I send my stcp-header with a call to network_send(),I
just pass in the address of the my stcpHdr struct, and use
sizeof(stcpHdr) as the length. I thought this might have something to do
with the fact that the struct is "packed"?

Does anyone have any suggestions?

Thanks,
Tim
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Michael Kahlil Madison 
Newsgroups: su.class.cs244a
Subject: Midterm discrepancy
Date: Thu, 31 Jan 2002 08:21:29 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga18.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2907

On Handout #1 it says our midterm is Feb 19th... however, on the website
calendar it says our midterm is Feb 14th.

Which is correct? and what about the final -- the web site says the 18th
is that correct?

-mike
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: windows
Date: Thu, 31 Jan 2002 08:31:22 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:2908

We count only data in our windows, right? That is, we don't include the
header byte count as part of the max number of bytes that can be
outstanding?

Thanks,
Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Submit script?
Date: Thu, 31 Jan 2002 08:34:48 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2909


Hi,

  I think the submit script might be broken.

elaine15:~/projects/hw2> /afs/ir/class/cs244a/bin/submit xwang00 hw2.A
Submit: "hw2.A" is not an assignment for this class.

Format: submit TA Assignment

TA is one of: hondroul appenz akishan xwang00 holliman

Assignment is one of: hw1
hw1             - FTP Directory Copy
elaine15:~/projects/hw2>


                                            ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth10.Stanford.EDU!cpjones
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: ACK packet fields
Date: Thu, 31 Jan 2002 09:43:55 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2910

For an ack packet, you would send the ack number in the th_ack field.  For
a syn, syn-ack, or data packet you would send the sequence number in the
th_seq field.

Clayton

On Thu, 31 Jan 2002, Bret Taylor wrote:

> From the assignment:
>
> "Data is acknowledged by setting the ACK bit in the flags field in the
> packet header. If this bit is set, then the ack_seq field contains the
> sequence number of the next byte of data the receiver expects..."
>
> There is no "ack_seq" field in the packet: is this the th_seq or th_ack
> field?  It does not matter when both clients use my transport layer, but it
> will when they interact with other implemenations.
>
> Thanks,
> Bret
>
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine32.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: fid in print_congestion_window.
Date: Thu, 31 Jan 2002 09:55:30 -0800
Lines: 29
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine32.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Guido Appenzeller 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2911

Hi,
This function takes fid as parameter. But it can be called from both
network.c and mysock.c. Since we have both the fds stored in our
connection structure what should we use this fid as. Is this local_data_sd
or is it sockfd.
Thanks
Shankar
 On Thu, 31 Jan 2002, Guido
Appenzeller wrote:

> Peter,
>
> there is actually nothing to implement here. The function is complete as
> is. You should only update/use the global variables and that's it.
>
>   Guido
>
> > Do we need to implement void  print_congestion_window_and_estimatedRTT(int
> > fid) ?  I ask because from what I can tell no other parts of the program
> > are calling it (I did a grep).
> >
> > If so, does it need to be done for part A?  I wouldn't think so... RTT
> > doesn't apply until part B.  What exactly is the congestion window?
> >
> > thanks,
> > Pete
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Race condition!
Date: Thu, 31 Jan 2002 10:07:13 -0800
Lines: 57
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2912


For passive side, child has to write to syn_sd only after receiving SYN

"Madhup Gulati"  wrote in message

>
> Hi all,
>
> I am stuck!!
>
> As soon as the parent gets the peer address it puts data on the data
> socket for child to read.
>
> In my program,in case of the active side the child sends the peer address
> to parent after recieving SYN_ACK but for the passive side I don't know
when the
> child should send the peer address?
>
> When I send it at the start in transport_init in if(is_passive != 0) the
> parent at once gives the data.....the child instead of reading SYN from
> network goes to read the parent!!
>
> So depending on which of the active of passive sides gets the address
> first the data transfer starts from that end....
>
> Where do I put the code for the child to send peer address to parent??
>
> Thanks,
>
> Madhup
>
>
>
>
>
>
> --------------------------------------------------------------------------
-----
> Madhup Gulati
> Graduate Student              Home:
> Computer Science Dept.        141 K
> William Gates Building        Escondido Road
> Stanford University           Stanford
> CA 94305                      CA 94305
>
>           Ph: 650-497-4701
>
> --------------------------------------------------------------------------
-----
>
>
>
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine42.Stanford.EDU!rxuan
From: Ruozhong Xuan 
Newsgroups: su.class.cs244a
Subject: Re: ACK packet fields
Date: Thu, 31 Jan 2002 10:12:53 -0800
Lines: 37
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine42.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Clayton Pierce Jones 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2913

Hi,
   Do we need to set the ACK flag bit on and set the ACK value when we are
sending a data packet (which contains data) ?

Thanks.
Ruozhong


On Thu, 31 Jan 2002, Clayton Pierce Jones wrote:

> For an ack packet, you would send the ack number in the th_ack field.  For
> a syn, syn-ack, or data packet you would send the sequence number in the
> th_seq field.
>
> Clayton
>
> On Thu, 31 Jan 2002, Bret Taylor wrote:
>
> > From the assignment:
> >
> > "Data is acknowledged by setting the ACK bit in the flags field in the
> > packet header. If this bit is set, then the ack_seq field contains the
> > sequence number of the next byte of data the receiver expects..."
> >
> > There is no "ack_seq" field in the packet: is this the th_seq or th_ack
> > field?  It does not matter when both clients use my transport layer, but it
> > will when they interact with other implemenations.
> >
> > Thanks,
> > Bret
> >
> >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth10.Stanford.EDU!cpjones
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: ACK packet fields
Date: Thu, 31 Jan 2002 10:15:42 -0800
Lines: 47
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: myth10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2914

An ack packet is sent to acknowledge that some data was actually
received... so it would not really make sense for a data packet to have an
ack flag (because it is not acknowledging anything... it is sending
something).

Clayton

On Thu, 31 Jan 2002, Ruozhong Xuan wrote:

> Hi,
>    Do we need to set the ACK flag bit on and set the ACK value when we are
> sending a data packet (which contains data) ?
>
> Thanks.
> Ruozhong
>
>
> On Thu, 31 Jan 2002, Clayton Pierce Jones wrote:
>
> > For an ack packet, you would send the ack number in the th_ack field.  For
> > a syn, syn-ack, or data packet you would send the sequence number in the
> > th_seq field.
> >
> > Clayton
> >
> > On Thu, 31 Jan 2002, Bret Taylor wrote:
> >
> > > From the assignment:
> > >
> > > "Data is acknowledged by setting the ACK bit in the flags field in the
> > > packet header. If this bit is set, then the ack_seq field contains the
> > > sequence number of the next byte of data the receiver expects..."
> > >
> > > There is no "ack_seq" field in the packet: is this the th_seq or th_ack
> > > field?  It does not matter when both clients use my transport layer, but it
> > > will when they interact with other implemenations.
> > >
> > > Thanks,
> > > Bret
> > >
> > >
> > >
> > >
> >
> >
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Race condition!
Date: Thu, 31 Jan 2002 10:11:36 -0800
Lines: 69
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2915


I was just expressing my opinion..
I was being so terse that I forgot to mention that..
Just hope that you're not stuck any more..

"BAEHOPIL"  wrote in message

>
> For passive side, child has to write to syn_sd only after receiving SYN
>
> "Madhup Gulati"  wrote in message
> 
> >
> > Hi all,
> >
> > I am stuck!!
> >
> > As soon as the parent gets the peer address it puts data on the data
> > socket for child to read.
> >
> > In my program,in case of the active side the child sends the peer
address
> > to parent after recieving SYN_ACK but for the passive side I don't know
> when the
> > child should send the peer address?
> >
> > When I send it at the start in transport_init in if(is_passive != 0) the
> > parent at once gives the data.....the child instead of reading SYN from
> > network goes to read the parent!!
> >
> > So depending on which of the active of passive sides gets the address
> > first the data transfer starts from that end....
> >
> > Where do I put the code for the child to send peer address to parent??
> >
> > Thanks,
> >
> > Madhup
> >
> >
> >
> >
> >
> >
>
> --------------------------------------------------------------------------
> -----
> > Madhup Gulati
> > Graduate Student              Home:
> > Computer Science Dept.        141 K
> > William Gates Building        Escondido Road
> > Stanford University           Stanford
> > CA 94305                      CA 94305
> >
> >           Ph: 650-497-4701
> >
>
> --------------------------------------------------------------------------
> -----
> >
> >
> >
> >
> >
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga14.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: Problem with test script
Date: Thu, 31 Jan 2002 10:53:24 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2916



  Hi,

  I think the test script have some problems again, haven't it?

  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Sriram Viji" 
Newsgroups: su.class.cs244a
Subject: network termination
Date: Thu, 31 Jan 2002 10:57:08 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: glaurung.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2615.200
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200
Xref: nntp.stanford.edu su.class.cs244a:2917


when a application closes its socket:
the child process sends out a FIN packet and waits for a FIN_ACK from the
peer

Q:  does it consider any of the packets received between sending the FIN and
receiving the FIN_ACK?

thanks,
sriram


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: script results
Date: Thu, 31 Jan 2002 11:36:00 -0800
Lines: 83
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2918

Took me a while to figure this one out, too.  My clue however was that 
when I ran large files between my client and server on hosts that are 
sufficiently distant from one another (elaine7 and myth7 for example), 
repeated runs would make the allegedly reliable UDP drop packets.  I put 
some delays in my code to not cause so much congestion and that cleared 
things up.

Scary stuff when you really can't depend on a reliable underlying transport.

-Sandeep

Luis Robles wrote:

> Hey folks,
> 
> Thankfully the script is running again...
> 
> I'm getting the following results even though my client
> and server seem quite happy to exchange files, even
> really large ones when I run them...
> 
> ===========================
> ===(T1.Ba) Byte count differences found (INCORRECT!)
> Student's Tx against our Rx (reliable mode)
> 
> ---
> 
> ---
> Difference between expected result and your result:
> 1c1,3
> <       0       0       0 TESTDATA.0
> ---
> 
>>   2666   15720  110230 TESTDATA.0
>>   1812   10049  299328 TESTDATA.1
>>    131     501    5098 TESTDATA.2
>>
> 
> ---
> 
> ===========================
> 
> --
> 
> and
> 
> ===========================
> ===(T1.Bb) Checksum differences found (INCORRECT!)
> Student's Tx against our Rx (reliable mode)
> 
> ---
> 
> ---
> Difference between expected result and your result:
> 1c1,3
> < 4294967295 0 TESTDATA.0
> ---
> 
>>1705403052 110230 TESTDATA.0
>>4265638025 299328 TESTDATA.1
>>2181702833 5098 TESTDATA.2
>>
> 
> ---
> 
> What is my Tx and their Rx? And what do these printouts mean
> (are the numbers file sizes and the names filenames) ?
> 
> Also is it that they expected one file called TESTDATA.0 and
> my code resulted in three files?
> 
> Thanks
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Disk full again
Date: Thu, 31 Jan 2002 11:37:26 -0800
Lines: 29
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
To: Guido Appenzeller 
Xref: nntp.stanford.edu su.class.cs244a:2919

This is still a problem.  And the submit script is broken, too.  Still 
complains about hw2.A being an invalid project.

-Sandeep

Guido Appenzeller wrote:

> It should be ok now, please try again and send me an email (not news) if 
> you still have problems.
> 
>   Guido
> 
> 
>>Hi...
>>
>>I cannot submit_test my code now, because of "no space left on device"..
>>I'm submitting to  akishan..
>>
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Question about connection termination
Date: Thu, 31 Jan 2002 11:41:29 -0800
Lines: 32
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2920

When the parent closes the pipe, your child's select WILL trigger.  And 
you'll do the read, see the 0 bytes, etc...

-Sandeep

Fang Sun wrote:

> In the FAQ, the explaination about termination is:
> "When you call myclose, it calls transport_close, which closes the
> parent's end of application socket and waits for the child process to terminate.
> When the child is doing a read on the localpipe, it should return a zero.
> That is an indication to initiate the FIN exchange. "
> But I am confused. Since we use "select" to know if network socket or the
> localpipe has pending data, and then do the read. If the parent has
> closed the localpipe, there must not have pending data in localpipe, then
> select can't tell us to read the localpipe.
> So what makes the child process to read on the localpipe and return a
> zero, and then initiate the FIN? (the zero also indicates there is no
> pending data in the localpipe)
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth3.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Optimizations on transmission
Date: 31 Jan 2002 20:38:18 GMT
Organization: Stanford University, CA 94305, USA
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth3.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2921

The program sheet says under Data Packets that STCP performs no
optimizations for grouping small amounts of data on the first
transmission. I've noticed that (without trying) my code does this. So

1)
should I add code to try to prevent this.
2)
as my code does not work with the script client/server at all, but work
fine together could this be why? (I am running out of ideas)


Thank you,

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!brandon
From: Brandon Badger 
Newsgroups: su.class.cs244a
Subject: Error on network_send
Date: Thu, 31 Jan 2002 12:52:44 -0800
Lines: 29
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2922

I'm getting stuck on an error where network_send() returns -1 when the
server tries to send the contents of the requested file to the client.

1) The handshake works correctly.
2) transport_appl_io() is called in the client and the filename is passed
to the client's child.  This sends the filename on sockfd.
3) transport_sock_io() is called in the server and the filename is passed
to the server's parent.
4) transport_appl_io() is called in the server and the contents of the
file are passed to the server's child.  At this point I'm able to print
out the contents of the file.  - But, when I try to send the file contents
through sockfd, the network_send function returns -1;

any ideas?  What are some causes for network_send to return -1?

In my transport_appl_io() function, the network_send code works correctly
for the client side when it sends the filename to the server.

stcpPacket = malloc(packetSize);
memcpy(stcpPacket, hdrPtr, sizeof(STCPHeader));
memcpy(stcpPacket + sizeof(STCPHeader), msg + dataStart, len);
/* This printf of the file contents works */
/* printf("\nMSG:%s", stcpPacket + sizeof(STCPHeader)); */
nSent = network_send(ctx->sockfd, stcpPacket, packetSize, NULL);

Thanks for any help,
Brandon


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic1.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: purify UMRs
Date: Thu, 31 Jan 2002 13:06:41 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Timothy Tay Chao 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2923


Hi Tim,

Are you initializing (bzero/memset) the stcp header before
u send it out ? It is possible that you set some of the
fields in struct like the th_seq, but leave rest of it
uninitialized, causing them to have unknown values.

hth,

Thanks, Arvind


On Thu, 31 Jan 2002, Timothy Tay Chao wrote:

> When I run purify, I get Uninitialized Mem Reads on my STCP header
> struct whenever i send use network_send(). My program runs fine, but
> this troubles me. I don't see why I would be reading uninitialized
> memory -- when I send my stcp-header with a call to network_send(),I
> just pass in the address of the my stcpHdr struct, and use
> sizeof(stcpHdr) as the length. I thought this might have something to do
> with the fact that the struct is "packed"?
>
> Does anyone have any suggestions?
>
> Thanks,
> Tim
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Regarding target address when using network_send
Date: Thu, 31 Jan 2002 13:52:12 -0800
Lines: 40
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2925


I wonder about how network_send() figures out the destination address..

For active side, it doesn't have to worry because we have already registered
the destination address using connect().
But for passive side, it seems to get the destination address from
global_net_ctx->fromaddr.

  if (global_net_ctx->is_passive)
    {
      mh.msg_name = (char*) &(global_net_ctx->fromaddr);
      mh.msg_namelen = global_net_ctx->fromlen;
    }

But global_net_ctx->fromaddr is set in network_recv()..

  len = recvfrom(sockfd, recvBuf, recvLen, 0,
                &(global_net_ctx->fromaddr), &(global_net_ctx->fromlen));

It seems to me that fromaddr is set to the source address of the received
packet..

Then, it seems to me that network_send() is sending to an IP address from
which
we received the last packet, not to an IP address from which we initially
accepted
connection.
It won't be a problem if there is only one client at a time,
but if another should try to connect while one is connected,
I think everything will be messed up..

My code seems to be working without this consideration,
but I really do wonder about what it's intended to be like..

And, when I used connect() at the passive side,
network_send() no longer worked.. Is there any reason we should not
use connect() on a UDP socket we already used bind() on?



.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: compile error in test script
Date: 31 Jan 2002 21:58:30 GMT
Organization: Stanford University, CA 94305, USA
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2926



Hi, I am getting this error message

    * Test 1.A: Fails to compile
  [out of 20]
 
[SUMMARY] NOT OK
[RESULT]  ERROR:  No server.sample in /afs/ir/class/cs244a/submissions/restricted/hw2.A/holliman/mdolan//compile.sun.


any idea?

Thanks


-mark


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Testing script/Submit script...
Date: Thu, 31 Jan 2002 14:06:36 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:2927

Everyone,

it looks like the testing script ran out of disk space a few minutes ago 
again and the submission script is having probles as well.

We are working on it and will post again once everything is fixed.

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine5.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: How should the server exit
Date: Thu, 31 Jan 2002 14:17:18 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2928

The client initiate the connection termination and both the parent and
child can exit. But what about the server? In the sever.c code, after it
has finished with one client, it go back to
 for(;;) {
	sd = myaccept(bindsd, (struct sockaddr*)&sin, &len, unreliable);
         .............

Then there is no client having connection request, so my program displays:

####Accepting a new connection now at port# 61188#### (sockfd=3)
parent: reading the peer address on a syn handshake (read only 0
bytes)recv: Error 0
myaccept: Error 0

Is that the correct exit status of the server, or it should wait there for
ever, but how to make it wait there for ever?

Thanks

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: how to submit hw2.A
Date: Thu, 31 Jan 2002 14:36:08 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:2929

Hi,

There seems to be some confusion about how to submit hw2.A. This time we 
are using a new script for doing the submission. From the directory 
where you have the code you have to call:

/afs/ir/class/cs244a/bin/submit.pl hw2.A TAname

where TAname is one of the akishan, appenz, holliman, hondroul or xwang00

Until 5 minutes ago there was a problem with the permissions of the 
directory. Now it has been solved, and you start submitting your code 
for grading.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Guido Appenzeller" 
Newsgroups: su.class.cs244a
Subject: Re: Control Loop and Multiple Child Processes
Date: Thu, 31 Jan 2002 15:27:11 -0800
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dnab4046a5.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2930

Ron,

the short answer is that there is only one connection for this exercise.

Generally, it depends how you code things. I would think that the select in
the server part of transport.c only has one thread for all clients, it has
to select on fd's for all of them. If you use select in the clients that's
different, however the perforomance hit you take by doing a select on all of
them and then having some additional checks is probably minimal so doing a
select on everything would work as well.

  Guido

> I'm slightly confused as to how things should work when there are multiple
> entries in the context array (and therefore multiple child processes.)
It
> seems to me that a single child process could handle all of the transport
> layer processing, yet we seem to fork a child process for each connection.
> While this eases the communications, it seems to me that it complicates
the
> control loop...
>
> Enough rambling; here's my question:
>
> Should we be coding the control loop so that it only 'selects' the file
> descriptors associated with this particular (child) process?    (I don't
see
> another way, but the skeleton code seems to imply otherwise.)
>
> Thanks,
>
> --Ron
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Guido Appenzeller" 
Newsgroups: su.class.cs244a
Subject: Re: Initial SYN Send
Date: Thu, 31 Jan 2002 15:30:29 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dnab4046a5.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2931

I have no guess what this could be. I have office hours today @6pm, if you
want to we can have a look at it then.

"Steven Siuhong Ngai"  schrieb im Newsbeitrag

> Interesting behavior:
>
> When my client sends out the SYN packet from ctx->sockfd on the very first
> call (22 bytes, successful), select immediately begins to register the
> presence of data to be read on ctx->sockfd. [This is true because I skip
the
> preceding code if I am passive, and select does the blocking/timeout in
that
> case.] This is weird because I have not implemented the SYN_ACK on the
other
> side. An attempt to investigate this mysterious data through network_recv
> produces a -1.
>
> Anyone encounter anything similar? I am correctly resetting the socket set
> before it heads into the select statement.
>
> Steve
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine4.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: when I ran student Tx - student Rx, ok.  test script said not ok ??
Date: Thu, 31 Jan 2002 16:34:23 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2932

Hi,

when I run client and server, they seem to exchange files well. However, I
just submitted a test script and it said
===========================
===(T1.Jb) Checksum differences found (INCORRECT!)
Student's Tx against student's Rx (reliable mode)

---

---
Difference between expected result and your result:
1a2,3
> 4265638025 299328 TESTDATA.1
> 2181702833 5098 TESTDATA.2


any hint?  Thanks!


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: scripts are working
Date: Thu, 31 Jan 2002 16:39:19 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:2933

Hi,

I just wanted to let you know that the both the script for testing and 
submitting are working fine.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth3.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: when I ran student Tx - student Rx, ok.  test script said not ok ??
Date: 1 Feb 2002 00:44:03 GMT
Organization: Stanford University, CA 94305, USA
Lines: 5
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth3.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2934



Hi, I am having the same problem. I'm thinking it must be because we're
not following the STCP protocol correctly.

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: The TA is in...
Date: Thu, 31 Jan 2002 17:08:02 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2935

Hi everyone,

as the assignment is due tomorrow, my office hours today will
start an hour earlier (they end as usual). I am now in Sweet 
Hall.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Midterm discrepancy
Date: 1 Feb 2002 01:17:48 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2936

The dates posted on the time schedule on the course web site are the
correct dates. Thus, the midterm is in fact two weeks from today (Feb
14th). 

Arun

Michael Kahlil Madison  wrote: : On Handout
#1 it says our midterm is Feb 19th... however, on the website : calendar
it says our midterm is Feb 14th.

: Which is correct? and what about the final -- the web site says the 18th
: is that correct?

: -mike
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: windows
Date: 1 Feb 2002 01:19:10 GMT
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2937

Yes, the bytes of the header do not count as data that needs to be
buffered, and thus do not need to consume sequence number space. There are
however some sequence numbers however that have no associated data, e.g.,
the SYN numbers.

Arun


Steven Siuhong Ngai  wrote:
: We count only data in our windows, right? That is, we don't include the
: header byte count as part of the max number of bytes that can be
: outstanding?

: Thanks,
: Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Submit script?
Date: 1 Feb 2002 01:20:42 GMT
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2938

Yes, we're working on this. It should be up and running now.

Arun

Martin Casado  wrote:

: Hi,

:   I think the submit script might be broken.

: elaine15:~/projects/hw2> /afs/ir/class/cs244a/bin/submit xwang00 hw2.A
: Submit: "hw2.A" is not an assignment for this class.

: Format: submit TA Assignment

: TA is one of: hondroul appenz akishan xwang00 holliman

: Assignment is one of: hw1
: hw1             - FTP Directory Copy
: elaine15:~/projects/hw2>


:                                             ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: network termination
Date: 1 Feb 2002 01:29:38 GMT
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2939

Once the FIN is sent, you can wait either for a FIN or a FIN_ACK (a FIN if 
both sides send it simultaneously rsulting in a deadlock situation -- 
though this is unlikely).

Arun

Sriram Viji  wrote:

: when a application closes its socket:
: the child process sends out a FIN packet and waits for a FIN_ACK from the
: peer

: Q:  does it consider any of the packets received between sending the FIN and
: receiving the FIN_ACK?

: thanks,
: sriram


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: scripts are working
Date: Thu, 31 Jan 2002 17:35:58 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2940


Hi...

I can't still do submit_test... Please check this...

>>>>/afs/ir/class/cs244a/bin/submit_test.pl hw2.A akishan

SubmitBase: /afs/ir/class/cs244a/submissions
Debug Level = 1
mkdir /afs/ir/class/cs244a/submissions/hw2.A/akishan/hpbae
mkdir /afs/ir/class/cs244a/submissions/hw2.A/akishan/hpbae: Permission
denied at /afs/ir/class/cs244a/bin/submit_test.pl line 135


"Pablo Molinero Fernandez"  wrote in message

> Hi,
>
> I just wanted to let you know that the both the script for testing and
> submitting are working fine.
>
> Pablo
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: Re: scripts are working
Date: Thu, 31 Jan 2002 17:43:16 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2941


I submitted my test more than an hour ago. I understand that there might
be other tests in the queue, but is it reasonable to expect that the wait
would be this long?

Thanks,
Alex

On Thu, 31 Jan 2002, Pablo Molinero Fernandez wrote:

> Hi,
>
> I just wanted to let you know that the both the script for testing and
> submitting are working fine.
>
> Pablo
>
>

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: How should the server exit
Date: 1 Feb 2002 01:45:29 GMT
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2942

The server should return to a state in which it is ready to accept a new 
connection. The client should close and exit, but the server will 
essentially return to its initial state, such that another client may 
connect to it.

Arun

Fang Sun  wrote:

: The client initiate the connection termination and both the parent and
: child can exit. But what about the server? In the sever.c code, after it
: has finished with one client, it go back to
:  for(;;) {
: 	sd = myaccept(bindsd, (struct sockaddr*)&sin, &len, unreliable);
:          .............

: Then there is no client having connection request, so my program displays:

: ####Accepting a new connection now at port# 61188#### (sockfd=3)
: parent: reading the peer address on a syn handshake (read only 0
: bytes)recv: Error 0
: myaccept: Error 0

: Is that the correct exit status of the server, or it should wait there for
: ever, but how to make it wait there for ever?

: Thanks

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Error on network_send
Date: 1 Feb 2002 01:48:35 GMT
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2943

Your code looks mostly correct....if the problem persists, please bring 
the code into office hour such that we may take a closer look at it.

Arun

Brandon Badger  wrote:
: I'm getting stuck on an error where network_send() returns -1 when the
: server tries to send the contents of the requested file to the client.

: 1) The handshake works correctly.
: 2) transport_appl_io() is called in the client and the filename is passed
: to the client's child.  This sends the filename on sockfd.
: 3) transport_sock_io() is called in the server and the filename is passed
: to the server's parent.
: 4) transport_appl_io() is called in the server and the contents of the
: file are passed to the server's child.  At this point I'm able to print
: out the contents of the file.  - But, when I try to send the file contents
: through sockfd, the network_send function returns -1;

: any ideas?  What are some causes for network_send to return -1?

: In my transport_appl_io() function, the network_send code works correctly
: for the client side when it sends the filename to the server.

: stcpPacket = malloc(packetSize);
: memcpy(stcpPacket, hdrPtr, sizeof(STCPHeader));
: memcpy(stcpPacket + sizeof(STCPHeader), msg + dataStart, len);
: /* This printf of the file contents works */
: /* printf("\nMSG:%s", stcpPacket + sizeof(STCPHeader)); */
: nSent = network_send(ctx->sockfd, stcpPacket, packetSize, NULL);

: Thanks for any help,
: Brandon


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: scripts are working
Date: Thu, 31 Jan 2002 17:49:34 -0800
Lines: 29
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2944

ditto

BAEHOPIL wrote:
> 
> Hi...
> 
> I can't still do submit_test... Please check this...
> 
> >>>>/afs/ir/class/cs244a/bin/submit_test.pl hw2.A akishan
> 
> SubmitBase: /afs/ir/class/cs244a/submissions
> Debug Level = 1
> mkdir /afs/ir/class/cs244a/submissions/hw2.A/akishan/hpbae
> mkdir /afs/ir/class/cs244a/submissions/hw2.A/akishan/hpbae: Permission
> denied at /afs/ir/class/cs244a/bin/submit_test.pl line 135
> 
> "Pablo Molinero Fernandez"  wrote in message
> 
> > Hi,
> >
> > I just wanted to let you know that the both the script for testing and
> > submitting are working fine.
> >
> > Pablo
> >

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: when I ran student Tx - student Rx, ok.  test script said not ok ??
Date: Thu, 31 Jan 2002 17:53:56 -0800
Lines: 42
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2945


I also had this mysterious error...
But when I tried to transfer a large file with my own TX and RX, it really
failed. ( blocked somewhere..)
And only after adding some (really) basic flow control, could I pass the
test..
It seemed to me that flooding packets without any control causes some part
to stop working..
(  I'm not sure where... I was just sure that I was following the spec
because other SYN/ACK/FIN related
tests were OK.. )
Maybe Sandeep had a similar experience with us.. although he took a
different solution..
( See his "Found it" posting.. )

Good luck...


"Huan Wanda Jiang"  wrote in message

> Hi,
>
> when I run client and server, they seem to exchange files well. However, I
> just submitted a test script and it said
> ===========================
> ===(T1.Jb) Checksum differences found (INCORRECT!)
> Student's Tx against student's Rx (reliable mode)
>
> ---
>
> ---
> Difference between expected result and your result:
> 1a2,3
> > 4265638025 299328 TESTDATA.1
> > 2181702833 5098 TESTDATA.2
>
>
> any hint?  Thanks!
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Vasco Chatalbashev" 
Newsgroups: su.class.cs244a
Subject: UDP
Date: Thu, 31 Jan 2002 18:03:21 -0800
Lines: 12
Distribution: su
Message-ID: 
Reply-To: "Vasco Chatalbashev" 
NNTP-Posting-Host: vassil.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2946

So, our network layer is essentially UDP.
We are assuming that the Stanford network is lightly loaded, so that
unless we purposely simulate dropping(or reordering )packets, packet
loss will not occur. Isn't it possible,however, that, say, while you are
running
tests on the submissions, a UDP datagram carrying a STCP packet does get
dropped (presumably it happens
sometimes,) thus making a correct implementation of STCP act unpredictably
for the first milestone? Am I missing something?



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Regarding target address when using network_send
Date: 1 Feb 2002 02:08:34 GMT
Lines: 48
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2947


Yes, as we have mentioned before, you will not need to worry about the 
multi-client situation for this assignment...the architecture does not 
lend itself to this. 

Arun

BAEHOPIL  wrote:

: I wonder about how network_send() figures out the destination address..

: For active side, it doesn't have to worry because we have already registered
: the destination address using connect().
: But for passive side, it seems to get the destination address from
: global_net_ctx->fromaddr.

:   if (global_net_ctx->is_passive)
:     {
:       mh.msg_name = (char*) &(global_net_ctx->fromaddr);
:       mh.msg_namelen = global_net_ctx->fromlen;
:     }

: But global_net_ctx->fromaddr is set in network_recv()..

:   len = recvfrom(sockfd, recvBuf, recvLen, 0,
:                 &(global_net_ctx->fromaddr), &(global_net_ctx->fromlen));

: It seems to me that fromaddr is set to the source address of the received
: packet..

: Then, it seems to me that network_send() is sending to an IP address from
: which
: we received the last packet, not to an IP address from which we initially
: accepted
: connection.
: It won't be a problem if there is only one client at a time,
: but if another should try to connect while one is connected,
: I think everything will be messed up..

: My code seems to be working without this consideration,
: but I really do wonder about what it's intended to be like..

: And, when I used connect() at the passive side,
: network_send() no longer worked.. Is there any reason we should not
: use connect() on a UDP socket we already used bind() on?



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: when I ran student Tx - student Rx, ok.  test script said not ok ??
Date: 1 Feb 2002 02:14:47 GMT
Organization: Stanford University, CA 94305, USA
Lines: 5
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine7.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2948


How big did you have to get before you saw problems?
I can successfully transfer 10M with no lock up.

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: when I ran student Tx - student Rx, ok.  test script said not ok ??
Date: Thu, 31 Jan 2002 18:42:43 -0800
Lines: 15
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2949


Not that large.. Just a few Mbytes..
( My code locked up no matter client and server are on the same machine or
not...)
If you're OK with 10M, maybe you have a different problem..

"Mark Joseph Dolan"  wrote in message

>
> How big did you have to get before you saw problems?
> I can successfully transfer 10M with no lock up.
>
> -mark


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Submit script?
Date: Thu, 31 Jan 2002 18:51:11 -0800
Lines: 41
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
To: Arun Upadhyaya Kishan 
Xref: nntp.stanford.edu su.class.cs244a:2950

No it's not; I'm getting an "Illegal instruction" error from it.

-Sandeep

Arun Upadhyaya Kishan wrote:

> Yes, we're working on this. It should be up and running now.
> 
> Arun
> 
> Martin Casado  wrote:
> 
> : Hi,
> 
> :   I think the submit script might be broken.
> 
> : elaine15:~/projects/hw2> /afs/ir/class/cs244a/bin/submit xwang00 hw2.A
> : Submit: "hw2.A" is not an assignment for this class.
> 
> : Format: submit TA Assignment
> 
> : TA is one of: hondroul appenz akishan xwang00 holliman
> 
> : Assignment is one of: hw1
> : hw1             - FTP Directory Copy
> : elaine15:~/projects/hw2>
> 
> 
> :                                             ~~m
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Submit script?
Date: Thu, 31 Jan 2002 18:58:22 -0800
Lines: 61
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
To: Sandeep Tamhankar 
CC: Arun Upadhyaya Kishan 
Xref: nntp.stanford.edu su.class.cs244a:2951

I take it back; I was running submit, not submit.pl.  You should've 
changed submit to be a shell script that executes submit.pl to avoid 
confusion.

Still, there's a warning/error stating:

fs: You don't have the required access rights on 
'/afs/ir/class/cs244a/submissions/grading/hw2.A/akishan/sandeept

But it seems like it submitted anyway because when I made a second 
attempt, it asked if I wanted to have this set of files graded instead 
of my original submission.  So it's mostly working now.

-Sandeep

Sandeep Tamhankar wrote:

> No it's not; I'm getting an "Illegal instruction" error from it.
> 
> -Sandeep
> 
> Arun Upadhyaya Kishan wrote:
> 
>> Yes, we're working on this. It should be up and running now.
>>
>> Arun
>>
>> Martin Casado  wrote:
>>
>> : Hi,
>>
>> :   I think the submit script might be broken.
>>
>> : elaine15:~/projects/hw2> /afs/ir/class/cs244a/bin/submit xwang00 hw2.A
>> : Submit: "hw2.A" is not an assignment for this class.
>>
>> : Format: submit TA Assignment
>>
>> : TA is one of: hondroul appenz akishan xwang00 holliman
>>
>> : Assignment is one of: hw1
>> : hw1             - FTP Directory Copy
>> : elaine15:~/projects/hw2>
>>
>>
>> :                                             ~~m
>>
>>
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: network termination
Date: Thu, 31 Jan 2002 18:59:26 -0800
Lines: 37
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2952

Since the spec says that only the "client" can close the connection, 
deadlock shouldn't be possible, right?

-Sandeep

Arun Upadhyaya Kishan wrote:

> Once the FIN is sent, you can wait either for a FIN or a FIN_ACK (a FIN if 
> both sides send it simultaneously rsulting in a deadlock situation -- 
> though this is unlikely).
> 
> Arun
> 
> Sriram Viji  wrote:
> 
> : when a application closes its socket:
> : the child process sends out a FIN packet and waits for a FIN_ACK from the
> : peer
> 
> : Q:  does it consider any of the packets received between sending the FIN and
> : receiving the FIN_ACK?
> 
> : thanks,
> : sriram
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: when I ran student Tx - student Rx, ok.  test script said not ok ??
Date: Thu, 31 Jan 2002 19:04:35 -0800
Lines: 31
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2953

As ... BAEHOPIL (wish I knew his/her real name) said earlier, I ran into 
the blocking problem, too, and implemented some very rudimentary flow 
control (i.e. delays) to keep UDP datagrams from getting lost in the ether.

I was able to make things block when sending a 50k file REPEATEDLY 
between elaine7 and myth8.  I just kept my server running, and everytime 
my client finished, I'd up-arrow-and-enter to do it again.  Usually the 
third-time would make it block.

So if you're successfully transferring a large amount of data 
repeatedly, then you probably don't have the same problem that we did.

-Sandeep

Mark Joseph Dolan wrote:

> How big did you have to get before you saw problems?
> I can successfully transfer 10M with no lock up.
> 
> -mark
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: UDP
Date: Thu, 31 Jan 2002 19:06:14 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2954

No, you're pretty much right.  A few of us had problems with network 
congestion -- I ended up sending packets with delays between so that the 
congestion is low enough to keep things reliable.

-Sandeep

Vasco Chatalbashev wrote:

> So, our network layer is essentially UDP.
> We are assuming that the Stanford network is lightly loaded, so that
> unless we purposely simulate dropping(or reordering )packets, packet
> loss will not occur. Isn't it possible,however, that, say, while you are
> running
> tests on the submissions, a UDP datagram carrying a STCP packet does get
> dropped (presumably it happens
> sometimes,) thus making a correct implementation of STCP act unpredictably
> for the first milestone? Am I missing something?
> 
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sutin Chen 
Newsgroups: su.class.cs244a
Subject: parent searching for peeraddress on an empty socket
Date: Thu, 31 Jan 2002 20:23:24 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic2.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2955

Am having problems with select and the parent thread trying to read the
peeraddress....

Basically, my parent thread for some reason is selecting the syn_sd
socket even though I have not sent it anything yet.  Since I haven't
sent anything to it, the thread of course reads 0 bytes on a recv call
and then craps out since it didn't get a peeraddress.

Any ideas would be very much appreciated!

Thanks,
Sutin
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine17.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: help!! script errors!!
Date: 1 Feb 2002 04:31:21 GMT
Organization: Stanford University, CA 94305, USA
Lines: 29
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine17.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2956

Hi, if anybody has any suggestions they would be greatly appreciated!!

I've been having this problem for a day now and I can't seem to solve it.
I pass all the tests except T1.B and T1.C which fail completly.
And no matter what I do I can't seem cause my code to run incorrectly, I even
created files of the same sizes as the tests. But this

> ===(T1.Ba) Byte count differences found (INCORRECT!)
> Student's Tx against our Rx (reliable mode)
> 
> ---
> 
> ---
> Difference between expected result and your result:
> 0a1,3
> >    2666   15720  110230 TESTDATA.0
> >    1812   10049  299328 TESTDATA.1
> >     131     501    5098 TESTDATA.2
> 
> ---
> 
> ===========================

suggests to me that I'm not sending/recieving any of them. 
Any ideas?

Thank you

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine17.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: parent searching for peeraddress on an empty socket
Date: 1 Feb 2002 04:39:08 GMT
Organization: Stanford University, CA 94305, USA
Lines: 4
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine17.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2957

I've had that problem. I found I had a memmory error and fixing that
solved my problem. Just an idea.

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Submit script?
Date: Thu, 31 Jan 2002 20:50:41 -0800
Lines: 28
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2958

I'm having the same problem...
And I'm afraid that there was no actual submission but only the submission
flag is set, so the script is asking if we want to renew the submission...
Please.. check this..

Hopil

"Sandeep Tamhankar"  wrote in message

> I take it back; I was running submit, not submit.pl.  You should've
> changed submit to be a shell script that executes submit.pl to avoid
> confusion.
>
> Still, there's a warning/error stating:
>
> fs: You don't have the required access rights on
> '/afs/ir/class/cs244a/submissions/grading/hw2.A/akishan/sandeept
>
> But it seems like it submitted anyway because when I made a second
> attempt, it asked if I wanted to have this set of files graded instead
> of my original submission.  So it's mostly working now.
>
> -Sandeep
>
> Sandeep Tamhankar wrote:
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Desperately trying to pass the script? Read this.
Date: Thu, 31 Jan 2002 21:13:10 -0800
Lines: 95
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012540391 26624 128.12.133.48 (1 Feb 2002 05:13:11 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2959

Sandeep and others,
    I really do not believe that network congestion is causing everyone's
problems, and I think that your repeated suggestion to the contrary will end
up misleading and hurting a lot of people. Best I can tell, the test script
is run over a local UDP connection, from epic5 to epic5. (Or at worst, f rom
epic5 to epic6 or 7.) I find it extremely unlikely that the test script
would drop packets regularly over such a short hop--on a local machine they
don't even get as far as a network adapter! Not to mention, putting a delay
in sending packets doesn't help you if you're failing against their
transmitter.

    I also observe that my own code has never had a problem with the test
script for 2A. If there were dropped packets on such a regular basis for
everyone else, I don't think I'd have been so lucky as to escape. I suspect
there is some subtlety in the protocol that many people are somehow missing.
(I've seen a couple of situations where a subtle protocol glitch can lead to
test script failures that are not easily reproducible with your own code or
others'.)

    Now, one thing that I find interesting to think about is this. Suppose
that the transport layer used in the test script is a full STCP
implementation; that is, suppose it implements timeouts and retransmissions.
Suppose also that on a given test run, all packets that were sent were
received, in the order they were sent. Suppose, however, that sometimes
things ran a little slowly, and packets  took a little while to arrive. (A
slight increase in processor load  on the machine might cause the network
layer to not process the packets quite as fast, but it would still do so in
the right order.) So, this test run would be fine according to the
milestone: all packets were delivered, in order.

    An STCP implementation that supports timeouts might very well time out
and resend in this circumstance. (I've actually observed this with my
milestone 2 code with as little as 45K of data.) So, the receiver might get
a bunch of packets that they've already received. It seems from the
discussion on this group that many people are completely ignoring the
sliding window part of the protocol, with elaborate justifications. However,
if your code doesn't correctly ignore data that you've already received, you
could have too much data coming out, memory scribbles, or segfaults that are
causing problems with the script.
You'd probably never see this with another student's code, because they
never do timeouts.

    One thing you could try is making your code transmit every window twice
and see if that breaks your receiver, but it might just be quicker to put
the already-received test in and try the script again.

DISCLAIMER: all of that is pulled straight out of my overactive imagination.
The test script may be doing it this way, and it may not. (It's kind of
amusing to me that we're all sitting here trying to reverse-engineer and
second-guess a black-box script.) But the scenario I just described makes
some kind of sense to me. You could argue that the test script should be
using a barely-passing implementation of 2A, and I might agree with you.
However, the code change required to ignore already-received data should be
minor, and hey, you've tried everything else, right?

Comments?

Jonathan

"Sandeep Tamhankar"  wrote in message

> No, you're pretty much right.  A few of us had problems with network
> congestion -- I ended up sending packets with delays between so that the
> congestion is low enough to keep things reliable.
>
> -Sandeep
>
> Vasco Chatalbashev wrote:
>
> > So, our network layer is essentially UDP.
> > We are assuming that the Stanford network is lightly loaded, so that
> > unless we purposely simulate dropping(or reordering )packets, packet
> > loss will not occur. Isn't it possible,however, that, say, while you are
> > running
> > tests on the submissions, a UDP datagram carrying a STCP packet does get
> > dropped (presumably it happens
> > sometimes,) thus making a correct implementation of STCP act
unpredictably
> > for the first milestone? Am I missing something?
> >
> >
> >
> >
>
>
>
> --
> ---------------------------------------------
> Sandeep V. Tamhankar
> M.S. Student
> Computer Science
> Email: 
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Desperately trying to pass the script? Read this.
Date: Thu, 31 Jan 2002 21:30:00 -0800
Lines: 358
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0343_01C1AA9E.7056DCA0"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2960

This is a multi-part message in MIME format.

------=_NextPart_000_0343_01C1AA9E.7056DCA0
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


I think I agree with Jonathan..
I made a similar assumption about test_script, and added some window =
control
which was absent at first.. Then I could pass the test..
Still, my code looks like kind of hack, but at least, it's NOT sending =
and receiving
everything given to it.

Of course, I should attach a big DISCLAIMER, too,
but as Jonathan said, putting some basic window checking isn't so =
complex
(at least there is no timeout or retransmission (I mean, for our TX), or =
reassembly )
that I think it's worth trying if you haven't done it yet..

Hopil

"Jonathan Keljo"  wrote in message =

> Sandeep and others,
>     I really do not believe that network congestion is causing =
everyone's
> problems, and I think that your repeated suggestion to the contrary =
will end
> up misleading and hurting a lot of people. Best I can tell, the test =
script
> is run over a local UDP connection, from epic5 to epic5. (Or at worst, =
f rom
> epic5 to epic6 or 7.) I find it extremely unlikely that the test =
script
> would drop packets regularly over such a short hop--on a local machine =
they
> don't even get as far as a network adapter! Not to mention, putting a =
delay
> in sending packets doesn't help you if you're failing against their
> transmitter.
>=20
>     I also observe that my own code has never had a problem with the =
test
> script for 2A. If there were dropped packets on such a regular basis =
for
> everyone else, I don't think I'd have been so lucky as to escape. I =
suspect
> there is some subtlety in the protocol that many people are somehow =
missing.
> (I've seen a couple of situations where a subtle protocol glitch can =
lead to
> test script failures that are not easily reproducible with your own =
code or
> others'.)
>=20
>     Now, one thing that I find interesting to think about is this. =
Suppose
> that the transport layer used in the test script is a full STCP
> implementation; that is, suppose it implements timeouts and =
retransmissions.
> Suppose also that on a given test run, all packets that were sent were
> received, in the order they were sent. Suppose, however, that =
sometimes
> things ran a little slowly, and packets  took a little while to =
arrive. (A
> slight increase in processor load  on the machine might cause the =
network
> layer to not process the packets quite as fast, but it would still do =
so in
> the right order.) So, this test run would be fine according to the
> milestone: all packets were delivered, in order.
>=20
>     An STCP implementation that supports timeouts might very well time =
out
> and resend in this circumstance. (I've actually observed this with my
> milestone 2 code with as little as 45K of data.) So, the receiver =
might get
> a bunch of packets that they've already received. It seems from the
> discussion on this group that many people are completely ignoring the
> sliding window part of the protocol, with elaborate justifications. =
However,
> if your code doesn't correctly ignore data that you've already =
received, you
> could have too much data coming out, memory scribbles, or segfaults =
that are
> causing problems with the script.
> You'd probably never see this with another student's code, because =
they
> never do timeouts.
>=20
>     One thing you could try is making your code transmit every window =
twice
> and see if that breaks your receiver, but it might just be quicker to =
put
> the already-received test in and try the script again.
>=20
> DISCLAIMER: all of that is pulled straight out of my overactive =
imagination.
> The test script may be doing it this way, and it may not. (It's kind =
of
> amusing to me that we're all sitting here trying to reverse-engineer =
and
> second-guess a black-box script.) But the scenario I just described =
makes
> some kind of sense to me. You could argue that the test script should =
be
> using a barely-passing implementation of 2A, and I might agree with =
you.
> However, the code change required to ignore already-received data =
should be
> minor, and hey, you've tried everything else, right?
>=20
> Comments?
>=20
> Jonathan
>=20
> "Sandeep Tamhankar"  wrote in message
> 
> > No, you're pretty much right.  A few of us had problems with network
> > congestion -- I ended up sending packets with delays between so that =
the
> > congestion is low enough to keep things reliable.
> >
> > -Sandeep
> >
> > Vasco Chatalbashev wrote:
> >
> > > So, our network layer is essentially UDP.
> > > We are assuming that the Stanford network is lightly loaded, so =
that
> > > unless we purposely simulate dropping(or reordering )packets, =
packet
> > > loss will not occur. Isn't it possible,however, that, say, while =
you are
> > > running
> > > tests on the submissions, a UDP datagram carrying a STCP packet =
does get
> > > dropped (presumably it happens
> > > sometimes,) thus making a correct implementation of STCP act
> unpredictably
> > > for the first milestone? Am I missing something?
> > >
> > >
> > >
> > >
> >
> >
> >
> > --
> > ---------------------------------------------
> > Sandeep V. Tamhankar
> > M.S. Student
> > Computer Science
> > Email: 
> >
>=20
>=20

------=_NextPart_000_0343_01C1AA9E.7056DCA0
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>I think I agree with =
Jonathan..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I made a similar assumption about =
test_script, and=20
added some window control</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>which was absent at first.. Then I =
could pass the=20
test..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Still, my code looks like kind of hack, =
but at=20
least, it's NOT sending and receiving</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>everything given to it.</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Of course, I should attach a big =
DISCLAIMER,=20
too,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>but as Jonathan said, putting some =
basic window=20
checking isn't so complex</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>(at least there is no timeout or =
retransmission (I=20
mean, for our TX), or reassembly )</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>that I think it's worth trying if =
you haven't=20
done it yet..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>"Jonathan Keljo" <</FONT><A=20
 face=3DArial=20
 face=3DArial size=3D2>> =
wrote in=20
message </FONT><A  =
face=3DArial=20
 =
face=3DArial=20
size=3D2>...</FONT></DIV><FONT face=3DArial size=3D2>> Sandeep and =
others,<BR>>=20
    I really do not believe that network congestion is =
causing=20
everyone's<BR>> problems, and I think that your repeated suggestion =
to the=20
contrary will end<BR>> up misleading and hurting a lot of people. =
Best I can=20
tell, the test script<BR>> is run over a local UDP connection, from =
epic5 to=20
epic5. (Or at worst, f rom<BR>> epic5 to epic6 or 7.) I find it =
extremely=20
unlikely that the test script<BR>> would drop packets regularly over =
such a=20
short hop--on a local machine they<BR>> don't even get as far as a =
network=20
adapter! Not to mention, putting a delay<BR>> in sending packets =
doesn't help=20
you if you're failing against their<BR>> transmitter.<BR>> =
<BR>>=20
    I also observe that my own code has never had a =
problem with=20
the test<BR>> script for 2A. If there were dropped packets on such a =
regular=20
basis for<BR>> everyone else, I don't think I'd have been so lucky as =
to=20
escape. I suspect<BR>> there is some subtlety in the protocol that =
many=20
people are somehow missing.<BR>> (I've seen a couple of situations =
where a=20
subtle protocol glitch can lead to<BR>> test script failures that are =
not=20
easily reproducible with your own code or<BR>> others'.)<BR>> =
<BR>>=20
    Now, one thing that I find interesting to think about =
is=20
this. Suppose<BR>> that the transport layer used in the test script =
is a full=20
STCP<BR>> implementation; that is, suppose it implements timeouts and =

retransmissions.<BR>> Suppose also that on a given test run, all =
packets that=20
were sent were<BR>> received, in the order they were sent. Suppose, =
however,=20
that sometimes<BR>> things ran a little slowly, and packets  =
took a=20
little while to arrive. (A<BR>> slight increase in processor =
load  on=20
the machine might cause the network<BR>> layer to not process the =
packets=20
quite as fast, but it would still do so in<BR>> the right order.) So, =
this=20
test run would be fine according to the<BR>> milestone: all packets =
were=20
delivered, in order.<BR>> <BR>>     An STCP =
implementation=20
that supports timeouts might very well time out<BR>> and resend in =
this=20
circumstance. (I've actually observed this with my<BR>> milestone 2 =
code with=20
as little as 45K of data.) So, the receiver might get<BR>> a bunch of =
packets=20
that they've already received. It seems from the<BR>> discussion on =
this=20
group that many people are completely ignoring the<BR>> sliding =
window part=20
of the protocol, with elaborate justifications. However,<BR>> if your =
code=20
doesn't correctly ignore data that you've already received, you<BR>> =
could=20
have too much data coming out, memory scribbles, or segfaults that =
are<BR>>=20
causing problems with the script.<BR>> You'd probably never see this =
with=20
another student's code, because they<BR>> never do timeouts.<BR>> =
<BR>>=20
    One thing you could try is making your code transmit =
every=20
window twice<BR>> and see if that breaks your receiver, but it might =
just be=20
quicker to put<BR>> the already-received test in and try the script=20
again.<BR>> <BR>> DISCLAIMER: all of that is pulled straight out =
of my=20
overactive imagination.<BR>> The test script may be doing it this =
way, and it=20
may not. (It's kind of<BR>> amusing to me that we're all sitting here =
trying=20
to reverse-engineer and<BR>> second-guess a black-box script.) But =
the=20
scenario I just described makes<BR>> some kind of sense to me. You =
could=20
argue that the test script should be<BR>> using a barely-passing=20
implementation of 2A, and I might agree with you.<BR>> However, the =
code=20
change required to ignore already-received data should be<BR>> minor, =
and=20
hey, you've tried everything else, right?<BR>> <BR>> =
Comments?<BR>>=20
<BR>> Jonathan<BR>> <BR>> "Sandeep Tamhankar" <</FONT><A=20
 face=3DArial=20
 face=3DArial =
size=3D2>> wrote in=20
message<BR>> </FONT><A =

face=3DArial =

face=3DArial size=3D2>...<BR>> > No, you're pretty much =
right.  A few of=20
us had problems with network<BR>> > congestion -- I ended up =
sending=20
packets with delays between so that the<BR>> > congestion is low =
enough to=20
keep things reliable.<BR>> ><BR>> > -Sandeep<BR>> =
><BR>>=20
> Vasco Chatalbashev wrote:<BR>> ><BR>> > > So, our =
network=20
layer is essentially UDP.<BR>> > > We are assuming that the =
Stanford=20
network is lightly loaded, so that<BR>> > > unless we purposely =

simulate dropping(or reordering )packets, packet<BR>> > > loss =
will not=20
occur. Isn't it possible,however, that, say, while you are<BR>> > =
>=20
running<BR>> > > tests on the submissions, a UDP datagram =
carrying a=20
STCP packet does get<BR>> > > dropped (presumably it =
happens<BR>>=20
> > sometimes,) thus making a correct implementation of STCP =
act<BR>>=20
unpredictably<BR>> > > for the first milestone? Am I missing=20
something?<BR>> > ><BR>> > ><BR>> > ><BR>> =
>=20
><BR>> ><BR>> ><BR>> ><BR>> > --<BR>> > =

---------------------------------------------<BR>> > Sandeep V.=20
Tamhankar<BR>> > M.S. Student<BR>> > Computer =
Science<BR>> >=20
Email: </FONT><A  =
face=3DArial=20
 face=3DArial =
size=3D2>>=20
><BR>> <BR>> </FONT></BODY></HTML>

------=_NextPart_000_0343_01C1AA9E.7056DCA0--

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga6.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: comm_sd
Date: Thu, 31 Jan 2002 22:18:32 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2961

I think a socket at the server side remains open always. Putting simply,
the server parent doesn't close comm_sd. So, when the server gets
<CTRL-C>, this socket descriptor still remains open!!

Any comments are welcome.

Abhishek Das

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: comm_sd
Date: Thu, 31 Jan 2002 22:30:11 -0800
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012545011 27208 128.12.133.48 (1 Feb 2002 06:30:11 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2962

I'm not positive, but I'm fairly sure that when a process exits for whatever
reason, all of its file and socket descriptors are freed by the operating
system.

Jonathan

"Abhishek Das"  wrote in message

> I think a socket at the server side remains open always. Putting simply,
> the server parent doesn't close comm_sd. So, when the server gets
> <CTRL-C>, this socket descriptor still remains open!!
>
> Any comments are welcome.
>
> Abhishek Das
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!rgustin
From: Reid Gustin 
Newsgroups: su.class.cs244a
Subject: Re: comm_sd
Date: Thu, 31 Jan 2002 23:00:50 -0800
Lines: 12
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2963

On Thu, 31 Jan 2002, Jonathan Keljo wrote:
> I'm not positive, but I'm fairly sure that when a process exits for whatever
> reason, all of its file and socket descriptors are freed by the operating
> system.

This is true, at least for Solaris, BSD, and Linux (which ought to pretty
much cover this assignment, at least), as long as the process exits. The
only catch is that if the child process lingers, it won't get cleaned up
until you 'kill -9' it (which can be a verb, I'm pretty sure...).

Reid

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sutin Chen 
Newsgroups: su.class.cs244a
Subject: Re: parent searching for peeraddress on an empty socket
Date: Thu, 31 Jan 2002 23:02:34 -0800
Lines: 9
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: epic6.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2964

Wow.  How did I not ever realize before what a beautiful beautiful thing
purify is?  Thanks for the advice - it took care of my bug.

Mark Joseph Dolan wrote:
> 
> I've had that problem. I found I had a memmory error and fixing that
> solved my problem. Just an idea.
> 
> -mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: network termination
Date: 1 Feb 2002 07:34:55 GMT
Lines: 43
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine14.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2965

Yes, you will not need to worry about that particular case for this 
assignment. I was mentioning it as something to possibly consider.

Arun

Sandeep Tamhankar  wrote:
: Since the spec says that only the "client" can close the connection, 
: deadlock shouldn't be possible, right?

: -Sandeep

: Arun Upadhyaya Kishan wrote:

:> Once the FIN is sent, you can wait either for a FIN or a FIN_ACK (a FIN if 
:> both sides send it simultaneously rsulting in a deadlock situation -- 
:> though this is unlikely).
:> 
:> Arun
:> 
:> Sriram Viji  wrote:
:> 
:> : when a application closes its socket:
:> : the child process sends out a FIN packet and waits for a FIN_ACK from the
:> : peer
:> 
:> : Q:  does it consider any of the packets received between sending the FIN and
:> : receiving the FIN_ACK?
:> 
:> : thanks,
:> : sriram
:> 
:> 
:> 



: -- 
: ---------------------------------------------
: Sandeep V. Tamhankar			
: M.S. Student
: Computer Science
: Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Wong 
Newsgroups: su.class.cs244a
Subject: multiple connections
Date: 1 Feb 2002 07:41:24 GMT
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2966

Hi,

So I think I'm confused by a pretty fundamental question. When multiple connections between the server and client are opened, a child process is spawned for each end of each connection. However, the block of comments before the control_loop function states that the control loop should use select to see if "any data on any of the sockets to our peer, ... etc." is available. So am I operating under the false impression that each child process is only supposed to take care of one end of one connection and tha
t the network layer would route the correct packet to the rightchild process? Or am I reading too into the comments?

I guess what I don't understand is how multiple connections the same port via UDP works. It seems like as UDP is connectionless, the packets from different connetions will be arriving multiplexed at the server post.

Help?!

Thanks,
Victor.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: multiple connections
Date: 1 Feb 2002 07:52:53 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2967

As we've mentioned, the existing network layer infrastructure does not 
lend itself to multiple client connectoins. In particular you will see the 
server manages only one client connection at a time. You do not need to 
worry about dealing with this -- consider the interactoin between one 
server and one client process only.

Arun

Victor Wong  wrote:
: Hi,

: So I think I'm confused by a pretty fundamental question. When multiple connections between the server and client are opened, a child process is spawned for each end of each connection. However, the block of comments before the control_loop function states that the control loop should use select to see if "any data on any of the sockets to our peer, ... etc." is available. So am I operating under the false impression that each child process is only supposed to take care of one end of one connection and t
ha
: t the network layer would route the correct packet to the rightchild process? Or am I reading too into the comments?

: I guess what I don't understand is how multiple connections the same
port via UDP works. It seems like as UDP is connectionless, the packets
from different connetions will be arriving multiplexed at the server post.

: Help?!

: Thanks,
: Victor.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine21.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Error on network_send
Date: Fri, 1 Feb 2002 00:15:53 -0800
Lines: 45
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2968

Every now and then I also keep getting errors in network_send (-1) on
server side. The server would work for 10 times and fail the 11'th time I
try to connect and retrieve a file. Any ideas?

Dhawal Kumar

On 1 Feb 2002, Arun Upadhyaya Kishan wrote:

> Your code looks mostly correct....if the problem persists, please bring
> the code into office hour such that we may take a closer look at it.
>
> Arun
>
> Brandon Badger  wrote:
> : I'm getting stuck on an error where network_send() returns -1 when the
> : server tries to send the contents of the requested file to the client.
>
> : 1) The handshake works correctly.
> : 2) transport_appl_io() is called in the client and the filename is passed
> : to the client's child.  This sends the filename on sockfd.
> : 3) transport_sock_io() is called in the server and the filename is passed
> : to the server's parent.
> : 4) transport_appl_io() is called in the server and the contents of the
> : file are passed to the server's child.  At this point I'm able to print
> : out the contents of the file.  - But, when I try to send the file contents
> : through sockfd, the network_send function returns -1;
>
> : any ideas?  What are some causes for network_send to return -1?
>
> : In my transport_appl_io() function, the network_send code works correctly
> : for the client side when it sends the filename to the server.
>
> : stcpPacket = malloc(packetSize);
> : memcpy(stcpPacket, hdrPtr, sizeof(STCPHeader));
> : memcpy(stcpPacket + sizeof(STCPHeader), msg + dataStart, len);
> : /* This printf of the file contents works */
> : /* printf("\nMSG:%s", stcpPacket + sizeof(STCPHeader)); */
> : nSent = network_send(ctx->sockfd, stcpPacket, packetSize, NULL);
>
> : Thanks for any help,
> : Brandon
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Test script for hw2.B
Date: Fri, 1 Feb 2002 01:08:58 -0800
Lines: 56
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0416_01C1AABD.0712D4A0"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2969

This is a multi-part message in MIME format.

------=_NextPart_000_0416_01C1AABD.0712D4A0
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

 Hi..

Is test_script for hw2.B ready for use? I tried it with my code, whose =
RX code is revised for hw2.B
but TX code is still for hw2.A... But the script said my TX works well =
and my RX doesn't..
I know that my TX should NOT work yet..And it reports Go-Back-N is OK, =
but I never implemented it..
Is there just some typo, or should I just wait? If so, when will it be =
available?

Hopil


------=_NextPart_000_0416_01C1AABD.0712D4A0
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2> Hi..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Is test_script for hw2.B ready for use? =
I tried it=20
with my code, whose RX code is revised for hw2.B</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>but </FONT><FONT face=3DArial =
size=3D2>TX code is still=20
for hw2.A... But the script said my TX works well and my RX=20
doesn't..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I know that my TX should NOT work=20
yet..</FONT><FONT face=3DArial size=3D2>And it reports Go-Back-N is OK, =
but I never=20
implemented it..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Is there just some typo, or =
should I just=20
wait? If so, when will it be available?</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_0416_01C1AABD.0712D4A0--

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga8.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Reading 0 bytes from application
Date: Fri, 1 Feb 2002 01:16:17 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2970

Hi,

After the first few packets are sent between the client and the server,
whenever I wait on the select statement, it tells me that there is data
to read from the application. However, when I use the read function, the
number of read bytes is equal to 0 (As it is the case for network
termination). Knowing that the client will keep the connection open till
it read all the bytes of the file, I have no idea why the select tells me
to read from the client when there is nothing to read. My code thinks that
it is the termination case and sends a FIN package !

Hope that someone has an idea?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: Reading 0 bytes from application
Date: 1 Feb 2002 09:33:55 GMT
Organization: Stanford University, CA 94305, USA
Lines: 4
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2971

Just a suggestion, but I've noticed strange select behavior when there are
memory errors.

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga6.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: comm_sd
Date: Fri, 1 Feb 2002 01:39:34 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: saga6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2972

Actually, purify points out the 2 extra file descriptors open,
local_data_sd and sockfd of parent for the server.
So, I was just wondering why it was happening and I agree with you people.

abhishek

On Thu, 31 Jan 2002, Reid Gustin wrote:

> On Thu, 31 Jan 2002, Jonathan Keljo wrote:
> > I'm not positive, but I'm fairly sure that when a process exits for whatever
> > reason, all of its file and socket descriptors are freed by the operating
> > system.
>
> This is true, at least for Solaris, BSD, and Linux (which ought to pretty
> much cover this assignment, at least), as long as the process exits. The
> only catch is that if the child process lingers, it won't get cleaned up
> until you 'kill -9' it (which can be a verb, I'm pretty sure...).
>
> Reid
>
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga4.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: Reading 0 bytes from application
Date: Fri, 1 Feb 2002 01:50:38 -0800
Lines: 12
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2973

Thanks, I solved it

Moh'd
On 1 Feb 2002, Mark Joseph Dolan wrote:

> Just a suggestion, but I've noticed strange select behavior when there are
> memory errors.
>
> -mark
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Test script for hw2.B
Date: Fri, 1 Feb 2002 02:30:40 -0800
Lines: 94
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0013_01C1AAC8.70DC6E40"
X-Trace: news.Stanford.EDU 1012559441 29404 128.12.133.48 (1 Feb 2002 10:30:41 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2974

This is a multi-part message in MIME format.

------=_NextPart_000_0013_01C1AAC8.70DC6E40
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

Hm. Well, that's discouraging. And here I thought I was all hot shit =
when I passed all the tests... :-)

I'm fairly sure the maximum-retransmission-count test is not working, =
and is recently broken. It told me a couple days ago I failed and that I =
sent between 12 and 18 retransmissions, but you can look at the output =
and clearly count only 5. (But an hour before that it was totally =
happy.)

Jonathan
  "BAEHOPIL"  wrote in message =

   Hi..

  Is test_script for hw2.B ready for use? I tried it with my code, whose =
RX code is revised for hw2.B
  but TX code is still for hw2.A... But the script said my TX works well =
and my RX doesn't..
  I know that my TX should NOT work yet..And it reports Go-Back-N is OK, =
but I never implemented it..
  Is there just some typo, or should I just wait? If so, when will it be =
available?

  Hopil


------=_NextPart_000_0013_01C1AAC8.70DC6E40
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META http-equiv=3DContent-Type content=3D"text/html; =
charset=3Dks_c_5601-1987">
<META content=3D"MSHTML 6.00.2712.300" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2>Hm. Well, that's discouraging. And here =
I thought I=20
was all hot shit when I passed all the tests... :-)</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>I'm fairly sure the =
maximum-retransmission-count=20
test is not working, and is recently broken. It told me a couple =
days ago I=20
failed and that I sent between 12 and 18 retransmissions, but you can =
look at=20
the output and clearly count only 5. (But an hour before that it was =
totally=20
happy.)</FONT></DIV>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>Jonathan</FONT></DIV>
<BLOCKQUOTE dir=3Dltr=20
style=3D"PADDING-RIGHT: 0px; PADDING-LEFT: 5px; MARGIN-LEFT: 5px; =
BORDER-LEFT: #000000 2px solid; MARGIN-RIGHT: 0px">
  <DIV>"BAEHOPIL" <<A=20
   =
wrote in=20
  message <A=20
  =

tanford.EDU</A>...</DIV>
  <DIV><FONT face=3DArial size=3D2> Hi..</FONT></DIV>
  <DIV> </DIV>
  <DIV><FONT face=3DArial size=3D2>Is test_script for hw2.B ready for =
use? I tried=20
  it with my code, whose RX code is revised for hw2.B</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>but </FONT><FONT face=3DArial =
size=3D2>TX code is=20
  still for hw2.A... But the script said my TX works well and my RX=20
  doesn't..</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>I know that my TX should NOT =
work=20
  yet..</FONT><FONT face=3DArial size=3D2>And it reports Go-Back-N is =
OK, but I=20
  never implemented it..</FONT></DIV>
  <DIV><FONT face=3DArial size=3D2>Is there just some typo, or =
should I just=20
  wait? If so, when will it be available?</FONT></DIV>
  <DIV> </DIV>
  <DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
  <DIV> </DIV></BLOCKQUOTE></BODY></HTML>

------=_NextPart_000_0013_01C1AAC8.70DC6E40--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: comm_sd
Date: Fri, 1 Feb 2002 02:33:42 -0800
Lines: 46
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012559622 29412 128.12.133.48 (1 Feb 2002 10:33:42 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2975

Well, there's no way to nicely shut down the server. It's in this big
infinite loop that keeps accepting connections until Armageddon. So a ctrl-c
pretty much kills it off without any cleanup. If someone implemented a
ctrl-c handler that correctly ran myclose(), the sockets would be properly
closed.

Jonathan

"Abhishek Das"  wrote in message

> Actually, purify points out the 2 extra file descriptors open,
> local_data_sd and sockfd of parent for the server.
> So, I was just wondering why it was happening and I agree with you people.
>
> abhishek
>
> On Thu, 31 Jan 2002, Reid Gustin wrote:
>
> > On Thu, 31 Jan 2002, Jonathan Keljo wrote:
> > > I'm not positive, but I'm fairly sure that when a process exits for
whatever
> > > reason, all of its file and socket descriptors are freed by the
operating
> > > system.
> >
> > This is true, at least for Solaris, BSD, and Linux (which ought to
pretty
> > much cover this assignment, at least), as long as the process exits. The
> > only catch is that if the child process lingers, it won't get cleaned up
> > until you 'kill -9' it (which can be a verb, I'm pretty sure...).
> >
> > Reid
> >
> >
>
> Abhishek Das
> Graduate Research Assistant
> Computer Systems Lab
> Stanford University
>
> Address:-
> Escondido Village 33B
> Stanford CA 94305
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga4.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Problem with the read function
Date: Fri, 1 Feb 2002 02:40:01 -0800
Lines: 29
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2976

Hi,

I am testing my code by trasmitting a 1 megabyte of binary. However, when
the server starts to receive packets from the application ( by using the
read function when the select says that there should be data from the
application), I found that sometimes the number of readBytes is 0. We know
that this is an indication to close the socket and go to finish
state. The strange thing is that if the server igonores this 0 bytes
message and continue reading from the application, there will be many
coming packets to send. In other words, I am not able to terminate on the
0 return of the read, since in my above case, it happens that there are
are still many packets to send and the application doesn't close the
socket with the trasnport.

I don't know what to do with this, if I ignore the 0 bytes, I will be able
to send the whole of the 1MByte file correctly, however, the client then
must initiate the end of connection, and the server won't be able to
detect a finish state according to the 0 byte indication from read
function.

Any ideas if I am on the wrong track? or should the client always be the
party which closes the connection (Although the handout says that the
passive party maybe the active later and close the connection).

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: TAs: Is the result from Programming Assignment #1 emailed out?
Date: Fri, 1 Feb 2002 02:42:09 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2977

Hi TAs,

Just wondering if the result from PA1 is emailed out? I would like to take a
look at the PA1 comments from the prof/TA before submitting HW2.A, in case I
have anything I need to change.

Thanks!!!!
Henry


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: submit_test.pl is not working
Date: Fri, 1 Feb 2002 02:46:00 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2978

Looks like the submit_test.pl script is not working. It says "No space left
on device".

Please take a look.
Henry


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga8.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: submit_test.pl is not working
Date: Fri, 1 Feb 2002 02:50:50 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2979

I am facing same problem, I sent my test from one hour and till now no
reply

Moh'd
On Fri, 1 Feb 2002, Henry Fu wrote:

> Looks like the submit_test.pl script is not working. It says "No space left
> on device".
>
> Please take a look.
> Henry
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Victor Wong" 
Newsgroups: su.class.cs244a
Subject: select and read
Date: Fri, 1 Feb 2002 03:05:38 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: summerbridge.stanford.edu
X-Trace: news.Stanford.EDU 1012561543 29616 128.12.82.21 (1 Feb 2002 11:05:43 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2980

The answer to the second question on the FAQ states that we should not be
reading any more data from the local socket if we can't send it out. This
means that the socket will still be ready for reading after reading a
portion of the data to fill the send queue. Doesn't this mean that the call
to select would return immediately and the control loop would just end up
polling to see when the ACK arrives and the window opens up? Isn't it more
efficient to implement this with a semaphore/lock so that there's no busy
waiting? Or should we not care?

Thanks,
Victor.


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Leonard Sibille 
Newsgroups: su.class.cs244a
Subject: Re: submit_test.pl is not working
Date: Fri, 01 Feb 2002 03:26:10 -0800
Lines: 23
Distribution: su
Message-ID: 
References:  
Reply-To: 
NNTP-Posting-Host: saclay.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1012563064 29755 171.64.68.193 (1 Feb 2002 11:31:04 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.18 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2981

Same error. I could not submit the test.


Leo



Moh'd Saleem Saleem Alrawashdeh wrote:
> 
> I am facing same problem, I sent my test from one hour and till now no
> reply
> 
> Moh'd
> On Fri, 1 Feb 2002, Henry Fu wrote:
> 
> > Looks like the submit_test.pl script is not working. It says "No space left
> > on device".
> >
> > Please take a look.
> > Henry
> >
> >
> >
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Yichen Xie" 
Newsgroups: su.class.cs244a
Subject: window size
Date: Fri, 1 Feb 2002 03:51:36 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: dn800cb605.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2982

Since th_win field in the tcp header is unused according to the
specification, is it safe to assume that the application reads whatever is
available to it as soon as the data has arrived (otherwise, writes to the
local_data_sd may block at the child's end, and sender will time out, and
all sorts of things could happen...)?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Yichen Xie" 
Newsgroups: su.class.cs244a
Subject: test script
Date: Fri, 1 Feb 2002 03:52:50 -0800
Lines: 3
Distribution: su
Message-ID: 
NNTP-Posting-Host: dn800cb605.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:2983

BTW, test script doesn't seem to respond...


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Submit Script - Illegal Instruction?
Date: 1 Feb 2002 11:59:39 GMT
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine37.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2984

I noticed all the messages regarding the submit_test.pl script, but has 
anyone recently been able to actually submit their hw2.A? I ran
/usr/class/cs244a/bin/submit and it immediately gives me an 'Illegal 
Instruction' message.
Anyone know what is going on with that script?

Thanks in advance for any help.

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Leonard Sibille 
Newsgroups: su.class.cs244a
Subject: Re: submit_test.pl is not working
Date: Fri, 01 Feb 2002 04:32:48 -0800
Lines: 31
Distribution: su
Message-ID: 
References:   
Reply-To: 
NNTP-Posting-Host: saclay.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1012567064 677 171.64.68.193 (1 Feb 2002 12:37:44 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.18 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:2985

Now the test script seems to be working properly. 

The only change I made since last time is to include my README file.


Leo



Leonard Sibille wrote:
> 
> Same error. I could not submit the test.
> 
> Leo
> 
> Moh'd Saleem Saleem Alrawashdeh wrote:
> >
> > I am facing same problem, I sent my test from one hour and till now no
> > reply
> >
> > Moh'd
> > On Fri, 1 Feb 2002, Henry Fu wrote:
> >
> > > Looks like the submit_test.pl script is not working. It says "No space left
> > > on device".
> > >
> > > Please take a look.
> > > Henry
> > >
> > >
> > >
.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Test script - Try again...
Date: Fri, 1 Feb 2002 04:51:00 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2986

Hi everyone,

I cleaned up the disk and the test script should work again.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: Submit Script - Illegal Instruction?
Date: 1 Feb 2002 13:07:59 GMT
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine37.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:2987

Doh! I just read Pablo's earlier post and we should use 
/usr/class/cs244a/bin/submit.pl instead of submit.

That should teach me to read the previous posts more carefully. 

-Victor

Victor Tung  wrote:
> I noticed all the messages regarding the submit_test.pl script, but has 
> anyone recently been able to actually submit their hw2.A? I ran
> /usr/class/cs244a/bin/submit and it immediately gives me an 'Illegal 
> Instruction' message.
> Anyone know what is going on with that script?

> Thanks in advance for any help.


-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine30.Stanford.EDU!kichan
From:  (Shu Kit Francis Chan)
Newsgroups: su.class.cs244a
Subject: test script problem ?
Date: 1 Feb 2002 16:32:09 GMT
Organization: Stanford University, CA 94305, USA
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine30.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:2988


I have submitted the test for more than a couple of hours. Still I have
not got any result back. Is it down ?

Is the following command right ?

/usr/class/cs244a/bin/submit_test.pl hw2.A <TA name>
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regardin the FIN Error.
Date: Fri, 1 Feb 2002 08:33:59 -0800
Lines: 260
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2989

Hi,
Hi,
I am seeing this error for my test.It says that failure to send FIN_ACK
for FIN. But i am not seeing any FIN at the end of the send and recv
sequence. Can you please have a loot at this sequence and tell me how it
can be explained.
Thanks
Shankar

    * Test 1.I: Failure to send a FIN_ACK after a FIN
  [out of 5]

[SUMMARY] NOT OK
[RESULT]  Incorrect FIN_ACK behaviour after FIN

---
network_send: a SYN packet with seq=198
network_recv: a SYN-ACK packet with seq=198 and ack=199 and
timestamp=3230007382
network_recv: a DATA-ACK packet with ack=211
network_recv: a DATA packet with seq=199, len=536 and timestamp=3230007841
network_send: a DATA-ACK packet with ack=735
network_recv: a DATA packet with seq=735, len=536 and timestamp=3230007841
network_send: a DATA-ACK packet with ack=1271
network_recv: a DATA packet with seq=1271, len=536 and
timestamp=3230007842
network_send: a DATA-ACK packet with ack=1807
network_recv: a DATA packet with seq=1807, len=536 and
timestamp=3230007842
network_send: a DATA-ACK packet with ack=2343
network_recv: a DATA packet with seq=2343, len=536 and
timestamp=3230007843
network_send: a DATA-ACK packet with ack=2879
network_recv: a DATA packet with seq=2879, len=392 and
timestamp=3230007843
network_send: a DATA-ACK packet with ack=3271
network_recv: a DATA packet with seq=3271, len=536 and
timestamp=3230007844
network_send: a DATA-ACK packet with ack=3807
network_recv: a DATA packet with seq=3807, len=536 and
timestamp=3230007844
network_send: a DATA-ACK packet with ack=4343
network_recv: a DATA packet with seq=4343, len=536 and
timestamp=3230007845
network_send: a DATA-ACK packet with ack=4879
network_recv: a DATA packet with seq=4879, len=536 and
timestamp=3230007846
network_send: a DATA-ACK packet with ack=5415
network_recv: a DATA packet with seq=5415, len=536 and
timestamp=3230007846
network_send: a DATA-ACK packet with ack=5951
network_recv: a DATA packet with seq=5951, len=392 and
timestamp=3230007847
network_send: a DATA-ACK packet with ack=6343
network_recv: a DATA packet with seq=6343, len=536 and
timestamp=3230007847
network_send: a DATA-ACK packet with ack=6879
network_recv: a DATA packet with seq=6879, len=536 and
timestamp=3230007848
network_send: a DATA-ACK packet with ack=7415
network_recv: a DATA packet with seq=7415, len=536 and
timestamp=3230007849
network_send: a DATA-ACK packet with ack=7951
network_recv: a DATA packet with seq=7951, len=536 and
timestamp=3230007849
network_send: a DATA-ACK packet with ack=8487
network_recv: a DATA packet with seq=8487, len=536 and
timestamp=3230007850
network_send: a DATA-ACK packet with ack=9023
network_recv: a DATA packet with seq=9023, len=392 and
timestamp=3230007850
network_send: a DATA-ACK packet with ack=9415
network_recv: a DATA packet with seq=9415, len=536 and
timestamp=3230007851
network_send: a DATA-ACK packet with ack=9951
network_recv: a DATA packet with seq=9951, len=536 and
timestamp=3230007851
network_send: a DATA-ACK packet with ack=10487
network_recv: a DATA packet with seq=10487, len=536 and
timestamp=3230007852
network_send: a DATA-ACK packet with ack=11023
network_recv: a DATA packet with seq=11023, len=536 and
timestamp=3230007852
network_send: a DATA-ACK packet with ack=11559
network_recv: a DATA packet with seq=11559, len=536 and
timestamp=3230007853
network_send: a DATA-ACK packet with ack=12095
network_recv: a DATA packet with seq=12095, len=392 and
timestamp=3230007854
network_send: a DATA-ACK packet with ack=12487
network_recv: a DATA packet with seq=12487, len=536 and
timestamp=3230007854
network_send: a DATA-ACK packet with ack=13023
network_recv: a DATA packet with seq=13023, len=536 and
timestamp=3230007855
network_send: a DATA-ACK packet with ack=13559
network_recv: a DATA packet with seq=13559, len=536 and
timestamp=3230007855
network_send: a DATA-ACK packet with ack=14095
network_recv: a DATA packet with seq=14095, len=536 and
timestamp=3230007856
network_send: a DATA-ACK packet with ack=14631
network_recv: a DATA packet with seq=14631, len=536 and
timestamp=3230007856
network_send: a DATA-ACK packet with ack=15167
network_recv: a DATA packet with seq=15167, len=392 and
timestamp=3230007857
network_send: a DATA-ACK packet with ack=15559
network_recv: a DATA packet with seq=15559, len=536 and
timestamp=3230007857
network_send: a DATA-ACK packet with ack=16095
network_recv: a DATA packet with seq=16095, len=536 and
timestamp=3230007858
network_send: a DATA-ACK packet with ack=16631
network_recv: a DATA packet with seq=16631, len=536 and
timestamp=3230007859
network_send: a DATA-ACK packet with ack=17167
network_recv: a DATA packet with seq=17167, len=536 and
timestamp=3230007859
network_send: a DATA-ACK packet with ack=17703
network_recv: a DATA packet with seq=17703, len=536 and
timestamp=3230007860
network_send: a DATA-ACK packet with ack=18239
network_recv: a DATA packet with seq=18239, len=392 and
timestamp=3230007860
network_send: a DATA-ACK packet with ack=18631
network_recv: a DATA packet with seq=18631, len=536 and
timestamp=3230007861
network_send: a DATA-ACK packet with ack=19167
network_recv: a DATA packet with seq=19167, len=536 and
timestamp=3230007861
network_send: a DATA-ACK packet with ack=19703
network_recv: a DATA packet with seq=19703, len=536 and
timestamp=3230007862
network_send: a DATA-ACK packet with ack=20239
network_recv: a DATA packet with seq=20239, len=536 and
timestamp=3230007862
network_send: a DATA-ACK packet with ack=20775
network_recv: a DATA packet with seq=20775, len=536 and
timestamp=3230007863
network_send: a DATA-ACK packet with ack=21311
network_recv: a DATA packet with seq=21311, len=392 and
timestamp=3230007864
network_send: a DATA-ACK packet with ack=21703
network_recv: a DATA packet with seq=21703, len=536 and
timestamp=3230007864
network_send: a DATA-ACK packet with ack=22239
network_recv: a DATA packet with seq=22239, len=536 and
timestamp=3230007865
network_send: a DATA-ACK packet with ack=22775
network_recv: a DATA packet with seq=22775, len=536 and
timestamp=3230007865
network_send: a DATA-ACK packet with ack=23311
network_recv: a DATA packet with seq=23311, len=536 and
timestamp=3230007866
network_send: a DATA-ACK packet with ack=23847
network_recv: a DATA packet with seq=23847, len=536 and
timestamp=3230007867
network_send: a DATA-ACK packet with ack=24383
network_recv: a DATA packet with seq=24383, len=392 and
timestamp=3230007867
network_send: a DATA-ACK packet with ack=24775
network_recv: a DATA packet with seq=24775, len=536 and
timestamp=3230007868
network_send: a DATA-ACK packet with ack=25311
network_recv: a DATA packet with seq=25311, len=536 and
timestamp=3230007868
network_send: a DATA-ACK packet with ack=25847
network_recv: a DATA packet with seq=25847, len=536 and
timestamp=3230007869
network_send: a DATA-ACK packet with ack=26383
network_recv: a DATA packet with seq=26383, len=536 and
timestamp=3230007870
network_send: a DATA-ACK packet with ack=26919
network_recv: a DATA packet with seq=26919, len=536 and
timestamp=3230007870
network_send: a DATA-ACK packet with ack=27455
network_recv: a DATA packet with seq=27455, len=392 and
timestamp=3230007871
network_send: a DATA-ACK packet with ack=27847
network_recv: a DATA packet with seq=27847, len=536 and
timestamp=3230007871
network_recv: a DATA packet with seq=28383, len=536 and
timestamp=3230007872
network_send: a DATA-ACK packet with ack=28919
network_recv: a DATA packet with seq=28919, len=536 and
timestamp=3230007872
network_send: a DATA-ACK packet with ack=29455
network_recv: a DATA packet with seq=29455, len=536 and
timestamp=3230007873
network_send: a DATA-ACK packet with ack=29991
network_recv: a DATA packet with seq=29991, len=536 and
timestamp=3230007873
network_send: a DATA-ACK packet with ack=30527
network_recv: a DATA packet with seq=30527, len=392 and
timestamp=3230007874
network_send: a DATA-ACK packet with ack=30919
network_recv: a DATA packet with seq=30919, len=536 and
timestamp=3230007874
network_send: a DATA-ACK packet with ack=31455
network_recv: a DATA packet with seq=31455, len=536 and
timestamp=3230007875
network_send: a DATA-ACK packet with ack=31991
network_recv: a DATA packet with seq=31991, len=536 and
timestamp=3230007876
network_send: a DATA-ACK packet with ack=32527
network_recv: a DATA packet with seq=32527, len=536 and
timestamp=3230008276
network_send: a DATA-ACK packet with ack=33063
network_recv: a DATA packet with seq=33063, len=536 and
timestamp=3230008277
network_send: a DATA-ACK packet with ack=33599
network_recv: a DATA packet with seq=33599, len=392 and
timestamp=3230008278
network_send: a DATA-ACK packet with ack=33991
network_recv: a DATA packet with seq=33991, len=536 and
timestamp=3230008279
network_send: a DATA-ACK packet with ack=34527
network_recv: a DATA packet with seq=34527, len=536 and
timestamp=3230008279
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=31991, len=536 and
timestamp=3230008279
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=32527, len=536 and
timestamp=3230008280
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=33063, len=536 and
timestamp=3230008280
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=33599, len=392 and
timestamp=3230008280
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=33991, len=536 and
timestamp=3230008281
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=34527, len=536 and
timestamp=3230008281
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=31991, len=536 and
timestamp=3230008281
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=32527, len=536 and
timestamp=3230008282
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=33063, len=536 and
timestamp=3230008282
network_send: a DATA-ACK packet with ack=35063
network_recv: a DATA packet with seq=33599, len=392 and
timestamp=3230008282
network_send: a DATA-ACK packet with ack=35063


---
 (INCORRECT!)


===============


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: TO TAs--Re: test script problem ?
Date: Fri, 1 Feb 2002 08:51:37 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2990

I too submitted my test a couple hours ago but no reply yet.There was no
error during the test submission.

Can someone please look into this?
Thanks
ashmi


On 1 Feb 2002, Shu Kit Francis Chan wrote:

>
> I have submitted the test for more than a couple of hours. Still I have
> not got any result back. Is it down ?
>
> Is the following command right ?
>
> /usr/class/cs244a/bin/submit_test.pl hw2.A <TA name>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Regardin the FIN Error--need help
Date: Fri, 1 Feb 2002 09:30:06 -0800
Lines: 362
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2991

hi,
I am getting a similar error, except that I don't see any network_send
output in my test result! But I know the network_send should be
happening, because, for example, the first line of my output says
'SYN_ACK' received and second line is 'DATA_ACK received' so that
means(according to my code structure) that  I should have sent a SYN and
sent a data packet before processing SYN_ACK and DATA_ACK respectively.

An abbreviated version of the output I get:

-----------TEST SCRIPT OUTPUT-------------------
  * Test 1.H: Failure to send all packets and make sure they are ACKed
before the FIN.
  [out of 5]

[SUMMARY] NOT OK
[RESULT]  FIN sent before Acking all data

---
network_recv: a SYN-ACK packet with seq=1 and ack=2 and
timestamp=3216151756
network_recv: a DATA-ACK packet with ack=14
network_recv: a DATA packet with seq=2, len=22 and timestamp=3216151795
network_recv: a DATA packet with seq=24, len=536 and timestamp=3216151796
network_recv: a DATA packet with seq=560, len=536 and timestamp=3216151796
network_recv: a DATA packet with seq=1096, len=536 and
timestamp=3216151796.
..
..
..
..
..
network_recv: a DATA packet with seq=109450, len=536 and
timestamp=3216152781
network_recv: a DATA packet with seq=109986, len=268 and
timestamp=3216152782
network_recv: a FIN-ACK packet packet with seq=0 and ack=4294967295 and
timestamp=3216152828


---
 (INCORRECT!)


===============

    * Test 1.I: Failure to send a FIN_ACK after a FIN
  [out of 5]

[SUMMARY] NOT OK
[RESULT]  Incorrect FIN_ACK behaviour after FIN

---
network_recv: a SYN-ACK packet with seq=1 and ack=2 and
timestamp=3216151756
network_recv: a DATA-ACK packet with ack=14
network_recv: a DATA packet with seq=2, len=22 and timestamp=3216151795
network_recv: a DATA packet with seq=24, len=536 and timestamp=3216151796
network_recv: a DATA packet with seq=560, len=536 and timestamp=3216151796
..
..
..
..
..
network_recv: a DATA packet with seq=109450, len=536 and
timestamp=3216152781
network_recv: a DATA packet with seq=109986, len=268 and
timestamp=3216152782
timestamp=3216152828


---
 (INCORRECT!)
------------***********output ends******************-------------------


Any ideas anyone? Especially the TAs? I have run out of ideas!
Thanks
Ashmi





-----***************SHANKAR'S MAIL
**************--------------------------
On Fri, 1 Feb 2002, Shankar Agarwal wrote:

> Hi,
> Hi,
> I am seeing this error for my test.It says that failure to send FIN_ACK
> for FIN. But i am not seeing any FIN at the end of the send and recv
> sequence. Can you please have a loot at this sequence and tell me how it
> can be explained.
> Thanks
> Shankar
>
>     * Test 1.I: Failure to send a FIN_ACK after a FIN
>   [out of 5]
>
> [SUMMARY] NOT OK
> [RESULT]  Incorrect FIN_ACK behaviour after FIN
>
> ---
> network_send: a SYN packet with seq=198
> network_recv: a SYN-ACK packet with seq=198 and ack=199 and
> timestamp=3230007382
> network_recv: a DATA-ACK packet with ack=211
> network_recv: a DATA packet with seq=199, len=536 and timestamp=3230007841
> network_send: a DATA-ACK packet with ack=735
> network_recv: a DATA packet with seq=735, len=536 and timestamp=3230007841
> network_send: a DATA-ACK packet with ack=1271
> network_recv: a DATA packet with seq=1271, len=536 and
> timestamp=3230007842
> network_send: a DATA-ACK packet with ack=1807
> network_recv: a DATA packet with seq=1807, len=536 and
> timestamp=3230007842
> network_send: a DATA-ACK packet with ack=2343
> network_recv: a DATA packet with seq=2343, len=536 and
> timestamp=3230007843
> network_send: a DATA-ACK packet with ack=2879
> network_recv: a DATA packet with seq=2879, len=392 and
> timestamp=3230007843
> network_send: a DATA-ACK packet with ack=3271
> network_recv: a DATA packet with seq=3271, len=536 and
> timestamp=3230007844
> network_send: a DATA-ACK packet with ack=3807
> network_recv: a DATA packet with seq=3807, len=536 and
> timestamp=3230007844
> network_send: a DATA-ACK packet with ack=4343
> network_recv: a DATA packet with seq=4343, len=536 and
> timestamp=3230007845
> network_send: a DATA-ACK packet with ack=4879
> network_recv: a DATA packet with seq=4879, len=536 and
> timestamp=3230007846
> network_send: a DATA-ACK packet with ack=5415
> network_recv: a DATA packet with seq=5415, len=536 and
> timestamp=3230007846
> network_send: a DATA-ACK packet with ack=5951
> network_recv: a DATA packet with seq=5951, len=392 and
> timestamp=3230007847
> network_send: a DATA-ACK packet with ack=6343
> network_recv: a DATA packet with seq=6343, len=536 and
> timestamp=3230007847
> network_send: a DATA-ACK packet with ack=6879
> network_recv: a DATA packet with seq=6879, len=536 and
> timestamp=3230007848
> network_send: a DATA-ACK packet with ack=7415
> network_recv: a DATA packet with seq=7415, len=536 and
> timestamp=3230007849
> network_send: a DATA-ACK packet with ack=7951
> network_recv: a DATA packet with seq=7951, len=536 and
> timestamp=3230007849
> network_send: a DATA-ACK packet with ack=8487
> network_recv: a DATA packet with seq=8487, len=536 and
> timestamp=3230007850
> network_send: a DATA-ACK packet with ack=9023
> network_recv: a DATA packet with seq=9023, len=392 and
> timestamp=3230007850
> network_send: a DATA-ACK packet with ack=9415
> network_recv: a DATA packet with seq=9415, len=536 and
> timestamp=3230007851
> network_send: a DATA-ACK packet with ack=9951
> network_recv: a DATA packet with seq=9951, len=536 and
> timestamp=3230007851
> network_send: a DATA-ACK packet with ack=10487
> network_recv: a DATA packet with seq=10487, len=536 and
> timestamp=3230007852
> network_send: a DATA-ACK packet with ack=11023
> network_recv: a DATA packet with seq=11023, len=536 and
> timestamp=3230007852
> network_send: a DATA-ACK packet with ack=11559
> network_recv: a DATA packet with seq=11559, len=536 and
> timestamp=3230007853
> network_send: a DATA-ACK packet with ack=12095
> network_recv: a DATA packet with seq=12095, len=392 and
> timestamp=3230007854
> network_send: a DATA-ACK packet with ack=12487
> network_recv: a DATA packet with seq=12487, len=536 and
> timestamp=3230007854
> network_send: a DATA-ACK packet with ack=13023
> network_recv: a DATA packet with seq=13023, len=536 and
> timestamp=3230007855
> network_send: a DATA-ACK packet with ack=13559
> network_recv: a DATA packet with seq=13559, len=536 and
> timestamp=3230007855
> network_send: a DATA-ACK packet with ack=14095
> network_recv: a DATA packet with seq=14095, len=536 and
> timestamp=3230007856
> network_send: a DATA-ACK packet with ack=14631
> network_recv: a DATA packet with seq=14631, len=536 and
> timestamp=3230007856
> network_send: a DATA-ACK packet with ack=15167
> network_recv: a DATA packet with seq=15167, len=392 and
> timestamp=3230007857
> network_send: a DATA-ACK packet with ack=15559
> network_recv: a DATA packet with seq=15559, len=536 and
> timestamp=3230007857
> network_send: a DATA-ACK packet with ack=16095
> network_recv: a DATA packet with seq=16095, len=536 and
> timestamp=3230007858
> network_send: a DATA-ACK packet with ack=16631
> network_recv: a DATA packet with seq=16631, len=536 and
> timestamp=3230007859
> network_send: a DATA-ACK packet with ack=17167
> network_recv: a DATA packet with seq=17167, len=536 and
> timestamp=3230007859
> network_send: a DATA-ACK packet with ack=17703
> network_recv: a DATA packet with seq=17703, len=536 and
> timestamp=3230007860
> network_send: a DATA-ACK packet with ack=18239
> network_recv: a DATA packet with seq=18239, len=392 and
> timestamp=3230007860
> network_send: a DATA-ACK packet with ack=18631
> network_recv: a DATA packet with seq=18631, len=536 and
> timestamp=3230007861
> network_send: a DATA-ACK packet with ack=19167
> network_recv: a DATA packet with seq=19167, len=536 and
> timestamp=3230007861
> network_send: a DATA-ACK packet with ack=19703
> network_recv: a DATA packet with seq=19703, len=536 and
> timestamp=3230007862
> network_send: a DATA-ACK packet with ack=20239
> network_recv: a DATA packet with seq=20239, len=536 and
> timestamp=3230007862
> network_send: a DATA-ACK packet with ack=20775
> network_recv: a DATA packet with seq=20775, len=536 and
> timestamp=3230007863
> network_send: a DATA-ACK packet with ack=21311
> network_recv: a DATA packet with seq=21311, len=392 and
> timestamp=3230007864
> network_send: a DATA-ACK packet with ack=21703
> network_recv: a DATA packet with seq=21703, len=536 and
> timestamp=3230007864
> network_send: a DATA-ACK packet with ack=22239
> network_recv: a DATA packet with seq=22239, len=536 and
> timestamp=3230007865
> network_send: a DATA-ACK packet with ack=22775
> network_recv: a DATA packet with seq=22775, len=536 and
> timestamp=3230007865
> network_send: a DATA-ACK packet with ack=23311
> network_recv: a DATA packet with seq=23311, len=536 and
> timestamp=3230007866
> network_send: a DATA-ACK packet with ack=23847
> network_recv: a DATA packet with seq=23847, len=536 and
> timestamp=3230007867
> network_send: a DATA-ACK packet with ack=24383
> network_recv: a DATA packet with seq=24383, len=392 and
> timestamp=3230007867
> network_send: a DATA-ACK packet with ack=24775
> network_recv: a DATA packet with seq=24775, len=536 and
> timestamp=3230007868
> network_send: a DATA-ACK packet with ack=25311
> network_recv: a DATA packet with seq=25311, len=536 and
> timestamp=3230007868
> network_send: a DATA-ACK packet with ack=25847
> network_recv: a DATA packet with seq=25847, len=536 and
> timestamp=3230007869
> network_send: a DATA-ACK packet with ack=26383
> network_recv: a DATA packet with seq=26383, len=536 and
> timestamp=3230007870
> network_send: a DATA-ACK packet with ack=26919
> network_recv: a DATA packet with seq=26919, len=536 and
> timestamp=3230007870
> network_send: a DATA-ACK packet with ack=27455
> network_recv: a DATA packet with seq=27455, len=392 and
> timestamp=3230007871
> network_send: a DATA-ACK packet with ack=27847
> network_recv: a DATA packet with seq=27847, len=536 and
> timestamp=3230007871
> network_recv: a DATA packet with seq=28383, len=536 and
> timestamp=3230007872
> network_send: a DATA-ACK packet with ack=28919
> network_recv: a DATA packet with seq=28919, len=536 and
> timestamp=3230007872
> network_send: a DATA-ACK packet with ack=29455
> network_recv: a DATA packet with seq=29455, len=536 and
> timestamp=3230007873
> network_send: a DATA-ACK packet with ack=29991
> network_recv: a DATA packet with seq=29991, len=536 and
> timestamp=3230007873
> network_send: a DATA-ACK packet with ack=30527
> network_recv: a DATA packet with seq=30527, len=392 and
> timestamp=3230007874
> network_send: a DATA-ACK packet with ack=30919
> network_recv: a DATA packet with seq=30919, len=536 and
> timestamp=3230007874
> network_send: a DATA-ACK packet with ack=31455
> network_recv: a DATA packet with seq=31455, len=536 and
> timestamp=3230007875
> network_send: a DATA-ACK packet with ack=31991
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230007876
> network_send: a DATA-ACK packet with ack=32527
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008276
> network_send: a DATA-ACK packet with ack=33063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008277
> network_send: a DATA-ACK packet with ack=33599
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008278
> network_send: a DATA-ACK packet with ack=33991
> network_recv: a DATA packet with seq=33991, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=34527
> network_recv: a DATA packet with seq=34527, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33991, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=34527, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
>
>
> ---
>  (INCORRECT!)
>
>
> ===============
>
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: Re: Regardin the FIN Error.
Date: Fri, 1 Feb 2002 09:36:54 -0800
Lines: 278
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:2992


I have the same problem. Actually, I tested this in my local folder, which
works fine because the active part will print "get FIN_ACK" and the
passive part print "send FIN_ACK" after FIN_ACK id received and send.


I guess the script use student's STCP as Tx while standoard STCP as Rx.
The server sends FIN_ACK< but somehow (maybe test connection is too busy)
the FIN_ACK gets lost . So the client will time out and exit without
getting a FIN_ACK.

Is there any clarification on this? Thanks a  lot!!


On Fri, 1 Feb 2002, Shankar Agarwal wrote:

> Hi,
> Hi,
> I am seeing this error for my test.It says that failure to send FIN_ACK
> for FIN. But i am not seeing any FIN at the end of the send and recv
> sequence. Can you please have a loot at this sequence and tell me how it
> can be explained.
> Thanks
> Shankar
>
>     * Test 1.I: Failure to send a FIN_ACK after a FIN
>   [out of 5]
>
> [SUMMARY] NOT OK
> [RESULT]  Incorrect FIN_ACK behaviour after FIN
>
> ---
> network_send: a SYN packet with seq=198
> network_recv: a SYN-ACK packet with seq=198 and ack=199 and
> timestamp=3230007382
> network_recv: a DATA-ACK packet with ack=211
> network_recv: a DATA packet with seq=199, len=536 and timestamp=3230007841
> network_send: a DATA-ACK packet with ack=735
> network_recv: a DATA packet with seq=735, len=536 and timestamp=3230007841
> network_send: a DATA-ACK packet with ack=1271
> network_recv: a DATA packet with seq=1271, len=536 and
> timestamp=3230007842
> network_send: a DATA-ACK packet with ack=1807
> network_recv: a DATA packet with seq=1807, len=536 and
> timestamp=3230007842
> network_send: a DATA-ACK packet with ack=2343
> network_recv: a DATA packet with seq=2343, len=536 and
> timestamp=3230007843
> network_send: a DATA-ACK packet with ack=2879
> network_recv: a DATA packet with seq=2879, len=392 and
> timestamp=3230007843
> network_send: a DATA-ACK packet with ack=3271
> network_recv: a DATA packet with seq=3271, len=536 and
> timestamp=3230007844
> network_send: a DATA-ACK packet with ack=3807
> network_recv: a DATA packet with seq=3807, len=536 and
> timestamp=3230007844
> network_send: a DATA-ACK packet with ack=4343
> network_recv: a DATA packet with seq=4343, len=536 and
> timestamp=3230007845
> network_send: a DATA-ACK packet with ack=4879
> network_recv: a DATA packet with seq=4879, len=536 and
> timestamp=3230007846
> network_send: a DATA-ACK packet with ack=5415
> network_recv: a DATA packet with seq=5415, len=536 and
> timestamp=3230007846
> network_send: a DATA-ACK packet with ack=5951
> network_recv: a DATA packet with seq=5951, len=392 and
> timestamp=3230007847
> network_send: a DATA-ACK packet with ack=6343
> network_recv: a DATA packet with seq=6343, len=536 and
> timestamp=3230007847
> network_send: a DATA-ACK packet with ack=6879
> network_recv: a DATA packet with seq=6879, len=536 and
> timestamp=3230007848
> network_send: a DATA-ACK packet with ack=7415
> network_recv: a DATA packet with seq=7415, len=536 and
> timestamp=3230007849
> network_send: a DATA-ACK packet with ack=7951
> network_recv: a DATA packet with seq=7951, len=536 and
> timestamp=3230007849
> network_send: a DATA-ACK packet with ack=8487
> network_recv: a DATA packet with seq=8487, len=536 and
> timestamp=3230007850
> network_send: a DATA-ACK packet with ack=9023
> network_recv: a DATA packet with seq=9023, len=392 and
> timestamp=3230007850
> network_send: a DATA-ACK packet with ack=9415
> network_recv: a DATA packet with seq=9415, len=536 and
> timestamp=3230007851
> network_send: a DATA-ACK packet with ack=9951
> network_recv: a DATA packet with seq=9951, len=536 and
> timestamp=3230007851
> network_send: a DATA-ACK packet with ack=10487
> network_recv: a DATA packet with seq=10487, len=536 and
> timestamp=3230007852
> network_send: a DATA-ACK packet with ack=11023
> network_recv: a DATA packet with seq=11023, len=536 and
> timestamp=3230007852
> network_send: a DATA-ACK packet with ack=11559
> network_recv: a DATA packet with seq=11559, len=536 and
> timestamp=3230007853
> network_send: a DATA-ACK packet with ack=12095
> network_recv: a DATA packet with seq=12095, len=392 and
> timestamp=3230007854
> network_send: a DATA-ACK packet with ack=12487
> network_recv: a DATA packet with seq=12487, len=536 and
> timestamp=3230007854
> network_send: a DATA-ACK packet with ack=13023
> network_recv: a DATA packet with seq=13023, len=536 and
> timestamp=3230007855
> network_send: a DATA-ACK packet with ack=13559
> network_recv: a DATA packet with seq=13559, len=536 and
> timestamp=3230007855
> network_send: a DATA-ACK packet with ack=14095
> network_recv: a DATA packet with seq=14095, len=536 and
> timestamp=3230007856
> network_send: a DATA-ACK packet with ack=14631
> network_recv: a DATA packet with seq=14631, len=536 and
> timestamp=3230007856
> network_send: a DATA-ACK packet with ack=15167
> network_recv: a DATA packet with seq=15167, len=392 and
> timestamp=3230007857
> network_send: a DATA-ACK packet with ack=15559
> network_recv: a DATA packet with seq=15559, len=536 and
> timestamp=3230007857
> network_send: a DATA-ACK packet with ack=16095
> network_recv: a DATA packet with seq=16095, len=536 and
> timestamp=3230007858
> network_send: a DATA-ACK packet with ack=16631
> network_recv: a DATA packet with seq=16631, len=536 and
> timestamp=3230007859
> network_send: a DATA-ACK packet with ack=17167
> network_recv: a DATA packet with seq=17167, len=536 and
> timestamp=3230007859
> network_send: a DATA-ACK packet with ack=17703
> network_recv: a DATA packet with seq=17703, len=536 and
> timestamp=3230007860
> network_send: a DATA-ACK packet with ack=18239
> network_recv: a DATA packet with seq=18239, len=392 and
> timestamp=3230007860
> network_send: a DATA-ACK packet with ack=18631
> network_recv: a DATA packet with seq=18631, len=536 and
> timestamp=3230007861
> network_send: a DATA-ACK packet with ack=19167
> network_recv: a DATA packet with seq=19167, len=536 and
> timestamp=3230007861
> network_send: a DATA-ACK packet with ack=19703
> network_recv: a DATA packet with seq=19703, len=536 and
> timestamp=3230007862
> network_send: a DATA-ACK packet with ack=20239
> network_recv: a DATA packet with seq=20239, len=536 and
> timestamp=3230007862
> network_send: a DATA-ACK packet with ack=20775
> network_recv: a DATA packet with seq=20775, len=536 and
> timestamp=3230007863
> network_send: a DATA-ACK packet with ack=21311
> network_recv: a DATA packet with seq=21311, len=392 and
> timestamp=3230007864
> network_send: a DATA-ACK packet with ack=21703
> network_recv: a DATA packet with seq=21703, len=536 and
> timestamp=3230007864
> network_send: a DATA-ACK packet with ack=22239
> network_recv: a DATA packet with seq=22239, len=536 and
> timestamp=3230007865
> network_send: a DATA-ACK packet with ack=22775
> network_recv: a DATA packet with seq=22775, len=536 and
> timestamp=3230007865
> network_send: a DATA-ACK packet with ack=23311
> network_recv: a DATA packet with seq=23311, len=536 and
> timestamp=3230007866
> network_send: a DATA-ACK packet with ack=23847
> network_recv: a DATA packet with seq=23847, len=536 and
> timestamp=3230007867
> network_send: a DATA-ACK packet with ack=24383
> network_recv: a DATA packet with seq=24383, len=392 and
> timestamp=3230007867
> network_send: a DATA-ACK packet with ack=24775
> network_recv: a DATA packet with seq=24775, len=536 and
> timestamp=3230007868
> network_send: a DATA-ACK packet with ack=25311
> network_recv: a DATA packet with seq=25311, len=536 and
> timestamp=3230007868
> network_send: a DATA-ACK packet with ack=25847
> network_recv: a DATA packet with seq=25847, len=536 and
> timestamp=3230007869
> network_send: a DATA-ACK packet with ack=26383
> network_recv: a DATA packet with seq=26383, len=536 and
> timestamp=3230007870
> network_send: a DATA-ACK packet with ack=26919
> network_recv: a DATA packet with seq=26919, len=536 and
> timestamp=3230007870
> network_send: a DATA-ACK packet with ack=27455
> network_recv: a DATA packet with seq=27455, len=392 and
> timestamp=3230007871
> network_send: a DATA-ACK packet with ack=27847
> network_recv: a DATA packet with seq=27847, len=536 and
> timestamp=3230007871
> network_recv: a DATA packet with seq=28383, len=536 and
> timestamp=3230007872
> network_send: a DATA-ACK packet with ack=28919
> network_recv: a DATA packet with seq=28919, len=536 and
> timestamp=3230007872
> network_send: a DATA-ACK packet with ack=29455
> network_recv: a DATA packet with seq=29455, len=536 and
> timestamp=3230007873
> network_send: a DATA-ACK packet with ack=29991
> network_recv: a DATA packet with seq=29991, len=536 and
> timestamp=3230007873
> network_send: a DATA-ACK packet with ack=30527
> network_recv: a DATA packet with seq=30527, len=392 and
> timestamp=3230007874
> network_send: a DATA-ACK packet with ack=30919
> network_recv: a DATA packet with seq=30919, len=536 and
> timestamp=3230007874
> network_send: a DATA-ACK packet with ack=31455
> network_recv: a DATA packet with seq=31455, len=536 and
> timestamp=3230007875
> network_send: a DATA-ACK packet with ack=31991
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230007876
> network_send: a DATA-ACK packet with ack=32527
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008276
> network_send: a DATA-ACK packet with ack=33063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008277
> network_send: a DATA-ACK packet with ack=33599
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008278
> network_send: a DATA-ACK packet with ack=33991
> network_recv: a DATA packet with seq=33991, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=34527
> network_recv: a DATA packet with seq=34527, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33991, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=34527, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
>
>
> ---
>  (INCORRECT!)
>
>
> ===============
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: testscripts not working
Date: Fri, 1 Feb 2002 10:01:49 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine43.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2993

Hi,

I havent got response to last 2 of my test runs - 1 I fired last night (10
hours back), the other I fired around 1 hour back.

Is anybody else seeing the problem?

thanks
-anurag


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Testscripts definitely down!
Date: Fri, 1 Feb 2002 10:08:12 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:2994

hi
i submitted a test like 3 hours ago, no reply yet.
what's happening?
~ashmi

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: test script up and running
Date: Fri, 01 Feb 2002 10:25:07 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:2995

Hi,

I have restarted the test script. It seems to have died yesterday at 
2:30am.There is a backlog of 21 students, so please be patient when 
waiting for your results.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: submit.pl
Date: Fri, 1 Feb 2002 10:26:15 -0800
Lines: 50
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0166_01C1AB0A.E0E0B510"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:2996

This is a multi-part message in MIME format.

------=_NextPart_000_0166_01C1AB0A.E0E0B510
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


I'm getting this error with submit script...

saga13:~/proj2a> /usr/class/cs244a/bin/submit.pl hw2.A akishan
SubmitBase: /afs/ir/class/cs244a/submissions
Debug Level =3D 1
fs: You don't have the required access rights on =
'/afs/ir/class/cs244a/submissions/grading/hw2.A/akishan/hpbae'
fs: You don't have the required access rights on =
'/afs/ir/class/cs244a/submissions/grading/hw2.A/akishan/hpbae'


------=_NextPart_000_0166_01C1AB0A.E0E0B510
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>I'm getting this error with submit=20
script...</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>saga13:~/proj2a> =
/usr/class/cs244a/bin/submit.pl=20
hw2.A akishan<BR>SubmitBase: /afs/ir/class/cs244a/submissions<BR>Debug =
Level =3D=20
1<BR>fs: You don't have the required access rights on=20
'/afs/ir/class/cs244a/submissions/grading/hw2.A/akishan/hpbae'<BR>fs: =
You don't=20
have the required access rights on=20
'/afs/ir/class/cs244a/submissions/grading/hw2.A/akishan/hpbae'</FONT></DI=
V>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_0166_01C1AB0A.E0E0B510--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: help!! script errors!!
Date: Fri, 01 Feb 2002 10:38:02 -0800
Lines: 59
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2997

Maybe your transmitter is sending data too fast, UDP is dropping some of 
the datagrams, and their receiver is blocking on the read (because 
they're not getting as many bytes as they expect), and ultimately they 
kill the process and report that you gave them nothing.

Try adding a tiny delay before each network_send in your transmitter.

I ran into something like this, although for me they at least said that 
I transferred 0 bytes of TESTDATA.0.

Good luck.

-Sandeep

Mark Joseph Dolan wrote:

> Hi, if anybody has any suggestions they would be greatly appreciated!!
> 
> I've been having this problem for a day now and I can't seem to solve it.
> I pass all the tests except T1.B and T1.C which fail completly.
> And no matter what I do I can't seem cause my code to run incorrectly, I even
> created files of the same sizes as the tests. But this
> 
> 
>>===(T1.Ba) Byte count differences found (INCORRECT!)
>>Student's Tx against our Rx (reliable mode)
>>
>>---
>>
>>---
>>Difference between expected result and your result:
>>0a1,3
>>
>>>   2666   15720  110230 TESTDATA.0
>>>   1812   10049  299328 TESTDATA.1
>>>    131     501    5098 TESTDATA.2
>>>
>>---
>>
>>===========================
>>
> 
> suggests to me that I'm not sending/recieving any of them. 
> Any ideas?
> 
> Thank you
> 
> -mark
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: multiple connections
Date: Fri, 01 Feb 2002 10:41:00 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:2998

You are reading a little too much into this.  What they really mean is 
that you're doing a select over your one local data socket and one 
"external entity" socket.

-Sandeep

Victor Wong wrote:

> Hi,
> 
> So I think I'm confused by a pretty fundamental question. When multiple connections between the server and client are opened, a child process is spawned for each end of each connection. However, the block of comments before the control_loop function states that the control loop should use select to see if "any data on any of the sockets to our peer, ... etc." is available. So am I operating under the false impression that each child process is only supposed to take care of one end of one connection and tha
> t the network layer would route the correct packet to the rightchild process? Or am I reading too into the comments?
> 
> I guess what I don't understand is how multiple connections the same port via UDP works. It seems like as UDP is connectionless, the packets from different connetions will be arriving multiplexed at the server post.
> 
> Help?!
> 
> Thanks,
> Victor.
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Connection termination
Date: 1 Feb 2002 18:43:50 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:2999

 writes:

>If I send a FIN, should I just ignore all packets that are not a FIN or
>FIN+ACK. In normal TCP I'd still be required to receive and acknowledge
>data packets but that is because both sides need to agree on the
>termination.

>Thanks,
>Josh


Please see the discussion about network termination in the assignment
specification.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Regardin the FIN Error.
Date: Fri, 1 Feb 2002 10:46:25 -0800
Lines: 275
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012589186 2565 128.12.133.48 (1 Feb 2002 18:46:26 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3000

This is interesting. If the output here is the output from YOUR code (I was
assuming it was the view from the script's transmitter/receiver), then it
shows that the script is indeed doing a retransmit like I thought. It looks
like your receiver is ACKing those correctly as already received packets,
but only to a point. Notice that the last packet sent has a sequence number
far lower than the size of the test script files. (Which we've seen from
earlier posts are 100K and 300K.) Make sure the repeated retransmits from
the script's transmitter aren't crashing your receiver.

Jonathan

"Shankar Agarwal"  wrote in message

> Hi,
> Hi,
> I am seeing this error for my test.It says that failure to send FIN_ACK
> for FIN. But i am not seeing any FIN at the end of the send and recv
> sequence. Can you please have a loot at this sequence and tell me how it
> can be explained.
> Thanks
> Shankar
>
>     * Test 1.I: Failure to send a FIN_ACK after a FIN
>   [out of 5]
>
> [SUMMARY] NOT OK
> [RESULT]  Incorrect FIN_ACK behaviour after FIN
>
> ---
> network_send: a SYN packet with seq=198
> network_recv: a SYN-ACK packet with seq=198 and ack=199 and
> timestamp=3230007382
> network_recv: a DATA-ACK packet with ack=211
> network_recv: a DATA packet with seq=199, len=536 and timestamp=3230007841
> network_send: a DATA-ACK packet with ack=735
> network_recv: a DATA packet with seq=735, len=536 and timestamp=3230007841
> network_send: a DATA-ACK packet with ack=1271
> network_recv: a DATA packet with seq=1271, len=536 and
> timestamp=3230007842
> network_send: a DATA-ACK packet with ack=1807
> network_recv: a DATA packet with seq=1807, len=536 and
> timestamp=3230007842
> network_send: a DATA-ACK packet with ack=2343
> network_recv: a DATA packet with seq=2343, len=536 and
> timestamp=3230007843
> network_send: a DATA-ACK packet with ack=2879
> network_recv: a DATA packet with seq=2879, len=392 and
> timestamp=3230007843
> network_send: a DATA-ACK packet with ack=3271
> network_recv: a DATA packet with seq=3271, len=536 and
> timestamp=3230007844
> network_send: a DATA-ACK packet with ack=3807
> network_recv: a DATA packet with seq=3807, len=536 and
> timestamp=3230007844
> network_send: a DATA-ACK packet with ack=4343
> network_recv: a DATA packet with seq=4343, len=536 and
> timestamp=3230007845
> network_send: a DATA-ACK packet with ack=4879
> network_recv: a DATA packet with seq=4879, len=536 and
> timestamp=3230007846
> network_send: a DATA-ACK packet with ack=5415
> network_recv: a DATA packet with seq=5415, len=536 and
> timestamp=3230007846
> network_send: a DATA-ACK packet with ack=5951
> network_recv: a DATA packet with seq=5951, len=392 and
> timestamp=3230007847
> network_send: a DATA-ACK packet with ack=6343
> network_recv: a DATA packet with seq=6343, len=536 and
> timestamp=3230007847
> network_send: a DATA-ACK packet with ack=6879
> network_recv: a DATA packet with seq=6879, len=536 and
> timestamp=3230007848
> network_send: a DATA-ACK packet with ack=7415
> network_recv: a DATA packet with seq=7415, len=536 and
> timestamp=3230007849
> network_send: a DATA-ACK packet with ack=7951
> network_recv: a DATA packet with seq=7951, len=536 and
> timestamp=3230007849
> network_send: a DATA-ACK packet with ack=8487
> network_recv: a DATA packet with seq=8487, len=536 and
> timestamp=3230007850
> network_send: a DATA-ACK packet with ack=9023
> network_recv: a DATA packet with seq=9023, len=392 and
> timestamp=3230007850
> network_send: a DATA-ACK packet with ack=9415
> network_recv: a DATA packet with seq=9415, len=536 and
> timestamp=3230007851
> network_send: a DATA-ACK packet with ack=9951
> network_recv: a DATA packet with seq=9951, len=536 and
> timestamp=3230007851
> network_send: a DATA-ACK packet with ack=10487
> network_recv: a DATA packet with seq=10487, len=536 and
> timestamp=3230007852
> network_send: a DATA-ACK packet with ack=11023
> network_recv: a DATA packet with seq=11023, len=536 and
> timestamp=3230007852
> network_send: a DATA-ACK packet with ack=11559
> network_recv: a DATA packet with seq=11559, len=536 and
> timestamp=3230007853
> network_send: a DATA-ACK packet with ack=12095
> network_recv: a DATA packet with seq=12095, len=392 and
> timestamp=3230007854
> network_send: a DATA-ACK packet with ack=12487
> network_recv: a DATA packet with seq=12487, len=536 and
> timestamp=3230007854
> network_send: a DATA-ACK packet with ack=13023
> network_recv: a DATA packet with seq=13023, len=536 and
> timestamp=3230007855
> network_send: a DATA-ACK packet with ack=13559
> network_recv: a DATA packet with seq=13559, len=536 and
> timestamp=3230007855
> network_send: a DATA-ACK packet with ack=14095
> network_recv: a DATA packet with seq=14095, len=536 and
> timestamp=3230007856
> network_send: a DATA-ACK packet with ack=14631
> network_recv: a DATA packet with seq=14631, len=536 and
> timestamp=3230007856
> network_send: a DATA-ACK packet with ack=15167
> network_recv: a DATA packet with seq=15167, len=392 and
> timestamp=3230007857
> network_send: a DATA-ACK packet with ack=15559
> network_recv: a DATA packet with seq=15559, len=536 and
> timestamp=3230007857
> network_send: a DATA-ACK packet with ack=16095
> network_recv: a DATA packet with seq=16095, len=536 and
> timestamp=3230007858
> network_send: a DATA-ACK packet with ack=16631
> network_recv: a DATA packet with seq=16631, len=536 and
> timestamp=3230007859
> network_send: a DATA-ACK packet with ack=17167
> network_recv: a DATA packet with seq=17167, len=536 and
> timestamp=3230007859
> network_send: a DATA-ACK packet with ack=17703
> network_recv: a DATA packet with seq=17703, len=536 and
> timestamp=3230007860
> network_send: a DATA-ACK packet with ack=18239
> network_recv: a DATA packet with seq=18239, len=392 and
> timestamp=3230007860
> network_send: a DATA-ACK packet with ack=18631
> network_recv: a DATA packet with seq=18631, len=536 and
> timestamp=3230007861
> network_send: a DATA-ACK packet with ack=19167
> network_recv: a DATA packet with seq=19167, len=536 and
> timestamp=3230007861
> network_send: a DATA-ACK packet with ack=19703
> network_recv: a DATA packet with seq=19703, len=536 and
> timestamp=3230007862
> network_send: a DATA-ACK packet with ack=20239
> network_recv: a DATA packet with seq=20239, len=536 and
> timestamp=3230007862
> network_send: a DATA-ACK packet with ack=20775
> network_recv: a DATA packet with seq=20775, len=536 and
> timestamp=3230007863
> network_send: a DATA-ACK packet with ack=21311
> network_recv: a DATA packet with seq=21311, len=392 and
> timestamp=3230007864
> network_send: a DATA-ACK packet with ack=21703
> network_recv: a DATA packet with seq=21703, len=536 and
> timestamp=3230007864
> network_send: a DATA-ACK packet with ack=22239
> network_recv: a DATA packet with seq=22239, len=536 and
> timestamp=3230007865
> network_send: a DATA-ACK packet with ack=22775
> network_recv: a DATA packet with seq=22775, len=536 and
> timestamp=3230007865
> network_send: a DATA-ACK packet with ack=23311
> network_recv: a DATA packet with seq=23311, len=536 and
> timestamp=3230007866
> network_send: a DATA-ACK packet with ack=23847
> network_recv: a DATA packet with seq=23847, len=536 and
> timestamp=3230007867
> network_send: a DATA-ACK packet with ack=24383
> network_recv: a DATA packet with seq=24383, len=392 and
> timestamp=3230007867
> network_send: a DATA-ACK packet with ack=24775
> network_recv: a DATA packet with seq=24775, len=536 and
> timestamp=3230007868
> network_send: a DATA-ACK packet with ack=25311
> network_recv: a DATA packet with seq=25311, len=536 and
> timestamp=3230007868
> network_send: a DATA-ACK packet with ack=25847
> network_recv: a DATA packet with seq=25847, len=536 and
> timestamp=3230007869
> network_send: a DATA-ACK packet with ack=26383
> network_recv: a DATA packet with seq=26383, len=536 and
> timestamp=3230007870
> network_send: a DATA-ACK packet with ack=26919
> network_recv: a DATA packet with seq=26919, len=536 and
> timestamp=3230007870
> network_send: a DATA-ACK packet with ack=27455
> network_recv: a DATA packet with seq=27455, len=392 and
> timestamp=3230007871
> network_send: a DATA-ACK packet with ack=27847
> network_recv: a DATA packet with seq=27847, len=536 and
> timestamp=3230007871
> network_recv: a DATA packet with seq=28383, len=536 and
> timestamp=3230007872
> network_send: a DATA-ACK packet with ack=28919
> network_recv: a DATA packet with seq=28919, len=536 and
> timestamp=3230007872
> network_send: a DATA-ACK packet with ack=29455
> network_recv: a DATA packet with seq=29455, len=536 and
> timestamp=3230007873
> network_send: a DATA-ACK packet with ack=29991
> network_recv: a DATA packet with seq=29991, len=536 and
> timestamp=3230007873
> network_send: a DATA-ACK packet with ack=30527
> network_recv: a DATA packet with seq=30527, len=392 and
> timestamp=3230007874
> network_send: a DATA-ACK packet with ack=30919
> network_recv: a DATA packet with seq=30919, len=536 and
> timestamp=3230007874
> network_send: a DATA-ACK packet with ack=31455
> network_recv: a DATA packet with seq=31455, len=536 and
> timestamp=3230007875
> network_send: a DATA-ACK packet with ack=31991
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230007876
> network_send: a DATA-ACK packet with ack=32527
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008276
> network_send: a DATA-ACK packet with ack=33063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008277
> network_send: a DATA-ACK packet with ack=33599
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008278
> network_send: a DATA-ACK packet with ack=33991
> network_recv: a DATA packet with seq=33991, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=34527
> network_recv: a DATA packet with seq=34527, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230008279
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008280
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33991, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=34527, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=31991, len=536 and
> timestamp=3230008281
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=32527, len=536 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33063, len=536 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
> network_recv: a DATA packet with seq=33599, len=392 and
> timestamp=3230008282
> network_send: a DATA-ACK packet with ack=35063
>
>
> ---
>  (INCORRECT!)
>
>
> ===============
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Regardin the FIN Error.
Date: Fri, 1 Feb 2002 10:59:45 -0800
Lines: 296
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Jonathan Keljo 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3001

I do think the output Shankar talks about is from the test script: I get a
similar error and output.
does that give anyone any ideas? pls let me know!
~ashmi

On Fri, 1 Feb 2002, Jonathan Keljo wrote:

> This is interesting. If the output here is the output from YOUR code (I was
> assuming it was the view from the script's transmitter/receiver), then it
> shows that the script is indeed doing a retransmit like I thought. It looks
> like your receiver is ACKing those correctly as already received packets,
> but only to a point. Notice that the last packet sent has a sequence number
> far lower than the size of the test script files. (Which we've seen from
> earlier posts are 100K and 300K.) Make sure the repeated retransmits from
> the script's transmitter aren't crashing your receiver.
>
> Jonathan
>
> "Shankar Agarwal"  wrote in message
> 
> > Hi,
> > Hi,
> > I am seeing this error for my test.It says that failure to send FIN_ACK
> > for FIN. But i am not seeing any FIN at the end of the send and recv
> > sequence. Can you please have a loot at this sequence and tell me how it
> > can be explained.
> > Thanks
> > Shankar
> >
> >     * Test 1.I: Failure to send a FIN_ACK after a FIN
> >   [out of 5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  Incorrect FIN_ACK behaviour after FIN
> >
> > ---
> > network_send: a SYN packet with seq=198
> > network_recv: a SYN-ACK packet with seq=198 and ack=199 and
> > timestamp=3230007382
> > network_recv: a DATA-ACK packet with ack=211
> > network_recv: a DATA packet with seq=199, len=536 and timestamp=3230007841
> > network_send: a DATA-ACK packet with ack=735
> > network_recv: a DATA packet with seq=735, len=536 and timestamp=3230007841
> > network_send: a DATA-ACK packet with ack=1271
> > network_recv: a DATA packet with seq=1271, len=536 and
> > timestamp=3230007842
> > network_send: a DATA-ACK packet with ack=1807
> > network_recv: a DATA packet with seq=1807, len=536 and
> > timestamp=3230007842
> > network_send: a DATA-ACK packet with ack=2343
> > network_recv: a DATA packet with seq=2343, len=536 and
> > timestamp=3230007843
> > network_send: a DATA-ACK packet with ack=2879
> > network_recv: a DATA packet with seq=2879, len=392 and
> > timestamp=3230007843
> > network_send: a DATA-ACK packet with ack=3271
> > network_recv: a DATA packet with seq=3271, len=536 and
> > timestamp=3230007844
> > network_send: a DATA-ACK packet with ack=3807
> > network_recv: a DATA packet with seq=3807, len=536 and
> > timestamp=3230007844
> > network_send: a DATA-ACK packet with ack=4343
> > network_recv: a DATA packet with seq=4343, len=536 and
> > timestamp=3230007845
> > network_send: a DATA-ACK packet with ack=4879
> > network_recv: a DATA packet with seq=4879, len=536 and
> > timestamp=3230007846
> > network_send: a DATA-ACK packet with ack=5415
> > network_recv: a DATA packet with seq=5415, len=536 and
> > timestamp=3230007846
> > network_send: a DATA-ACK packet with ack=5951
> > network_recv: a DATA packet with seq=5951, len=392 and
> > timestamp=3230007847
> > network_send: a DATA-ACK packet with ack=6343
> > network_recv: a DATA packet with seq=6343, len=536 and
> > timestamp=3230007847
> > network_send: a DATA-ACK packet with ack=6879
> > network_recv: a DATA packet with seq=6879, len=536 and
> > timestamp=3230007848
> > network_send: a DATA-ACK packet with ack=7415
> > network_recv: a DATA packet with seq=7415, len=536 and
> > timestamp=3230007849
> > network_send: a DATA-ACK packet with ack=7951
> > network_recv: a DATA packet with seq=7951, len=536 and
> > timestamp=3230007849
> > network_send: a DATA-ACK packet with ack=8487
> > network_recv: a DATA packet with seq=8487, len=536 and
> > timestamp=3230007850
> > network_send: a DATA-ACK packet with ack=9023
> > network_recv: a DATA packet with seq=9023, len=392 and
> > timestamp=3230007850
> > network_send: a DATA-ACK packet with ack=9415
> > network_recv: a DATA packet with seq=9415, len=536 and
> > timestamp=3230007851
> > network_send: a DATA-ACK packet with ack=9951
> > network_recv: a DATA packet with seq=9951, len=536 and
> > timestamp=3230007851
> > network_send: a DATA-ACK packet with ack=10487
> > network_recv: a DATA packet with seq=10487, len=536 and
> > timestamp=3230007852
> > network_send: a DATA-ACK packet with ack=11023
> > network_recv: a DATA packet with seq=11023, len=536 and
> > timestamp=3230007852
> > network_send: a DATA-ACK packet with ack=11559
> > network_recv: a DATA packet with seq=11559, len=536 and
> > timestamp=3230007853
> > network_send: a DATA-ACK packet with ack=12095
> > network_recv: a DATA packet with seq=12095, len=392 and
> > timestamp=3230007854
> > network_send: a DATA-ACK packet with ack=12487
> > network_recv: a DATA packet with seq=12487, len=536 and
> > timestamp=3230007854
> > network_send: a DATA-ACK packet with ack=13023
> > network_recv: a DATA packet with seq=13023, len=536 and
> > timestamp=3230007855
> > network_send: a DATA-ACK packet with ack=13559
> > network_recv: a DATA packet with seq=13559, len=536 and
> > timestamp=3230007855
> > network_send: a DATA-ACK packet with ack=14095
> > network_recv: a DATA packet with seq=14095, len=536 and
> > timestamp=3230007856
> > network_send: a DATA-ACK packet with ack=14631
> > network_recv: a DATA packet with seq=14631, len=536 and
> > timestamp=3230007856
> > network_send: a DATA-ACK packet with ack=15167
> > network_recv: a DATA packet with seq=15167, len=392 and
> > timestamp=3230007857
> > network_send: a DATA-ACK packet with ack=15559
> > network_recv: a DATA packet with seq=15559, len=536 and
> > timestamp=3230007857
> > network_send: a DATA-ACK packet with ack=16095
> > network_recv: a DATA packet with seq=16095, len=536 and
> > timestamp=3230007858
> > network_send: a DATA-ACK packet with ack=16631
> > network_recv: a DATA packet with seq=16631, len=536 and
> > timestamp=3230007859
> > network_send: a DATA-ACK packet with ack=17167
> > network_recv: a DATA packet with seq=17167, len=536 and
> > timestamp=3230007859
> > network_send: a DATA-ACK packet with ack=17703
> > network_recv: a DATA packet with seq=17703, len=536 and
> > timestamp=3230007860
> > network_send: a DATA-ACK packet with ack=18239
> > network_recv: a DATA packet with seq=18239, len=392 and
> > timestamp=3230007860
> > network_send: a DATA-ACK packet with ack=18631
> > network_recv: a DATA packet with seq=18631, len=536 and
> > timestamp=3230007861
> > network_send: a DATA-ACK packet with ack=19167
> > network_recv: a DATA packet with seq=19167, len=536 and
> > timestamp=3230007861
> > network_send: a DATA-ACK packet with ack=19703
> > network_recv: a DATA packet with seq=19703, len=536 and
> > timestamp=3230007862
> > network_send: a DATA-ACK packet with ack=20239
> > network_recv: a DATA packet with seq=20239, len=536 and
> > timestamp=3230007862
> > network_send: a DATA-ACK packet with ack=20775
> > network_recv: a DATA packet with seq=20775, len=536 and
> > timestamp=3230007863
> > network_send: a DATA-ACK packet with ack=21311
> > network_recv: a DATA packet with seq=21311, len=392 and
> > timestamp=3230007864
> > network_send: a DATA-ACK packet with ack=21703
> > network_recv: a DATA packet with seq=21703, len=536 and
> > timestamp=3230007864
> > network_send: a DATA-ACK packet with ack=22239
> > network_recv: a DATA packet with seq=22239, len=536 and
> > timestamp=3230007865
> > network_send: a DATA-ACK packet with ack=22775
> > network_recv: a DATA packet with seq=22775, len=536 and
> > timestamp=3230007865
> > network_send: a DATA-ACK packet with ack=23311
> > network_recv: a DATA packet with seq=23311, len=536 and
> > timestamp=3230007866
> > network_send: a DATA-ACK packet with ack=23847
> > network_recv: a DATA packet with seq=23847, len=536 and
> > timestamp=3230007867
> > network_send: a DATA-ACK packet with ack=24383
> > network_recv: a DATA packet with seq=24383, len=392 and
> > timestamp=3230007867
> > network_send: a DATA-ACK packet with ack=24775
> > network_recv: a DATA packet with seq=24775, len=536 and
> > timestamp=3230007868
> > network_send: a DATA-ACK packet with ack=25311
> > network_recv: a DATA packet with seq=25311, len=536 and
> > timestamp=3230007868
> > network_send: a DATA-ACK packet with ack=25847
> > network_recv: a DATA packet with seq=25847, len=536 and
> > timestamp=3230007869
> > network_send: a DATA-ACK packet with ack=26383
> > network_recv: a DATA packet with seq=26383, len=536 and
> > timestamp=3230007870
> > network_send: a DATA-ACK packet with ack=26919
> > network_recv: a DATA packet with seq=26919, len=536 and
> > timestamp=3230007870
> > network_send: a DATA-ACK packet with ack=27455
> > network_recv: a DATA packet with seq=27455, len=392 and
> > timestamp=3230007871
> > network_send: a DATA-ACK packet with ack=27847
> > network_recv: a DATA packet with seq=27847, len=536 and
> > timestamp=3230007871
> > network_recv: a DATA packet with seq=28383, len=536 and
> > timestamp=3230007872
> > network_send: a DATA-ACK packet with ack=28919
> > network_recv: a DATA packet with seq=28919, len=536 and
> > timestamp=3230007872
> > network_send: a DATA-ACK packet with ack=29455
> > network_recv: a DATA packet with seq=29455, len=536 and
> > timestamp=3230007873
> > network_send: a DATA-ACK packet with ack=29991
> > network_recv: a DATA packet with seq=29991, len=536 and
> > timestamp=3230007873
> > network_send: a DATA-ACK packet with ack=30527
> > network_recv: a DATA packet with seq=30527, len=392 and
> > timestamp=3230007874
> > network_send: a DATA-ACK packet with ack=30919
> > network_recv: a DATA packet with seq=30919, len=536 and
> > timestamp=3230007874
> > network_send: a DATA-ACK packet with ack=31455
> > network_recv: a DATA packet with seq=31455, len=536 and
> > timestamp=3230007875
> > network_send: a DATA-ACK packet with ack=31991
> > network_recv: a DATA packet with seq=31991, len=536 and
> > timestamp=3230007876
> > network_send: a DATA-ACK packet with ack=32527
> > network_recv: a DATA packet with seq=32527, len=536 and
> > timestamp=3230008276
> > network_send: a DATA-ACK packet with ack=33063
> > network_recv: a DATA packet with seq=33063, len=536 and
> > timestamp=3230008277
> > network_send: a DATA-ACK packet with ack=33599
> > network_recv: a DATA packet with seq=33599, len=392 and
> > timestamp=3230008278
> > network_send: a DATA-ACK packet with ack=33991
> > network_recv: a DATA packet with seq=33991, len=536 and
> > timestamp=3230008279
> > network_send: a DATA-ACK packet with ack=34527
> > network_recv: a DATA packet with seq=34527, len=536 and
> > timestamp=3230008279
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=31991, len=536 and
> > timestamp=3230008279
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=32527, len=536 and
> > timestamp=3230008280
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=33063, len=536 and
> > timestamp=3230008280
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=33599, len=392 and
> > timestamp=3230008280
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=33991, len=536 and
> > timestamp=3230008281
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=34527, len=536 and
> > timestamp=3230008281
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=31991, len=536 and
> > timestamp=3230008281
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=32527, len=536 and
> > timestamp=3230008282
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=33063, len=536 and
> > timestamp=3230008282
> > network_send: a DATA-ACK packet with ack=35063
> > network_recv: a DATA packet with seq=33599, len=392 and
> > timestamp=3230008282
> > network_send: a DATA-ACK packet with ack=35063
> >
> >
> > ---
> >  (INCORRECT!)
> >
> >
> > ===============
> >
> >
>
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine17.Stanford.EDU!kichan
From:  (Shu Kit Francis Chan)
Newsgroups: su.class.cs244a
Subject: test script problem again ?
Date: 1 Feb 2002 19:12:37 GMT
Organization: Stanford University, CA 94305, USA
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine17.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3002

I have waited for another 2 hrs already and I still have not got my result
back. 

Do you mean I need to resubmit again or is it just the thing is down again
?
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: select and read
Date: 1 Feb 2002 19:25:02 GMT
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3003

"Victor Wong"  writes:

>The answer to the second question on the FAQ states that we should not be
>reading any more data from the local socket if we can't send it out. This
>means that the socket will still be ready for reading after reading a
>portion of the data to fill the send queue. Doesn't this mean that the call
>to select would return immediately and the control loop would just end up
>polling to see when the ACK arrives and the window opens up? Isn't it more
>efficient to implement this with a semaphore/lock so that there's no busy
>waiting? Or should we not care?

If your sender window is full, you will probably not want to wake up on
application data ready to transmit, as FAQ#2 is meant to suggest.  select()
will only wake up for a particular event if you explicitly requested to be
notified about it.  So if you aren't interested in what the application has
to send at the moment, don't ask select() to wake you up on that condition...

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: test script problem again ?
Date: 1 Feb 2002 19:25:15 GMT
Organization: Stanford University, CA 94305, USA
Lines: 2
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3004


Yes, I've been waiting for 4 hours. Do we need to resubmit???
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: window size
Date: 1 Feb 2002 19:25:35 GMT
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3005

"Yichen Xie"  writes:

>Since th_win field in the tcp header is unused according to the
>specification, is it safe to assume that the application reads whatever is
>available to it as soon as the data has arrived (otherwise, writes to the
>local_data_sd may block at the child's end, and sender will time out, and
>all sorts of things could happen...)?

Please see the earlier responses to this question.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Mikhail Seregine 
Newsgroups: su.class.cs244a
Subject: Re: test script problem again ?
Date: Fri, 01 Feb 2002 11:36:53 -0800
Lines: 16
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3006

You probably need to resubmit; I resubmitted at 10:40 and got
a result at 10:50

Good luck,

Mikhail

Mark Joseph Dolan wrote:
> 
> Yes, I've been waiting for 4 hours. Do we need to resubmit???

-- 
----------------
Mikhail Seregine

http://www.stanford.edu/~seregine
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Regardin the FIN Error.
Date: 1 Feb 2002 19:44:17 GMT
Lines: 27
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3007

"Jonathan Keljo"  writes:

>This is interesting. If the output here is the output from YOUR code (I was
>assuming it was the view from the script's transmitter/receiver), then it
>shows that the script is indeed doing a retransmit like I thought. It looks
>like your receiver is ACKing those correctly as already received packets,
>but only to a point. Notice that the last packet sent has a sequence number
>far lower than the size of the test script files. (Which we've seen from
>earlier posts are 100K and 300K.) Make sure the repeated retransmits from
>the script's transmitter aren't crashing your receiver.

This is quite possible--the "standard" receiver/transmitter is a full
STCP implementation.  During testing, you (a generic "you" for whomever's
reading) have to be careful that a UDP packet wasn't lost during
transmission--i.e. be paranoid.  (This is a good idea during development
anyway--if you don't expect a condition, assert on it and find the problem
as early as possible).

This shouldn't normally be a problem if you're testing between elaine
machines, but as was posted previously, if you test over other networks
you might observe some packet loss/retransmissions.

However, if the problem occurs repeatedly, chances are that it isn't due
to the unreliably underlying network, but rather a problem in your code...
so asserts/debugging output will be very helpful in verifying that your
code is behaving correctly.

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: test script problem again ?
Date: Fri, 1 Feb 2002 11:47:49 -0800
Lines: 40
Distribution: su
Message-ID: 
References:  
 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3008

when i try to resubmit it says my earlier test is still pending. if i
still choose to resubmit the code it just overwrites my earlier files... i
did this twice already in the last hour and nothing's come yet

Help, TAs!!!
ashmi


On Fri, 1 Feb 2002, Mikhail Seregine wrote:

> You probably need to resubmit; I resubmitted at 10:40 and got
> a result at 10:50
>
> Good luck,
>
> Mikhail
>
> Mark Joseph Dolan wrote:
> >
> > Yes, I've been waiting for 4 hours. Do we need to resubmit???
>
> --
> ----------------
> Mikhail Seregine
> 
> http://www.stanford.edu/~seregine
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!fable14.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: Re: test script problem again ?
Date: Fri, 1 Feb 2002 11:51:14 -0800
Lines: 62
Distribution: su
Message-ID: 
References:  
  
NNTP-Posting-Host: fable14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Ashmi 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3009

Hey,

  You move yourself to the back of the queue each time you resubmit the
tests.
  So be patient and wait for the results... dont resubmit it again...:)

-gaurav


On Fri, 1 Feb 2002, Ashmi wrote:

> when i try to resubmit it says my earlier test is still pending. if i
> still choose to resubmit the code it just overwrites my earlier files... i
> did this twice already in the last hour and nothing's come yet
>
> Help, TAs!!!
> ashmi
>
>
> On Fri, 1 Feb 2002, Mikhail Seregine wrote:
>
> > You probably need to resubmit; I resubmitted at 10:40 and got
> > a result at 10:50
> >
> > Good luck,
> >
> > Mikhail
> >
> > Mark Joseph Dolan wrote:
> > >
> > > Yes, I've been waiting for 4 hours. Do we need to resubmit???
> >
> > --
> > ----------------
> > Mikhail Seregine
> > 
> > http://www.stanford.edu/~seregine
> >
>
> ###############################################################################
>
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
>
> ###############################################################################
>
>
>
>

*************************************************************************
Gaurav Garg                          * Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             *       Escondido Village, Stanford
Department of Electrical Engineering *       CA-94305
Stanford University                  * Ph:650-498-1208
*************************************************************************

-Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine29.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: test script problem again ?
Date: Fri, 1 Feb 2002 11:52:18 -0800
Lines: 51
Distribution: su
Message-ID: 
References:  
  
NNTP-Posting-Host: elaine29.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3010

Same problem here. I had  submitted it about 8 hours ago and no results. I
would really appreciate if TAs can put their server and client in some
shared location for us to test. Can any of the students volunteer - of
course compiled without -g option so that I may not cheat.

Dhawal Kumar

On Fri, 1 Feb 2002, Ashmi wrote:

> when i try to resubmit it says my earlier test is still pending. if i
> still choose to resubmit the code it just overwrites my earlier files... i
> did this twice already in the last hour and nothing's come yet
>
> Help, TAs!!!
> ashmi
>
>
> On Fri, 1 Feb 2002, Mikhail Seregine wrote:
>
> > You probably need to resubmit; I resubmitted at 10:40 and got
> > a result at 10:50
> >
> > Good luck,
> >
> > Mikhail
> >
> > Mark Joseph Dolan wrote:
> > >
> > > Yes, I've been waiting for 4 hours. Do we need to resubmit???
> >
> > --
> > ----------------
> > Mikhail Seregine
> > 
> > http://www.stanford.edu/~seregine
> >
>
> ###############################################################################
>
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
>
> ###############################################################################
>
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: Re: Problem with the read function
Date: Fri, 1 Feb 2002 12:03:35 -0800
Lines: 59
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine43.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Moh'd Saleem Saleem Alrawashdeh 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3011

Hi,

See if this helps.

I discovered that while sending a file like "transport.o" (probably
because of being a binary file) or something, if I do

n = read(local_sd, buf, bufLen)
buf[n] = 0;
printf("bytes read = %d", n);
printf("buf length = %d", strlen(buf));

These 2 were different!

So, instead of doing

network_send(.., buf, strlen(buf), ...)

I did
network_send(.., buf, n)

and it worked for me.
-anurag



On Fri, 1 Feb 2002, Moh'd Saleem Saleem Alrawashdeh wrote:

> Hi,
>
> I am testing my code by trasmitting a 1 megabyte of binary. However, when
> the server starts to receive packets from the application ( by using the
> read function when the select says that there should be data from the
> application), I found that sometimes the number of readBytes is 0. We know
> that this is an indication to close the socket and go to finish
> state. The strange thing is that if the server igonores this 0 bytes
> message and continue reading from the application, there will be many
> coming packets to send. In other words, I am not able to terminate on the
> 0 return of the read, since in my above case, it happens that there are
> are still many packets to send and the application doesn't close the
> socket with the trasnport.
>
> I don't know what to do with this, if I ignore the 0 bytes, I will be able
> to send the whole of the 1MByte file correctly, however, the client then
> must initiate the end of connection, and the server won't be able to
> detect a finish state according to the 0 byte indication from read
> function.
>
> Any ideas if I am on the wrong track? or should the client always be the
> party which closes the connection (Although the handout says that the
> passive party maybe the active later and close the connection).
>
> Thanks,
>
> Moh'd
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sutin Chen 
Newsgroups: su.class.cs244a
Subject: test script - erased my submission?
Date: Fri, 01 Feb 2002 12:06:04 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic22.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3012

i sent in a submission at 10 and didn't get anything back.  went to
check and the test script ran as if i hadn't ever sent anything in!! 
anyway i resubmitted, but my friend submitted about two hours ago and
hasn't gotten anything back.  can someone check the test script please?
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Test Script problem, Just in
Date: Fri, 1 Feb 2002 12:20:48 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3013

I just submitted for testing, and i got this error:

RESULTS FOR THE DIFFERENT TESTS:

    * Test 1.A: Fails to compile
  [out of 20]

[SUMMARY] NOT OK
[RESULT]  ERROR:  No testserver.sample in
/afs/ir/class/cs244a/submissions/restricted/hw2.A/xwang00/hwfu//compile.sun.


so what's up?
Henry


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: only 3 people in the test queue
Date: Fri, 01 Feb 2002 12:32:25 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3014

Hi,

As of this moment there are only three people waiting in the queue to 
have their code tested. If you are one of them, you will receive your 
results in a few minutes.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Regardin the FIN Error--need help
Date: 1 Feb 2002 20:40:03 GMT
Lines: 377
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine3.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3015

Ashmi  writes:

>hi,
>I am getting a similar error, except that I don't see any network_send
>output in my test result! But I know the network_send should be
>happening, because, for example, the first line of my output says
>'SYN_ACK' received and second line is 'DATA_ACK received' so that
>means(according to my code structure) that  I should have sent a SYN and
>sent a data packet before processing SYN_ACK and DATA_ACK respectively.


It's very difficult to tell from what's posted; as mentioned before, I'd
generate all the diagnostic output you can (e.g. log to a file somewhere)
as to exactly what packets you send, which ACKs you receive, etc.  Once
you confirm that you seem to be doing what you think you should be doing,
you should double-check certain places (e.g. FIN or FIN-ACK sending) to make
sure the packet is being sent as you intended, and cross reference with the
output of the test script to make sure it's consistent and that packets
don't seem to have been dropped somewhere.  (It's very unlikely that packet
loss caused the problem, however).



>An abbreviated version of the output I get:

>-----------TEST SCRIPT OUTPUT-------------------
>  * Test 1.H: Failure to send all packets and make sure they are ACKed
>before the FIN.
>  [out of 5]

>[SUMMARY] NOT OK
>[RESULT]  FIN sent before Acking all data

>---
>network_recv: a SYN-ACK packet with seq=1 and ack=2 and
>timestamp=3216151756
>network_recv: a DATA-ACK packet with ack=14
>network_recv: a DATA packet with seq=2, len=22 and timestamp=3216151795
>network_recv: a DATA packet with seq=24, len=536 and timestamp=3216151796
>network_recv: a DATA packet with seq=560, len=536 and timestamp=3216151796
>network_recv: a DATA packet with seq=1096, len=536 and
>timestamp=3216151796.
>.
>.
>.
>.
>.
>network_recv: a DATA packet with seq=109450, len=536 and
>timestamp=3216152781
>network_recv: a DATA packet with seq=109986, len=268 and
>timestamp=3216152782
>network_recv: a FIN-ACK packet packet with seq=0 and ack=4294967295 and
>timestamp=3216152828


>---
> (INCORRECT!)


>===============

>    * Test 1.I: Failure to send a FIN_ACK after a FIN
>  [out of 5]

>[SUMMARY] NOT OK
>[RESULT]  Incorrect FIN_ACK behaviour after FIN

>---
>network_recv: a SYN-ACK packet with seq=1 and ack=2 and
>timestamp=3216151756
>network_recv: a DATA-ACK packet with ack=14
>network_recv: a DATA packet with seq=2, len=22 and timestamp=3216151795
>network_recv: a DATA packet with seq=24, len=536 and timestamp=3216151796
>network_recv: a DATA packet with seq=560, len=536 and timestamp=3216151796
>.
>.
>.
>.
>.
>network_recv: a DATA packet with seq=109450, len=536 and
>timestamp=3216152781
>network_recv: a DATA packet with seq=109986, len=268 and
>timestamp=3216152782
>timestamp=3216152828


>---
> (INCORRECT!)
>------------***********output ends******************-------------------


>Any ideas anyone? Especially the TAs? I have run out of ideas!
>Thanks
>Ashmi





>-----***************SHANKAR'S MAIL
>**************--------------------------
>On Fri, 1 Feb 2002, Shankar Agarwal wrote:

>> Hi,
>> Hi,
>> I am seeing this error for my test.It says that failure to send FIN_ACK
>> for FIN. But i am not seeing any FIN at the end of the send and recv
>> sequence. Can you please have a loot at this sequence and tell me how it
>> can be explained.
>> Thanks
>> Shankar
>>
>>     * Test 1.I: Failure to send a FIN_ACK after a FIN
>>   [out of 5]
>>
>> [SUMMARY] NOT OK
>> [RESULT]  Incorrect FIN_ACK behaviour after FIN
>>
>> ---
>> network_send: a SYN packet with seq=198
>> network_recv: a SYN-ACK packet with seq=198 and ack=199 and
>> timestamp=3230007382
>> network_recv: a DATA-ACK packet with ack=211
>> network_recv: a DATA packet with seq=199, len=536 and timestamp=3230007841
>> network_send: a DATA-ACK packet with ack=735
>> network_recv: a DATA packet with seq=735, len=536 and timestamp=3230007841
>> network_send: a DATA-ACK packet with ack=1271
>> network_recv: a DATA packet with seq=1271, len=536 and
>> timestamp=3230007842
>> network_send: a DATA-ACK packet with ack=1807
>> network_recv: a DATA packet with seq=1807, len=536 and
>> timestamp=3230007842
>> network_send: a DATA-ACK packet with ack=2343
>> network_recv: a DATA packet with seq=2343, len=536 and
>> timestamp=3230007843
>> network_send: a DATA-ACK packet with ack=2879
>> network_recv: a DATA packet with seq=2879, len=392 and
>> timestamp=3230007843
>> network_send: a DATA-ACK packet with ack=3271
>> network_recv: a DATA packet with seq=3271, len=536 and
>> timestamp=3230007844
>> network_send: a DATA-ACK packet with ack=3807
>> network_recv: a DATA packet with seq=3807, len=536 and
>> timestamp=3230007844
>> network_send: a DATA-ACK packet with ack=4343
>> network_recv: a DATA packet with seq=4343, len=536 and
>> timestamp=3230007845
>> network_send: a DATA-ACK packet with ack=4879
>> network_recv: a DATA packet with seq=4879, len=536 and
>> timestamp=3230007846
>> network_send: a DATA-ACK packet with ack=5415
>> network_recv: a DATA packet with seq=5415, len=536 and
>> timestamp=3230007846
>> network_send: a DATA-ACK packet with ack=5951
>> network_recv: a DATA packet with seq=5951, len=392 and
>> timestamp=3230007847
>> network_send: a DATA-ACK packet with ack=6343
>> network_recv: a DATA packet with seq=6343, len=536 and
>> timestamp=3230007847
>> network_send: a DATA-ACK packet with ack=6879
>> network_recv: a DATA packet with seq=6879, len=536 and
>> timestamp=3230007848
>> network_send: a DATA-ACK packet with ack=7415
>> network_recv: a DATA packet with seq=7415, len=536 and
>> timestamp=3230007849
>> network_send: a DATA-ACK packet with ack=7951
>> network_recv: a DATA packet with seq=7951, len=536 and
>> timestamp=3230007849
>> network_send: a DATA-ACK packet with ack=8487
>> network_recv: a DATA packet with seq=8487, len=536 and
>> timestamp=3230007850
>> network_send: a DATA-ACK packet with ack=9023
>> network_recv: a DATA packet with seq=9023, len=392 and
>> timestamp=3230007850
>> network_send: a DATA-ACK packet with ack=9415
>> network_recv: a DATA packet with seq=9415, len=536 and
>> timestamp=3230007851
>> network_send: a DATA-ACK packet with ack=9951
>> network_recv: a DATA packet with seq=9951, len=536 and
>> timestamp=3230007851
>> network_send: a DATA-ACK packet with ack=10487
>> network_recv: a DATA packet with seq=10487, len=536 and
>> timestamp=3230007852
>> network_send: a DATA-ACK packet with ack=11023
>> network_recv: a DATA packet with seq=11023, len=536 and
>> timestamp=3230007852
>> network_send: a DATA-ACK packet with ack=11559
>> network_recv: a DATA packet with seq=11559, len=536 and
>> timestamp=3230007853
>> network_send: a DATA-ACK packet with ack=12095
>> network_recv: a DATA packet with seq=12095, len=392 and
>> timestamp=3230007854
>> network_send: a DATA-ACK packet with ack=12487
>> network_recv: a DATA packet with seq=12487, len=536 and
>> timestamp=3230007854
>> network_send: a DATA-ACK packet with ack=13023
>> network_recv: a DATA packet with seq=13023, len=536 and
>> timestamp=3230007855
>> network_send: a DATA-ACK packet with ack=13559
>> network_recv: a DATA packet with seq=13559, len=536 and
>> timestamp=3230007855
>> network_send: a DATA-ACK packet with ack=14095
>> network_recv: a DATA packet with seq=14095, len=536 and
>> timestamp=3230007856
>> network_send: a DATA-ACK packet with ack=14631
>> network_recv: a DATA packet with seq=14631, len=536 and
>> timestamp=3230007856
>> network_send: a DATA-ACK packet with ack=15167
>> network_recv: a DATA packet with seq=15167, len=392 and
>> timestamp=3230007857
>> network_send: a DATA-ACK packet with ack=15559
>> network_recv: a DATA packet with seq=15559, len=536 and
>> timestamp=3230007857
>> network_send: a DATA-ACK packet with ack=16095
>> network_recv: a DATA packet with seq=16095, len=536 and
>> timestamp=3230007858
>> network_send: a DATA-ACK packet with ack=16631
>> network_recv: a DATA packet with seq=16631, len=536 and
>> timestamp=3230007859
>> network_send: a DATA-ACK packet with ack=17167
>> network_recv: a DATA packet with seq=17167, len=536 and
>> timestamp=3230007859
>> network_send: a DATA-ACK packet with ack=17703
>> network_recv: a DATA packet with seq=17703, len=536 and
>> timestamp=3230007860
>> network_send: a DATA-ACK packet with ack=18239
>> network_recv: a DATA packet with seq=18239, len=392 and
>> timestamp=3230007860
>> network_send: a DATA-ACK packet with ack=18631
>> network_recv: a DATA packet with seq=18631, len=536 and
>> timestamp=3230007861
>> network_send: a DATA-ACK packet with ack=19167
>> network_recv: a DATA packet with seq=19167, len=536 and
>> timestamp=3230007861
>> network_send: a DATA-ACK packet with ack=19703
>> network_recv: a DATA packet with seq=19703, len=536 and
>> timestamp=3230007862
>> network_send: a DATA-ACK packet with ack=20239
>> network_recv: a DATA packet with seq=20239, len=536 and
>> timestamp=3230007862
>> network_send: a DATA-ACK packet with ack=20775
>> network_recv: a DATA packet with seq=20775, len=536 and
>> timestamp=3230007863
>> network_send: a DATA-ACK packet with ack=21311
>> network_recv: a DATA packet with seq=21311, len=392 and
>> timestamp=3230007864
>> network_send: a DATA-ACK packet with ack=21703
>> network_recv: a DATA packet with seq=21703, len=536 and
>> timestamp=3230007864
>> network_send: a DATA-ACK packet with ack=22239
>> network_recv: a DATA packet with seq=22239, len=536 and
>> timestamp=3230007865
>> network_send: a DATA-ACK packet with ack=22775
>> network_recv: a DATA packet with seq=22775, len=536 and
>> timestamp=3230007865
>> network_send: a DATA-ACK packet with ack=23311
>> network_recv: a DATA packet with seq=23311, len=536 and
>> timestamp=3230007866
>> network_send: a DATA-ACK packet with ack=23847
>> network_recv: a DATA packet with seq=23847, len=536 and
>> timestamp=3230007867
>> network_send: a DATA-ACK packet with ack=24383
>> network_recv: a DATA packet with seq=24383, len=392 and
>> timestamp=3230007867
>> network_send: a DATA-ACK packet with ack=24775
>> network_recv: a DATA packet with seq=24775, len=536 and
>> timestamp=3230007868
>> network_send: a DATA-ACK packet with ack=25311
>> network_recv: a DATA packet with seq=25311, len=536 and
>> timestamp=3230007868
>> network_send: a DATA-ACK packet with ack=25847
>> network_recv: a DATA packet with seq=25847, len=536 and
>> timestamp=3230007869
>> network_send: a DATA-ACK packet with ack=26383
>> network_recv: a DATA packet with seq=26383, len=536 and
>> timestamp=3230007870
>> network_send: a DATA-ACK packet with ack=26919
>> network_recv: a DATA packet with seq=26919, len=536 and
>> timestamp=3230007870
>> network_send: a DATA-ACK packet with ack=27455
>> network_recv: a DATA packet with seq=27455, len=392 and
>> timestamp=3230007871
>> network_send: a DATA-ACK packet with ack=27847
>> network_recv: a DATA packet with seq=27847, len=536 and
>> timestamp=3230007871
>> network_recv: a DATA packet with seq=28383, len=536 and
>> timestamp=3230007872
>> network_send: a DATA-ACK packet with ack=28919
>> network_recv: a DATA packet with seq=28919, len=536 and
>> timestamp=3230007872
>> network_send: a DATA-ACK packet with ack=29455
>> network_recv: a DATA packet with seq=29455, len=536 and
>> timestamp=3230007873
>> network_send: a DATA-ACK packet with ack=29991
>> network_recv: a DATA packet with seq=29991, len=536 and
>> timestamp=3230007873
>> network_send: a DATA-ACK packet with ack=30527
>> network_recv: a DATA packet with seq=30527, len=392 and
>> timestamp=3230007874
>> network_send: a DATA-ACK packet with ack=30919
>> network_recv: a DATA packet with seq=30919, len=536 and
>> timestamp=3230007874
>> network_send: a DATA-ACK packet with ack=31455
>> network_recv: a DATA packet with seq=31455, len=536 and
>> timestamp=3230007875
>> network_send: a DATA-ACK packet with ack=31991
>> network_recv: a DATA packet with seq=31991, len=536 and
>> timestamp=3230007876
>> network_send: a DATA-ACK packet with ack=32527
>> network_recv: a DATA packet with seq=32527, len=536 and
>> timestamp=3230008276
>> network_send: a DATA-ACK packet with ack=33063
>> network_recv: a DATA packet with seq=33063, len=536 and
>> timestamp=3230008277
>> network_send: a DATA-ACK packet with ack=33599
>> network_recv: a DATA packet with seq=33599, len=392 and
>> timestamp=3230008278
>> network_send: a DATA-ACK packet with ack=33991
>> network_recv: a DATA packet with seq=33991, len=536 and
>> timestamp=3230008279
>> network_send: a DATA-ACK packet with ack=34527
>> network_recv: a DATA packet with seq=34527, len=536 and
>> timestamp=3230008279
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=31991, len=536 and
>> timestamp=3230008279
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=32527, len=536 and
>> timestamp=3230008280
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=33063, len=536 and
>> timestamp=3230008280
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=33599, len=392 and
>> timestamp=3230008280
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=33991, len=536 and
>> timestamp=3230008281
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=34527, len=536 and
>> timestamp=3230008281
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=31991, len=536 and
>> timestamp=3230008281
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=32527, len=536 and
>> timestamp=3230008282
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=33063, len=536 and
>> timestamp=3230008282
>> network_send: a DATA-ACK packet with ack=35063
>> network_recv: a DATA packet with seq=33599, len=392 and
>> timestamp=3230008282
>> network_send: a DATA-ACK packet with ack=35063
>>
>>
>> ---
>>  (INCORRECT!)
>>
>>
>> ===============
>>
>>
>>

>###############################################################################

>				ASHMI CHOKSHI
>Graduate Student				    141L Escondido Village
>Dept. of Computer Science			    Stanford University
>Stanford University				    Stanford Ca 94305
>Stanford Ca 94305				    (650)498-1103

>###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Test Script problem, Just in
Date: Fri, 01 Feb 2002 12:40:49 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3016

Hi,

For all the students that got this result in the test. We will not count 
it towards the total number of tests for this milestone, since it was a 
problem in the testing server on our side.

Pablo

Henry Fu wrote:

> I just submitted for testing, and i got this error:
> 
> RESULTS FOR THE DIFFERENT TESTS:
> 
>     * Test 1.A: Fails to compile
>   [out of 20]
> 
> [SUMMARY] NOT OK
> [RESULT]  ERROR:  No testserver.sample in
> /afs/ir/class/cs244a/submissions/restricted/hw2.A/xwang00/hwfu//compile.sun.
> 
> 
> so what's up?
> Henry
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga22.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: How to figure out if peer doesn't exist
Date: Fri, 1 Feb 2002 13:07:19 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3017

The only indication we get is error in network_recv when we are waiting
for a reply. But that can be a temporary failure. I looked at the errno
variable and couldn't find something very relevant. In fact when I tested
errno I found that it matched none of the values in documentation of
recvfrom which is called by network_recv. Should I take that to be an
indication?

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic20.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Are the test script working?
Date: Fri, 1 Feb 2002 13:12:12 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic20.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3018

Please, is the script is working. I submitted twice, but nothing is
showing!!

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic5.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: Are the test script working?
Date: 1 Feb 2002 21:19:58 GMT
Organization: Stanford University, CA 94305, USA
Lines: 2
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic5.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3019


Yes, I am not getting any replies either.
.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Test script currently down - stay tuned.
Date: Fri, 1 Feb 2002 13:26:00 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3020

Hi everyone,

the test script is down again, we are working on getting it 
back up. One way you could help us to prevent these crashes
is trying to make sure your executable doesn't generate
excessive amounts of debug output. 100 lines are ok,
10000 lines are ok, however 50 MBytes (yes we did have that)
is more than the server can handle.

We'll pot again once the script is back up.

  Guido


---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Regarding timeout & retransmission
Date: Fri, 1 Feb 2002 13:27:36 -0800
Lines: 88
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0229_01C1AB24.366733B0"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3021

This is a multi-part message in MIME format.

------=_NextPart_000_0229_01C1AB24.366733B0
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


I want to confirm my understanding of the STCP specification...

Suppose there are 3 packets (A,B,C), which are first sent at 0ms, 300ms, =
400ms
and there's no more to send and somehow no ACK is coming back..

And suppose RTO =3D=3D500ms..
Then packet A will timeout at 500ms.. and then, because of Go-back-N,=20
all of A,B and C are retransmitted at this moment
And RTO will be updated...
If the updated RTO is 1000ms, then they will all timeout at 1500ms..and =
retransmit.. so on..

On the other hand, if we received ACK for A at 400ms, A is discarded =
from the queue...
B will timeout at 800ms... Again, if ACK for B is received before 800ms,
C will timeout at 900ms, so on...

Thank you in advance for any comments or corrections..

Hopil


------=_NextPart_000_0229_01C1AB24.366733B0
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>I want to confirm my understanding of =
the STCP=20
specification...</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Suppose there are 3 packets =
(A,B,C), which are=20
first sent at 0ms, 300ms, 400ms</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>and there's no more to send and somehow =
no ACK is=20
coming back..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>And suppose RTO =
=3D=3D500ms..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Then packet A will timeout at 500ms.. =
and then,=20
because of Go-back-N, </FONT></DIV>
<DIV><FONT face=3DArial size=3D2>all of A,B and C are retransmitted =
</FONT><FONT=20
face=3DArial size=3D2>at this moment</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>And RTO will be updated...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>If the updated RTO is 1000ms, then they =
will all=20
timeout at 1500ms..and retransmit.. so on..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>On the other hand, if we received ACK =
for A at=20
400ms, </FONT><FONT face=3DArial size=3D2>A is discarded from the=20
queue...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>B will timeout at 800ms... Again, if =
ACK for B is=20
received before 800ms,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>C will timeout at 900ms, so =
on...</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Thank you in advance for any comments =
or=20
corrections..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_0229_01C1AB24.366733B0--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Testing script/Submit script...
Date: Fri, 1 Feb 2002 13:39:18 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012599558 4582 128.12.133.48 (1 Feb 2002 21:39:18 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3022

Does the Makefile have -DDEBUG defined? Or is #ifdef DEBUG around all of our
printouts sufficent for stopping it?

Jonathan

"Guido Appenzeller"  wrote in message

> Everyone,
>
> it looks like the testing script ran out of disk space a few minutes ago
> again and the submission script is having probles as well.
>
> We are working on it and will post again once everything is fixed.
>
>   Guido


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Test script currently down - stay tuned.
Date: Fri, 1 Feb 2002 13:40:42 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012599642 4586 128.12.133.48 (1 Feb 2002 21:40:42 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3023

Does the Makefile have -DDEBUG defined? Or is #ifdef DEBUG around all of our
printouts sufficent for stopping it?

Jonathan

"Guido Appenzeller"  wrote in message

> Hi everyone,
>
> the test script is down again, we are working on getting it
> back up. One way you could help us to prevent these crashes
> is trying to make sure your executable doesn't generate
> excessive amounts of debug output. 100 lines are ok,
> 10000 lines are ok, however 50 MBytes (yes we did have that)
> is more than the server can handle.
>
> We'll pot again once the script is back up.
>
>   Guido
>
>
> ---------------------------------------------------------------
> Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
>  - office: 650 7253545  cell: 650 7042781
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: test server is up
Date: Fri, 01 Feb 2002 13:46:38 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3024

Hi,

The test server is up and running. Sometimes it seems to have stopped, 
it is because a fellow student's program just hangs, and we have to wait 
until we decide to kill it. This means that rather than taking 1 minute 
per student, it will take 10 minutes!! So, before you test your code, 
please make sure it runs to completion at least.

In future editions of the test server, we will test all students in 
parallel, so that there is no head-of-line blocking. However this will 
not happend before today's deadline.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Programming Assignment #1, Regrade Policy
Date: Fri, 1 Feb 2002 13:53:42 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
cc: 
Xref: nntp.stanford.edu su.class.cs244a:3025

Two announcements:

** All grades for the first programming assignment have been
   mailed out. If you haven't received your grade yet please
   contact your TA.

** If you think we missed something when grading your written
   hom work or programming assignment and would like to have
   it re-graded, contact the TA responsible for grading it
   within 7 days after the home work was given back. No re-
   grades will be accepted after 7 days! (For the first hw,
   please let us know by Tuesady next week).


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Test script currently down - stay tuned.
Date: 1 Feb 2002 21:55:21 GMT
Lines: 11
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine3.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3026

"Jonathan Keljo"  writes:

>Does the Makefile have -DDEBUG defined? Or is #ifdef DEBUG around all of our
>printouts sufficent for stopping it?

>Jonathan


Nope, DEBUG is not defined in the Makefile, so you can use this to
conditionally compile debugging output.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine12.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Check about the flags
Date: Fri, 1 Feb 2002 14:27:05 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3027

Hi,

I am striving to know why I have a FIN problem in grading, although it
works perfectly when I test my RX/TX

I want to check something, when I am expecting a FIN header, ALL THE FLAGS
SHOULD BE ZERO EXCEPT THE FIN BIT ( and it is the same for all the other
cases). What I mean is that I should expect only the relevant flag to be
set, while all the others are zero by default?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine12.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Error meaning
Date: Fri, 1 Feb 2002 14:31:05 -0800
Lines: 50
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3028

Sorry again to flood the list, but I read all the previous postings and
till now I cannot exactely understand the test script output. For example,
below is the test that says that my tx and rx aren't working with each
other. What is the meaning of 0 0 0 TestData2? What are these zeros?

Thanks,

Moh'd

===============

    * Test 1.J: If the Tx and Rx fails with our Rx and our Tx but works
with each other.
  [out of 14]


[SUMMARY] NOT OK
[RESULT]
===========================
===(T1.Ja) Byte count differences found (INCORRECT!)
Student's Tx against student's Rx (reliable mode)

---

---
Difference between expected result and your result:
3c3
<       0       0       0 TESTDATA.2
---
>     131     501    5098 TESTDATA.2

---

===========================
===(T1.Jb) Checksum differences found (INCORRECT!)
Student's Tx against student's Rx (reliable mode)

---

---
Difference between expected result and your result:
3c3
< 4294967295 0 TESTDATA.2
---
> 2181702833 5098 TESTDATA.2

---



.

Path: shelby.stanford.edu!nntp.stanford.edu!saga22.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: More test script problems
Date: Fri, 1 Feb 2002 14:34:39 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3029

I had submitted my code once and passed the test. Then I changed some
print statements in transport.c and all of a sudden the test script says
it can't compile. And the program is running fine and compiles on epic5
where the test script reported an error in compile. I made no changes to
any other file.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine12.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Last Question --- Question about specification
Date: Fri, 1 Feb 2002 14:38:33 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3030

Hi,

I noticed in the specification the following:

"i.e. all data already written to the connection must be sent before the
FIN is sent."

However, to send the FIN, the read function must return 0. That means
there will be no data to send. This is why I ignored this sentence and
thought that it may be for part2. Please correct me if I am wrong, maybe
this will solve the FIN script error that I got.

Thanks,

Moh'd

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Last Question --- Question about specification
Date: Fri, 1 Feb 2002 14:52:32 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3031


Hi...
I ignored that part like you, and tested OK for FIN/ACK...
So ignoring that sentence was safe... I mean, in my case..

Hopil

"Moh'd Saleem Saleem Alrawashdeh"  wrote in message

> Hi,
>
> I noticed in the specification the following:
>
> "i.e. all data already written to the connection must be sent before the
> FIN is sent."
>
> However, to send the FIN, the read function must return 0. That means
> there will be no data to send. This is why I ignored this sentence and
> thought that it may be for part2. Please correct me if I am wrong, maybe
> this will solve the FIN script error that I got.
>
> Thanks,
>
> Moh'd
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga22.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Last Question --- Question about specification
Date: Fri, 1 Feb 2002 15:07:03 -0800
Lines: 41
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3032

I too think so and am passing the tests. In the first part of this
assignment there can be no data outstanding since we send the data the
moment application gives to us (of course if there is space in our sender
window) and we never retransmit - that means we are not storing the
packets - at least I am not. If the sender win if full I don't read from
appl which means by the time I get a read of 0 from appl, there wouldn't
be any data in that pipe and no data in STCPs buffer to transmit.

Dhawal Kumar

 On Fri, 1 Feb 2002, BAEHOPIL wrote:

>
> Hi...
> I ignored that part like you, and tested OK for FIN/ACK...
> So ignoring that sentence was safe... I mean, in my case..
>
> Hopil
>
> "Moh'd Saleem Saleem Alrawashdeh"  wrote in message
> 
> > Hi,
> >
> > I noticed in the specification the following:
> >
> > "i.e. all data already written to the connection must be sent before the
> > FIN is sent."
> >
> > However, to send the FIN, the read function must return 0. That means
> > there will be no data to send. This is why I ignored this sentence and
> > thought that it may be for part2. Please correct me if I am wrong, maybe
> > this will solve the FIN script error that I got.
> >
> > Thanks,
> >
> > Moh'd
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga22.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Now purify problems
Date: Fri, 1 Feb 2002 15:08:38 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3033

Although client.purify works for me, when I run server.purify I get

ld.so.1: ./server.purify: fatal:
libsocket.so.1_pure_p3_c0_109072111_58_32_70800: open failed: No such file
or directory
Killed


Any ideas?

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga22.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: More test script problems
Date: Fri, 1 Feb 2002 15:10:15 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3034

If you haven't responded to the last one please don't. I tried submitting
again and this time it compiled without errors.

Dhawal Kumar

On Fri, 1 Feb 2002, Dhawal Kumar wrote:

> I had submitted my code once and passed the test. Then I changed some
> print statements in transport.c and all of a sudden the test script says
> it can't compile. And the program is running fine and compiles on epic5
> where the test script reported an error in compile. I made no changes to
> any other file.
>
> Dhawal Kumar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic5.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: Now purify problems
Date: 1 Feb 2002 23:37:10 GMT
Organization: Stanford University, CA 94305, USA
Lines: 2
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic5.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3035

Did you make all.purify on the same machine you're trying to run the
server on?
.

Path: shelby.stanford.edu!nntp.stanford.edu!myth9.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: hi, are the test scripts working ?
Date: Fri, 1 Feb 2002 15:48:22 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3036


I submitted one at 2.45pm and still have not got results.

Thanks, Arvind

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine32.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Now purify problems
Date: Fri, 1 Feb 2002 15:58:18 -0800
Lines: 10
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine32.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3037

Thanks a lot. Your suggestion worked.

Dhawal Kumar

On 1 Feb 2002, Mark Joseph Dolan wrote:

> Did you make all.purify on the same machine you're trying to run the
> server on?
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: URGENT: what is wrong with the test script!
Date: Fri, 1 Feb 2002 16:02:49 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3038

i submitted the test at 2:30 and nothing back yet!
what's going on? deadline is 1 hour away!

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tudor Andrei Bosman 
Newsgroups: su.class.cs244a
Subject: update deadline in submit_test.pl
Date: 2 Feb 2002 01:09:41 GMT
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine5.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3039

Although the deadline for hw2.A has been extended by 24 hours,
submit_test.pl thinks the deadline is still 2/1, and doesn't let me test
my code again.

Thanks,
Tudor.

elaine5:~/cs244a/2> /usr/class/cs244a/bin/submit_test.pl hw2.A xwang00
SubmitBase: /afs/ir/class/cs244a/submissions
Debug Level = 1
######################################################################
Past Submission Time
DEADLINE: 2/1/2002, 17:00
######################################################################
======================================================================
You cannot test your code because you have exceeded the maximum
amount of tests that you are allowed for this time and date.

If you believe this is an error or you are allowed to test the code by
your professor, please send an email to your TA
(xwang00; 
======================================================================

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: test script
Date: Fri, 01 Feb 2002 17:24:03 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3040

Hi,

The test script died at 2:41pm and was restarted (with the new deadline 
set by Prof. McKeown) at 5:14pm. I will be monitoring until the dealine 
to make sure it does not die again.

BTW, The deadline is also effective for the submit.pl script.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga22.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: update deadline in submit_test.pl
Date: Fri, 1 Feb 2002 17:51:52 -0800
Lines: 39
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3041

me too. infact i only got 2 replies...whereas the maximum was 3.


On 2 Feb 2002, Tudor Andrei Bosman wrote:

> Although the deadline for hw2.A has been extended by 24 hours,
> submit_test.pl thinks the deadline is still 2/1, and doesn't let me test
> my code again.
>
> Thanks,
> Tudor.
>
> elaine5:~/cs244a/2> /usr/class/cs244a/bin/submit_test.pl hw2.A xwang00
> SubmitBase: /afs/ir/class/cs244a/submissions
> Debug Level = 1
> ######################################################################
> Past Submission Time
> DEADLINE: 2/1/2002, 17:00
> ######################################################################
> ======================================================================
> You cannot test your code because you have exceeded the maximum
> amount of tests that you are allowed for this time and date.
>
> If you believe this is an error or you are allowed to test the code by
> your professor, please send an email to your TA
> (xwang00; 
> ======================================================================
>
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Leonard Sibille 
Newsgroups: su.class.cs244a
Subject: Re: update deadline in submit_test.pl
Date: Fri, 01 Feb 2002 17:58:48 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
Reply-To: 
NNTP-Posting-Host: saclay.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1012615424 7449 171.64.68.193 (2 Feb 2002 02:03:44 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.18 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3043

You probably need to resubmit your code. This way things worked fine for
me.

Leo



Tudor Andrei Bosman wrote:


> 
> Although the deadline for hw2.A has been extended by 24 hours,
> submit_test.pl thinks the deadline is still 2/1, and doesn't let me test
> my code again.
> 
> Thanks,
> Tudor.
> 
> elaine5:~/cs244a/2> /usr/class/cs244a/bin/submit_test.pl hw2.A xwang00
> SubmitBase: /afs/ir/class/cs244a/submissions
> Debug Level = 1
> ######################################################################
> Past Submission Time
> DEADLINE: 2/1/2002, 17:00
> ######################################################################
> ======================================================================
> You cannot test your code because you have exceeded the maximum
> amount of tests that you are allowed for this time and date.
> 
> If you believe this is an error or you are allowed to test the code by
> your professor, please send an email to your TA
> (xwang00; 
> ======================================================================
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Retransmissions
Date: Fri, 1 Feb 2002 20:19:38 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3044

Hi,

  In the Go-Back-N protocol if a segment N is retransmitted,
  all data greater than N is also retransmitted.  Does that
  mean that the timers for all segments above N is reset and
  their timeout counter is increased by one? Thoughts? Thanks :-)

                        ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Sundar Maddali 
Newsgroups: su.class.cs244a
Subject: Purify UMR Errors on the  access of stcpheader.data_len
Date: Fri, 01 Feb 2002 23:31:08 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3045

Hi,
   when i run my purified server/client, i get the following errors..

the  STCP_Header is defined as follows

#pragma pack(1)
typedef struct STCP_Header
{
  struct tcphdr tcp_hdr;
  u_short data_len;
} __attribute__ ((packed)) STCP_Header;

i see that 22 bytes are allocated for the header and
data_len starts at 21st byte.

but whenever i access header.data_len, the program reads 4 bytes instead
2, 
resulting in 2bytes of UMR.

any help will be appreciated.

~
Lakshman shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga14.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: review session feb 1
Date: Fri, 1 Feb 2002 23:54:43 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3046

Its for the TAs.
Can't the slides be put up in the web page? I think the dumps were pretty
useful to understand congestion window mechanism.

thanks
Abhishek Das

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga14.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Program Assignment Grades
Date: Sat, 2 Feb 2002 00:08:28 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3047


Hi
Again for TAs.
I just learned that the grades were were sent out for Assignment 1 via
email. I didn't get any such mail.
Could you please check this.

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: update deadline in submit_test.pl
Date: Sat, 02 Feb 2002 01:39:15 -0800
Lines: 34
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3048

Please reset the script, I thought waiting
until midnight would work - but it still
prevents me from testing,

Thanks....

Tudor Andrei Bosman wrote:
> 
> Although the deadline for hw2.A has been extended by 24 hours,
> submit_test.pl thinks the deadline is still 2/1, and doesn't let me test
> my code again.
> 
> Thanks,
> Tudor.
> 
> elaine5:~/cs244a/2> /usr/class/cs244a/bin/submit_test.pl hw2.A xwang00
> SubmitBase: /afs/ir/class/cs244a/submissions
> Debug Level = 1
> ######################################################################
> Past Submission Time
> DEADLINE: 2/1/2002, 17:00
> ######################################################################
> ======================================================================
> You cannot test your code because you have exceeded the maximum
> amount of tests that you are allowed for this time and date.
> 
> If you believe this is an error or you are allowed to test the code by
> your professor, please send an email to your TA
> (xwang00; 
> ======================================================================

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: To the Professor and TAs: test script doesnt allow submissions
Date: Sat, 2 Feb 2002 07:22:33 -0800
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3049

hi,
the test script is still not allowing me to submit today (saturday)
morning.
yesterday a couple of submitted tests didnt result in any replies, so
maybe those counted to the maximum number allowed? if that is the case
then i think it is only fair that we be allowed to test again. after all
it was a problem with the script!

please do something FAST!
thanks
ashmi

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Victor Wong" 
Newsgroups: su.class.cs244a
Subject: Termination
Date: Sat, 2 Feb 2002 09:14:56 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: summerbridge.stanford.edu
X-Trace: news.Stanford.EDU 1012670103 13892 128.12.82.21 (2 Feb 2002 17:15:03 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3050

According to the assignment specs, if the passive end  fails to send the
FIN-ACK, the active end should terminate its end of the connection anyway.
Can I conclude that it's sufficient for the active end to just send out the
FIN and disconnect? Otherwise how long should I wait before disconnecting
the parent?

Thanks,
Victor.

PS. Despite the problems with the testing script, it's been really useful to
have it around. Otherwise I don't know what sort of grade I'd get since "my
server and client work together before...." :)


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: IS ANYONE MONITORING THE NEWSGROUP?
Date: Sat, 2 Feb 2002 11:53:31 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3051



###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: similar problem when submitting test--exceed max number of submission.
Date: Sat, 2 Feb 2002 13:10:48 -0800
Lines: 50
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine18.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3052

I got a similar problem when submitting my test. When I submitted the
test, it did allow me to submit, but after a little while, I got a new
email said "you can not test our code because you have exceeded the max
amount of tests.". But the problem is this is my first time to submit
the test for today.

If someone is monitoring the test server, please take a look at this
problem.

Thank you very much.

Wanda

 On Fri, 1 Feb 2002, Leonard Sibille wrote:

> You probably need to resubmit your code. This way things worked fine for
> me.
>
> Leo
>
>
>
> Tudor Andrei Bosman wrote:
>
>
> >
> > Although the deadline for hw2.A has been extended by 24 hours,
> > submit_test.pl thinks the deadline is still 2/1, and doesn't let me test
> > my code again.
> >
> > Thanks,
> > Tudor.
> >
> > elaine5:~/cs244a/2> /usr/class/cs244a/bin/submit_test.pl hw2.A xwang00
> > SubmitBase: /afs/ir/class/cs244a/submissions
> > Debug Level = 1
> > ######################################################################
> > Past Submission Time
> > DEADLINE: 2/1/2002, 17:00
> > ######################################################################
> > ======================================================================
> > You cannot test your code because you have exceeded the maximum
> > amount of tests that you are allowed for this time and date.
> >
> > If you believe this is an error or you are allowed to test the code by
> > your professor, please send an email to your TA
> > (xwang00; 
> > ======================================================================
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Quick note to peeps developing on linux
Date: Sat, 2 Feb 2002 13:29:13 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3053


The semantics of select(..) are different in linux
than most other OSes, solaris included.  The struct
timeval argument is modified in linux, not in solaris,
so you should assume its value after the call
is indeterminant for cross compatibility.

            ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Can not submit for tesing.
Date: Sat, 2 Feb 2002 13:48:46 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3054

Hi,
Can someone please look into the matter. I am not able to submit my test
for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
time is ticking fast.
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 2 Feb 2002 14:00:56 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3055

same here... and no one seems to be doing anything about it...
the handout says "3 tests 24 hours before deadline", but i couldnt submit
even 1 today since morning!

ashmi

On Sat, 2 Feb 2002, Shankar Agarwal wrote:

> Hi,
> Can someone please look into the matter. I am not able to submit my test
> for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
> time is ticking fast.
> Shankar
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Sundar Maddali 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 02 Feb 2002 14:06:14 -0800
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3056

I am able to submit for testing. but the point to note is thati
submitted about 3-4 times
in this assignment cycle and was able to submit twice since 11 AM today.

~
Lakshman 

Shankar Agarwal wrote:
> 
> Hi,
> Can someone please look into the matter. I am not able to submit my test
> for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
> time is ticking fast.
> Shankar
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!spabrai
From: Shashank Pabrai 
Newsgroups: su.class.cs244a
Subject: Test results: what do these mean?
Date: Sat, 2 Feb 2002 14:18:02 -0800
Lines: 93
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3057


Hi Folks,

In the test results, I am supposed to see expected data vs data from my
code. I have copy-pasted below the test results from today ... can someone
tell me what the results mean?

(I pass all tests except 1.B & 1.C)

Thanks,
Shashank.

===============

    * Test 1.B: Transmitter fails to work in reliable mode with our
Receiver
	[out of 10]

[SUMMARY] NOT OK
[RESULT]
===========================
===(T1.Ba) Byte count differences found (INCORRECT!)
Student's Tx against our Rx (reliable mode)

---

---
Difference between expected result and your result:
0a1,3
>    2666   15720  110230 TESTDATA.0
>    1812   10049  299328 TESTDATA.1
>     131     501    5098 TESTDATA.2

---

===========================
===(T1.Bb) Checksum differences found (INCORRECT!)
Student's Tx against our Rx (reliable mode)

---

---
Difference between expected result and your result:
0a1,3
> 1705403052 110230 TESTDATA.0
> 4265638025 299328 TESTDATA.1
> 2181702833 5098 TESTDATA.2

---


===============

    * Test 1.C: Receiver fails to work in reliable mode with our
Transmitter
	[out of 10]

[SUMMARY] NOT OK
[RESULT]
===========================
===(T1.Ca) Byte count differences found (INCORRECT!)
Our Tx against student's Rx (reliable mode)

---

---
Difference between expected result and your result:
0a1,3
>    2666   15720  110230 TESTDATA.0
>    1812   10049  299328 TESTDATA.1
>     131     501    5098 TESTDATA.2

---

===========================
===(T1.Cb) Checksum differences found (INCORRECT!)
Our Tx against student's Rx (reliable mode)

---

---
Difference between expected result and your result:
0a1,3
> 1705403052 110230 TESTDATA.0
> 4265638025 299328 TESTDATA.1
> 2181702833 5098 TESTDATA.2

---


===============


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Sundar Maddali 
Newsgroups: su.class.cs244a
Subject: FIN- FIN_ACK behaviour
Date: Sat, 02 Feb 2002 14:19:44 -0800
Lines: 91
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3058

Hi
  i am submit for testing it shows errors in FIN, FIN_ACK handling.

the errors are

   * Test 1.H: Failure to send all packets and make sure they are ACKed
before the FIN.
  [out of 5]

[SUMMARY] NOT OK
[RESULT]  FIN sent before Acking all data

---
network_send: a SYN packet with seq=0
network_recv: a SYN-ACK packet with seq=0 and ack=1 and
timestamp=3369196918
network_recv: a DATA-ACK packet with ack=13
network_recv: a DATA packet with seq=1, len=536 and timestamp=3369196954
network_recv: a DATA packet with seq=537, len=536 and
timestamp=3369196974
network_recv: a DATA packet with seq=1073, len=536 and
timestamp=3369196994
network_recv: a DATA packet with seq=1609, len=536 and
timestamp=3369197014
network_recv: a DATA packet with seq=2145, len=536 and
timestamp=3369197034
network_recv: a DATA packet with seq=2681, len=392 and
timestamp=3369197054
network_recv: a DATA packet with seq=3073, len=536 and
timestamp=3369197074
network_recv: a DATA packet with seq=3609, len=536 and
timestamp=3369197094
network_recv: a DATA packet with seq=4145, len=536 and
timestamp=3369197114
................
.................
network_recv: a DATA packet with seq=109665, len=536 and
timestamp=3369201233
network_recv: a DATA packet with seq=110201, len=52 and
timestamp=3369201254
network_recv: a FIN-ACK packet packet with seq=0 and ack=0 and
timestamp=3369201313

ALSO

    * Test 1.I: Failure to send a FIN_ACK after a FIN
  [out of 5]

[SUMMARY] NOT OK
[RESULT]  Incorrect FIN_ACK behaviour after FIN

---
network_send: a SYN packet with seq=0
network_recv: a SYN-ACK packet with seq=0 and ack=1 and
timestamp=3369196918
network_recv: a DATA-ACK packet with ack=13
network_recv: a DATA packet with seq=1, len=536 and timestamp=3369196954
network_recv: a DATA packet with seq=537, len=536 and
timestamp=3369196974
network_recv: a DATA packet with seq=1073, len=536 and
timestamp=3369196994
network_recv: a DATA packet with seq=1609, len=536 and
timestamp=3369197014
network_recv: a DATA packet with seq=2145, len=536 and
timestamp=3369197034
......................
.....................

network_recv: a DATA packet with seq=109665, len=536 and
timestamp=3369201233
network_recv: a DATA packet with seq=110201, len=52 and
timestamp=3369201254
network_recv: a FIN-ACK packet packet with seq=0 and ack=0 and
timestamp=3369201313


---
 (INCORRECT!)

----------

It can also be observed that DATA_ACKS( network_send )  and FIN are not
printed. 
there are no network_send 's after the initial DATA_ACK. i think that my
program is sending
the DATA_ACKS, for otherwise it wouldnt have proceeded this far.

Can anyone help please.

Thanks
Lakshman Shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Termination
Date: 2 Feb 2002 22:23:24 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine32.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3059

Your retransmission mecahnism wil lautomatically handle this and terminate 
the connection if you do not receive a FIN_ACK.

You do not need to worry about this case for the first milestone.

Arun

Victor Wong  wrote:

: According to the assignment specs, if the passive end  fails to send the
: FIN-ACK, the active end should terminate its end of the connection anyway.
: Can I conclude that it's sufficient for the active end to just send out the
: FIN and disconnect? Otherwise how long should I wait before disconnecting
: the parent?

: Thanks,
: Victor.

: PS. Despite the problems with the testing script, it's been really useful to
: have it around. Otherwise I don't know what sort of grade I'd get since "my
: server and client work together before...." :)


.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: number of submissions today
Date: Sat, 2 Feb 2002 14:26:13 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3060

Hi,

It turns out that the script was configured to limit the number of 
submissions during the last 3 days (not the last one as stated in the 
assignment). This has just been changed.

Remember that for hw2.A you will only be allowed a total of 3 submissions 
since 5pm on Friday.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: 2 Feb 2002 22:29:44 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine32.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3061

I received notice that students have been able to successfully submit and 
receive their results as of 2:00 pm.

Please let us know if the problems continue.

Arun


Shankar Agarwal  wrote:
: Hi,
: Can someone please look into the matter. I am not able to submit my test
: for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
: time is ticking fast.
: Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Test results: what do these mean?
Date: 2 Feb 2002 22:34:00 GMT
Lines: 105
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine32.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3062

Please see earlier posts in the newsgroup where the output is explained.  
Essentially, when either your transmitter/receiver is matched with our
receiver/transmitter (respectively), the files do not checksum properly.

Possible explanations for this are also discussed earlier in earlier posts 
on this newsgroup.

Good luck,

Arun

Shashank Pabrai  wrote:

: Hi Folks,

: In the test results, I am supposed to see expected data vs data from my
: code. I have copy-pasted below the test results from today ... can someone
: tell me what the results mean?

: (I pass all tests except 1.B & 1.C)

: Thanks,
: Shashank.

: ===============

:     * Test 1.B: Transmitter fails to work in reliable mode with our
: Receiver
: 	[out of 10]

: [SUMMARY] NOT OK
: [RESULT]
: ===========================
: ===(T1.Ba) Byte count differences found (INCORRECT!)
: Student's Tx against our Rx (reliable mode)

: ---

: ---
: Difference between expected result and your result:
: 0a1,3
:>    2666   15720  110230 TESTDATA.0
:>    1812   10049  299328 TESTDATA.1
:>     131     501    5098 TESTDATA.2

: ---

: ===========================
: ===(T1.Bb) Checksum differences found (INCORRECT!)
: Student's Tx against our Rx (reliable mode)

: ---

: ---
: Difference between expected result and your result:
: 0a1,3
:> 1705403052 110230 TESTDATA.0
:> 4265638025 299328 TESTDATA.1
:> 2181702833 5098 TESTDATA.2

: ---


: ===============

:     * Test 1.C: Receiver fails to work in reliable mode with our
: Transmitter
: 	[out of 10]

: [SUMMARY] NOT OK
: [RESULT]
: ===========================
: ===(T1.Ca) Byte count differences found (INCORRECT!)
: Our Tx against student's Rx (reliable mode)

: ---

: ---
: Difference between expected result and your result:
: 0a1,3
:>    2666   15720  110230 TESTDATA.0
:>    1812   10049  299328 TESTDATA.1
:>     131     501    5098 TESTDATA.2

: ---

: ===========================
: ===(T1.Cb) Checksum differences found (INCORRECT!)
: Our Tx against student's Rx (reliable mode)

: ---

: ---
: Difference between expected result and your result:
: 0a1,3
:> 1705403052 110230 TESTDATA.0
:> 4265638025 299328 TESTDATA.1
:> 2181702833 5098 TESTDATA.2

: ---


: ===============


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: FIN- FIN_ACK behaviour
Date: 2 Feb 2002 22:39:31 GMT
Lines: 104
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine32.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3063

Before sending a FIN, you want to be sure that all outstanding data that 
has yet to have been ACKed has been. Thus you will wait for your send 
window to clear (i.e. become 3072 bytes), and then you will send a FIN.

The passive side will receive this FIN, send a FIN_ACK, and terminate 
immediately. Once the active side receives the FIN_ACK, it may terminate 
its end of the connection as well.

Good luck,

Arun

Lakshman Sundar Maddali  wrote:
: Hi
:   i am submit for testing it shows errors in FIN, FIN_ACK handling.

: the errors are

:    * Test 1.H: Failure to send all packets and make sure they are ACKed
: before the FIN.
:   [out of 5]

: [SUMMARY] NOT OK
: [RESULT]  FIN sent before Acking all data

: ---
: network_send: a SYN packet with seq=0
: network_recv: a SYN-ACK packet with seq=0 and ack=1 and
: timestamp=3369196918
: network_recv: a DATA-ACK packet with ack=13
: network_recv: a DATA packet with seq=1, len=536 and timestamp=3369196954
: network_recv: a DATA packet with seq=537, len=536 and
: timestamp=3369196974
: network_recv: a DATA packet with seq=1073, len=536 and
: timestamp=3369196994
: network_recv: a DATA packet with seq=1609, len=536 and
: timestamp=3369197014
: network_recv: a DATA packet with seq=2145, len=536 and
: timestamp=3369197034
: network_recv: a DATA packet with seq=2681, len=392 and
: timestamp=3369197054
: network_recv: a DATA packet with seq=3073, len=536 and
: timestamp=3369197074
: network_recv: a DATA packet with seq=3609, len=536 and
: timestamp=3369197094
: network_recv: a DATA packet with seq=4145, len=536 and
: timestamp=3369197114
: ...............
: ................
: network_recv: a DATA packet with seq=109665, len=536 and
: timestamp=3369201233
: network_recv: a DATA packet with seq=110201, len=52 and
: timestamp=3369201254
: network_recv: a FIN-ACK packet packet with seq=0 and ack=0 and
: timestamp=3369201313

: ALSO

:     * Test 1.I: Failure to send a FIN_ACK after a FIN
:   [out of 5]

: [SUMMARY] NOT OK
: [RESULT]  Incorrect FIN_ACK behaviour after FIN

: ---
: network_send: a SYN packet with seq=0
: network_recv: a SYN-ACK packet with seq=0 and ack=1 and
: timestamp=3369196918
: network_recv: a DATA-ACK packet with ack=13
: network_recv: a DATA packet with seq=1, len=536 and timestamp=3369196954
: network_recv: a DATA packet with seq=537, len=536 and
: timestamp=3369196974
: network_recv: a DATA packet with seq=1073, len=536 and
: timestamp=3369196994
: network_recv: a DATA packet with seq=1609, len=536 and
: timestamp=3369197014
: network_recv: a DATA packet with seq=2145, len=536 and
: timestamp=3369197034
: .....................
: ....................

: network_recv: a DATA packet with seq=109665, len=536 and
: timestamp=3369201233
: network_recv: a DATA packet with seq=110201, len=52 and
: timestamp=3369201254
: network_recv: a FIN-ACK packet packet with seq=0 and ack=0 and
: timestamp=3369201313


: ---
:  (INCORRECT!)

: ----------

: It can also be observed that DATA_ACKS( network_send )  and FIN are not
: printed. 
: there are no network_send 's after the initial DATA_ACK. i think that my
: program is sending
: the DATA_ACKS, for otherwise it wouldnt have proceeded this far.

: Can anyone help please.

: Thanks
: Lakshman Shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 2 Feb 2002 15:14:41 -0800
Lines: 39
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Arun Upadhyaya Kishan 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3064

hi,
I still can't submit for testing, I keep getting an email saying I
exceeded the number of submissions, but I haven't tested at all since
yesterday.
Can you find out what's happening? My SUID is ashmi.
Thanks
ashmi

 On 2 Feb 2002, Arun Upadhyaya Kishan wrote:

> I received notice that students have been able to successfully submit and
> receive their results as of 2:00 pm.
>
> Please let us know if the problems continue.
>
> Arun
>
>
> Shankar Agarwal  wrote:
> : Hi,
> : Can someone please look into the matter. I am not able to submit my test
> : for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
> : time is ticking fast.
> : Shankar
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine27.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 2 Feb 2002 15:18:29 -0800
Lines: 48
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: elaine27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3065

Me too. could somebody look at the problem? My suid is wjiang.

Thanks,
Wanda

On Sat, 2 Feb 2002, Ashmi wrote:

> hi,
> I still can't submit for testing, I keep getting an email saying I
> exceeded the number of submissions, but I haven't tested at all since
> yesterday.
> Can you find out what's happening? My SUID is ashmi.
> Thanks
> ashmi
>
>  On 2 Feb 2002, Arun Upadhyaya Kishan wrote:
>
> > I received notice that students have been able to successfully submit and
> > receive their results as of 2:00 pm.
> >
> > Please let us know if the problems continue.
> >
> > Arun
> >
> >
> > Shankar Agarwal  wrote:
> > : Hi,
> > : Can someone please look into the matter. I am not able to submit my test
> > : for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
> > : time is ticking fast.
> > : Shankar
> >
> >
>
> ###############################################################################
>
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
>
> ###############################################################################
>
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic8.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 2 Feb 2002 15:19:20 -0800
Lines: 49
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: epic8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Ashmi 
cc: Arun Upadhyaya Kishan 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3066


Same here...
I have already emailed my TA and no response from him either :(

-Arvind

On Sat, 2 Feb 2002, Ashmi wrote:

> hi,
> I still can't submit for testing, I keep getting an email saying I
> exceeded the number of submissions, but I haven't tested at all since
> yesterday.
> Can you find out what's happening? My SUID is ashmi.
> Thanks
> ashmi
>
>  On 2 Feb 2002, Arun Upadhyaya Kishan wrote:
>
> > I received notice that students have been able to successfully submit and
> > receive their results as of 2:00 pm.
> >
> > Please let us know if the problems continue.
> >
> > Arun
> >
> >
> > Shankar Agarwal  wrote:
> > : Hi,
> > : Can someone please look into the matter. I am not able to submit my test
> > : for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
> > : time is ticking fast.
> > : Shankar
> >
> >
>
> ###############################################################################
>
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
>
> ###############################################################################
>
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 2 Feb 2002 15:22:02 -0800
Lines: 69
Distribution: su
Message-ID: 
References: 
  
 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Arvind Bhat 
cc: Arun Upadhyaya Kishan 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3067

I haven't been able to test at all since yesterday; essentially the
extended deadline has not proved to be at all useful to me!
ashmi


On Sat, 2 Feb 2002, Arvind Bhat wrote:

>
> Same here...
> I have already emailed my TA and no response from him either :(
>
> -Arvind
>
> On Sat, 2 Feb 2002, Ashmi wrote:
>
> > hi,
> > I still can't submit for testing, I keep getting an email saying I
> > exceeded the number of submissions, but I haven't tested at all since
> > yesterday.
> > Can you find out what's happening? My SUID is ashmi.
> > Thanks
> > ashmi
> >
> >  On 2 Feb 2002, Arun Upadhyaya Kishan wrote:
> >
> > > I received notice that students have been able to successfully submit and
> > > receive their results as of 2:00 pm.
> > >
> > > Please let us know if the problems continue.
> > >
> > > Arun
> > >
> > >
> > > Shankar Agarwal  wrote:
> > > : Hi,
> > > : Can someone please look into the matter. I am not able to submit my test
> > > : for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
> > > : time is ticking fast.
> > > : Shankar
> > >
> > >
> >
> > ###############################################################################
> >
> > 				ASHMI CHOKSHI
> > Graduate Student				    141L Escondido Village
> > Dept. of Computer Science			    Stanford University
> > Stanford University				    Stanford Ca 94305
> > Stanford Ca 94305				    (650)498-1103
> >
> > ###############################################################################
> >
> >
> >
> >
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!saga0.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: karn's algorithm
Date: Sat, 2 Feb 2002 15:59:38 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga0.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3068


It is understood that we don't take the sampleRTT for retransmitted
segments. But are we supposed to double the RTO as explained in class or
keep the RTO same as before?

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 02 Feb 2002 16:13:40 -0800
Lines: 82
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, en, fr, de
To: Ashmi 
CC: Arvind Bhat 
	Arun Upadhyaya Kishan 
Xref: nntp.stanford.edu su.class.cs244a:3069

Hi,

When reporting an error, please, copy the message that you see on the 
screen. It makes debugging easier. Thank you.

Pablo

Ashmi wrote:

> I haven't been able to test at all since yesterday; essentially the
> extended deadline has not proved to be at all useful to me!
> ashmi
> 
> 
> On Sat, 2 Feb 2002, Arvind Bhat wrote:
> 
> 
>>Same here...
>>I have already emailed my TA and no response from him either :(
>>
>>-Arvind
>>
>>On Sat, 2 Feb 2002, Ashmi wrote:
>>
>>
>>>hi,
>>>I still can't submit for testing, I keep getting an email saying I
>>>exceeded the number of submissions, but I haven't tested at all since
>>>yesterday.
>>>Can you find out what's happening? My SUID is ashmi.
>>>Thanks
>>>ashmi
>>>
>>> On 2 Feb 2002, Arun Upadhyaya Kishan wrote:
>>>
>>>
>>>>I received notice that students have been able to successfully submit and
>>>>receive their results as of 2:00 pm.
>>>>
>>>>Please let us know if the problems continue.
>>>>
>>>>Arun
>>>>
>>>>
>>>>Shankar Agarwal  wrote:
>>>>: Hi,
>>>>: Can someone please look into the matter. I am not able to submit my test
>>>>: for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
>>>>: time is ticking fast.
>>>>: Shankar
>>>>
>>>>
>>>>
>>>###############################################################################
>>>
>>>				ASHMI CHOKSHI
>>>Graduate Student				    141L Escondido Village
>>>Dept. of Computer Science			    Stanford University
>>>Stanford University				    Stanford Ca 94305
>>>Stanford Ca 94305				    (650)498-1103
>>>
>>>###############################################################################
>>>
>>>
>>>
>>>
>>>
> 
> ###############################################################################
> 
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
> 
> ###############################################################################
> 
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: hw2.B test script
Date: Sat, 2 Feb 2002 16:18:45 -0800
Lines: 54
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_00A5_01C1AC05.497BF940"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3070

This is a multi-part message in MIME format.

------=_NextPart_000_00A5_01C1AC05.497BF940
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


It seems that hw2.B is not ready yet..
I test_submitted my dumb hw2.A code to hw2.B test script,
but it said OK for most of the tests, notably 2.B, 2.D, 2.E, 2.M, and =
more...
( It cannot be OK without any provision for timeout or reassembly, etc.. =
)

I wonder when we can actually use it..

Hopil


------=_NextPart_000_00A5_01C1AC05.497BF940
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>It seems that hw2.B is not ready =
yet..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I test_submitted my dumb hw2.A code to =
hw2.B test=20
script,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>but it said OK for most of the tests, =
notably 2.B,=20
2.D, 2.E, 2.M, and more...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>( It cannot be OK without any provision =
for timeout=20
or reassembly, etc.. )</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>I wonder when we can actually use=20
it..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_00A5_01C1AC05.497BF940--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lin Lu 
Newsgroups: su.class.cs244a
Subject: clarification
Date: Sat, 02 Feb 2002 16:42:45 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine5.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3071

Hi,

I am not sure about several things.  Could anyone clarify?

1. before send out FIN, STCP just need to flush all the data in send
window that has not been acknowledged. right ?  Does it have to wait for
all the acknowledgement?  If it does not receive the acknowledgement
again, reflush the data in send window again?
Also, the passive does not do anything but just close its connection and
send out FIN_ACK after it receives FIN. right?


2. On assignment discription page 4, under overview, it says it is
STCP's job to break up the data into packets and reassemble the data on
the other side.  I understand STCP will break up the data if it reads
more than 536 from application.  But if there is no packet reordering on
the receiver side, the receiver side STCP just send the data to its
application right away once it receives the packet, no matter how big
the packet is.  So what does the reassemble mean here?

Thanks a lot.

Lin
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Can not submit for tesing.
Date: Sat, 02 Feb 2002 16:46:57 -0800
Lines: 101
Distribution: su
Message-ID: 
References:      
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3072

Hi,

Some of the students that complained before have submitted their code 
for testing again, and it worked for them. If you still cannot test your 
code for hw2.A, please zwrite to me ("zwrite molinero" from a Sweet Hall 
machine)

Pablo

Pablo Molinero Fernandez wrote:

> Hi,
> 
> When reporting an error, please, copy the message that you see on the 
> screen. It makes debugging easier. Thank you.
> 
> Pablo
> 
> Ashmi wrote:
> 
>> I haven't been able to test at all since yesterday; essentially the
>> extended deadline has not proved to be at all useful to me!
>> ashmi
>>
>>
>> On Sat, 2 Feb 2002, Arvind Bhat wrote:
>>
>>
>>> Same here...
>>> I have already emailed my TA and no response from him either :(
>>>
>>> -Arvind
>>>
>>> On Sat, 2 Feb 2002, Ashmi wrote:
>>>
>>>
>>>> hi,
>>>> I still can't submit for testing, I keep getting an email saying I
>>>> exceeded the number of submissions, but I haven't tested at all since
>>>> yesterday.
>>>> Can you find out what's happening? My SUID is ashmi.
>>>> Thanks
>>>> ashmi
>>>>
>>>> On 2 Feb 2002, Arun Upadhyaya Kishan wrote:
>>>>
>>>>
>>>>> I received notice that students have been able to successfully 
>>>>> submit and
>>>>> receive their results as of 2:00 pm.
>>>>>
>>>>> Please let us know if the problems continue.
>>>>>
>>>>> Arun
>>>>>
>>>>>
>>>>> Shankar Agarwal  wrote:
>>>>> : Hi,
>>>>> : Can someone please look into the matter. I am not able to submit 
>>>>> my test
>>>>> : for testing. It says i have exceeded the limit. Its 1:00 o'clock 
>>>>> and the
>>>>> : time is ticking fast.
>>>>> : Shankar
>>>>>
>>>>>
>>>>>
>>>> ############################################################################### 
>>>>
>>>>
>>>>                 ASHMI CHOKSHI
>>>> Graduate Student                    141L Escondido Village
>>>> Dept. of Computer Science                Stanford University
>>>> Stanford University                    Stanford Ca 94305
>>>> Stanford Ca 94305                    (650)498-1103
>>>>
>>>> ############################################################################### 
>>>>
>>>>
>>>>
>>>>
>>>>
>>>>
>>
>> ############################################################################### 
>>
>>
>>                 ASHMI CHOKSHI
>> Graduate Student                    141L Escondido Village
>> Dept. of Computer Science                Stanford University
>> Stanford University                    Stanford Ca 94305
>> Stanford Ca 94305                    (650)498-1103
>>
>> ############################################################################### 
>>
>>
>>
>>
>>
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: How to figure out if peer doesn't exist
Date: 3 Feb 2002 00:56:48 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3073

Dhawal Kumar  writes:

>The only indication we get is error in network_recv when we are waiting
>for a reply. But that can be a temporary failure. I looked at the errno
>variable and couldn't find something very relevant. In fact when I tested
>errno I found that it matched none of the values in documentation of
>recvfrom which is called by network_recv. Should I take that to be an
>indication?


What error did you see from recvfrom()?
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: clarification
Date: Sat, 2 Feb 2002 16:58:23 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3074

On Sat, 2 Feb 2002, Lin Lu wrote:

> Hi,
>
> I am not sure about several things.  Could anyone clarify?
>
> 1. before send out FIN, STCP just need to flush all the data in send
> window that has not been acknowledged. right ?  Does it have to wait for
> all the acknowledgement?  If it does not receive the acknowledgement
> again, reflush the data in send window again?
> Also, the passive does not do anything but just close its connection and
> send out FIN_ACK after it receives FIN. right?

 I assume that all flushed data must be acknowledged before shutdown,
 however it the data is not acknowledged after 6 timeouts, the connection
 is dropped.

> 2. On assignment discription page 4, under overview, it says it is
> STCP's job to break up the data into packets and reassemble the data on
> the other side.  I understand STCP will break up the data if it reads
> more than 536 from application.  But if there is no packet reordering on
> the receiver side, the receiver side STCP just send the data to its
> application right away once it receives the packet, no matter how big
> the packet is.  So what does the reassemble mean here?

STCP gaurantees in order delivery so really it is responsible for
reassembling the stream in the advent that the packets get dropped,
or arrive out of order. The receiving end must have some mechanism
for tracking the ordering of packets, and cannot send packets right
upon receipt.

                        HTH,
                        ~~m


> Thanks a lot.
>
> Lin
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: hw2.B test script
Date: Sat, 02 Feb 2002 17:05:30 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3075

Some tests would just work fine with the client from hw2.A, so it should 
come to no one's surprise. In any case just to be 100% sure I will have 
yet another close look to all these cases. If it turns out to be broken, 
you will get a refund for your broken tests ;)

Pablo

BAEHOPIL wrote:

>  
> 
> It seems that hw2.B is not ready yet..
> 
> I test_submitted my dumb hw2.A code to hw2.B test script,
> 
> but it said OK for most of the tests, notably 2.B, 2.D, 2.E, 2.M, and 
> more...
> 
> ( It cannot be OK without any provision for timeout or reassembly, etc.. )
> 
>  
> 
> I wonder when we can actually use it..
> 
>  
> 
> Hopil
> 
>  
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Regarding timeout & retransmission
Date: 3 Feb 2002 01:08:59 GMT
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3076


>I want to confirm my understanding of the STCP specification...

>Suppose there are 3 packets (A,B,C), which are first sent at 0ms, 300ms, =
>400ms
>and there's no more to send and somehow no ACK is coming back..

>And suppose RTO =3D=3D500ms..
>Then packet A will timeout at 500ms.. and then, because of Go-back-N,=20
>all of A,B and C are retransmitted at this moment
>And RTO will be updated...
>If the updated RTO is 1000ms, then they will all timeout at 1500ms..and =
>retransmit.. so on..

>On the other hand, if we received ACK for A at 400ms, A is discarded =
>from the queue...
>B will timeout at 800ms... Again, if ACK for B is received before 800ms,
>C will timeout at 900ms, so on...

>Thank you in advance for any comments or corrections..

>Hopil


Yes, that's correct.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Check about the flags
Date: 3 Feb 2002 01:11:45 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3077


>I am striving to know why I have a FIN problem in grading, although it
>works perfectly when I test my RX/TX

>I want to check something, when I am expecting a FIN header, ALL THE FLAGS
>SHOULD BE ZERO EXCEPT THE FIN BIT ( and it is the same for all the other
>cases). What I mean is that I should expect only the relevant flag to be
>set, while all the others are zero by default?


Yes, all flags should be zero except for those that are explicitly intended
to be set, e.g. the FIN bit, or FIN + ACK bits in the case of a FIN-ACK.

(Flags are not zero "by default" unless you've explicitly initialized them).
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Last Question --- Question about specification
Date: 3 Feb 2002 01:20:51 GMT
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3078


>I noticed in the specification the following:

>"i.e. all data already written to the connection must be sent before the
>FIN is sent."

>However, to send the FIN, the read function must return 0. That means
>there will be no data to send. This is why I ignored this sentence and
>thought that it may be for part2. Please correct me if I am wrong, maybe
>this will solve the FIN script error that I got.


Not necessarily--depending on your implementation, you might have data
waiting to be sent in your sender window.  But the sentence should
really read that outstanding data should be *acknowledged* by the peer.

(Otherwise, since the FIN segment is not required to have a sequence number
in STCP, data could be lost if it were not ACKed before the FIN was sent).
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lin Lu 
Newsgroups: su.class.cs244a
Subject: Re: clarification
Date: Sat, 02 Feb 2002 17:27:07 -0800
Lines: 40
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3079

Martin Casado wrote:
> 
> On Sat, 2 Feb 2002, Lin Lu wrote:
> 
> > Hi,
> >
> > I am not sure about several things.  Could anyone clarify?
> >
> > 1. before send out FIN, STCP just need to flush all the data in send
> > window that has not been acknowledged. right ?  Does it have to wait for
> > all the acknowledgement?  If it does not receive the acknowledgement
> > again, reflush the data in send window again?
> > Also, the passive does not do anything but just close its connection and
> > send out FIN_ACK after it receives FIN. right?
> 
>  I assume that all flushed data must be acknowledged before shutdown,
>  however it the data is not acknowledged after 6 timeouts, the connection
>  is dropped.
> 
> > 2. On assignment discription page 4, under overview, it says it is
> > STCP's job to break up the data into packets and reassemble the data on
> > the other side.  I understand STCP will break up the data if it reads
> > more than 536 from application.  But if there is no packet reordering on
> > the receiver side, the receiver side STCP just send the data to its
> > application right away once it receives the packet, no matter how big
> > the packet is.  So what does the reassemble mean here?
> 
> STCP gaurantees in order delivery so really it is responsible for
> reassembling the stream in the advent that the packets get dropped,
> or arrive out of order. The receiving end must have some mechanism
> for tracking the ordering of packets, and cannot send packets right
> upon receipt.
> 
>                         HTH,
>                         ~~m

I understand STCP will reassemble data in the advent that the packets
get dropped or arrive out of order.  But what about when packets arrive
in the order receiver expects? What I am confused about is what is said
on discription page 4 under overview section.  Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Error in retransmission specification
Date: Sat, 2 Feb 2002 17:30:20 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3080


Hi,

In the retransmission specification it is stated that:

EstimatedRTT = min(20ms,0.8*EstimatedRTT + (.2O*SampleRTT))

Should that be instead max(..)? Especially since we set the
inital value to 250?  Pleas correct me is I am wrong. Thanks.

                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Retransmissions
Date: 3 Feb 2002 01:35:50 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3081


>  In the Go-Back-N protocol if a segment N is retransmitted,
>  all data greater than N is also retransmitted.  Does that
>  mean that the timers for all segments above N is reset and
>  their timeout counter is increased by one? Thoughts? Thanks :-)

>                        ~~m


It's up to you.  However, updating the timeout counter probably unfairly
penalizes segments retransmitted as part of Go-back-N--I only incremented
it when the segment's timer itself expired.  (I found it necessary to reset
the timer for segments retransmitted as part of Go-back-N; otherwise for
large file transfers I'd frequently find my connection timing out).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: clarification
Date: 3 Feb 2002 01:38:06 GMT
Lines: 8
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3082


>I understand STCP will reassemble data in the advent that the packets
>get dropped or arrive out of order.  But what about when packets arrive
>in the order receiver expects? What I am confused about is what is said
>on discription page 4 under overview section.  Thanks

Even if packets arrive in the order the receiver expects, the application
shouldn't see packets, but rather a reliable byte stream.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Error in retransmission specification
Date: 3 Feb 2002 01:39:07 GMT
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3083

Martin Casado  writes:


>Hi,

>In the retransmission specification it is stated that:

>EstimatedRTT = min(20ms,0.8*EstimatedRTT + (.2O*SampleRTT))

>Should that be instead max(..)? Especially since we set the
>inital value to 250?  Pleas correct me is I am wrong. Thanks.

>                    ~~m


Yep, thanks!

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: debug log from black box testing.
Date: Sat, 2 Feb 2002 17:41:58 -0800
Lines: 107
Distribution: su
Message-ID: 
References: 
  
 
 
 
NNTP-Posting-Host: myth4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Pablo Molinero Fernandez 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3084


Hi Pablo,

I wonder if the same holds good for black box testing :-)

We can stop making wild guesses and submitting to the test
server if we can get atleast a log of the full run with our
debug messages. The information from the failed test runs
are cryptic and far from useful.

I am sure others who had been desperately been trying to
pass the test scripts for subtle mistakes are feeling the
same way.

It will definitely help the learning process in the long run.

my 2 cents.

Thanks, Arvind


On Sat, 2 Feb 2002, Pablo Molinero Fernandez wrote:

> Hi,
>
> When reporting an error, please, copy the message that you see on the
> screen. It makes debugging easier. Thank you.
>
> Pablo
>
> Ashmi wrote:
>
> > I haven't been able to test at all since yesterday; essentially the
> > extended deadline has not proved to be at all useful to me!
> > ashmi
> >
> >
> > On Sat, 2 Feb 2002, Arvind Bhat wrote:
> >
> >
> >>Same here...
> >>I have already emailed my TA and no response from him either :(
> >>
> >>-Arvind
> >>
> >>On Sat, 2 Feb 2002, Ashmi wrote:
> >>
> >>
> >>>hi,
> >>>I still can't submit for testing, I keep getting an email saying I
> >>>exceeded the number of submissions, but I haven't tested at all since
> >>>yesterday.
> >>>Can you find out what's happening? My SUID is ashmi.
> >>>Thanks
> >>>ashmi
> >>>
> >>> On 2 Feb 2002, Arun Upadhyaya Kishan wrote:
> >>>
> >>>
> >>>>I received notice that students have been able to successfully submit and
> >>>>receive their results as of 2:00 pm.
> >>>>
> >>>>Please let us know if the problems continue.
> >>>>
> >>>>Arun
> >>>>
> >>>>
> >>>>Shankar Agarwal  wrote:
> >>>>: Hi,
> >>>>: Can someone please look into the matter. I am not able to submit my test
> >>>>: for testing. It says i have exceeded the limit. Its 1:00 o'clock and the
> >>>>: time is ticking fast.
> >>>>: Shankar
> >>>>
> >>>>
> >>>>
> >>>###############################################################################
> >>>
> >>>				ASHMI CHOKSHI
> >>>Graduate Student				    141L Escondido Village
> >>>Dept. of Computer Science			    Stanford University
> >>>Stanford University				    Stanford Ca 94305
> >>>Stanford Ca 94305				    (650)498-1103
> >>>
> >>>###############################################################################
> >>>
> >>>
> >>>
> >>>
> >>>
> >
> > ###############################################################################
> >
> > 				ASHMI CHOKSHI
> > Graduate Student				    141L Escondido Village
> > Dept. of Computer Science			    Stanford University
> > Stanford University				    Stanford Ca 94305
> > Stanford Ca 94305				    (650)498-1103
> >
> > ###############################################################################
> >
> >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine25.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: karn's algorithm
Date: 3 Feb 2002 01:42:26 GMT
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine25.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3085

Abhishek Das  writes:


>It is understood that we don't take the sampleRTT for retransmitted
>segments. But are we supposed to double the RTO as explained in class or
>keep the RTO same as before?


I suspect you'll find that your connections will time out quite frequently,
especially for large file transfers, if you don't use exponential backoff.
If this is the case, I'd suggest doubling RTO on retransmissions.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine25.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Error meaning
Date: 3 Feb 2002 01:55:20 GMT
Lines: 57
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine25.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3086

Moh'd Saleem Saleem Alrawashdeh  writes:

>Sorry again to flood the list, but I read all the previous postings and
>till now I cannot exactely understand the test script output. For example,
>below is the test that says that my tx and rx aren't working with each
>other. What is the meaning of 0 0 0 TestData2? What are these zeros?


The numbers correspond to the output of diff between the 'wc' result for
(1) a file transferred using your implementation and (2) the reference
original, and then likewise for the file checksums.

So in the example below, nothing was transferred, but 5098 bytes were
expected to have been.


>===============

>    * Test 1.J: If the Tx and Rx fails with our Rx and our Tx but works
>with each other.
>  [out of 14]


>[SUMMARY] NOT OK
>[RESULT]
>===========================
>===(T1.Ja) Byte count differences found (INCORRECT!)
>Student's Tx against student's Rx (reliable mode)

>---

>---
>Difference between expected result and your result:
>3c3
><       0       0       0 TESTDATA.2
>---
>>     131     501    5098 TESTDATA.2

>---

>===========================
>===(T1.Jb) Checksum differences found (INCORRECT!)
>Student's Tx against student's Rx (reliable mode)

>---

>---
>Difference between expected result and your result:
>3c3
>< 4294967295 0 TESTDATA.2
>---
>> 2181702833 5098 TESTDATA.2

>---



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Retransmission Question
Date: Sat, 2 Feb 2002 18:02:38 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3087

STCP is a Go-back-N protocol.  So say I transmit 5 segments and I get ACKs
back for the all the segments except for the first.  For my re-transmission
I start transmitting at the initial sequence number and re-transmit the data
covering all 5 segments.  For segments 2-5 does that re-transmission count
toward the total of 6 retransmission per segment or is that count only
updated for the first segment lost?  Thanks

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission Question
Date: Sat, 2 Feb 2002 18:34:00 -0800
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3088


I really don't see the point of the Go-back-N protocol... why would you
resend segments that have already been acknowledged?

 Can we just send all bytes from N that haven't been acknowledged? This
 makes quite a bit more sense to me.

                    ~~m

> STCP is a Go-back-N protocol.  So say I transmit 5 segments and I get ACKs
> back for the all the segments except for the first.  For my re-transmission
> I start transmitting at the initial sequence number and re-transmit the data
> covering all 5 segments.  For segments 2-5 does that re-transmission count
> toward the total of 6 retransmission per segment or is that count only
> updated for the first segment lost?  Thanks
>
> --Russ
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lin Lu 
Newsgroups: su.class.cs244a
Subject: Re: clarification
Date: Sat, 02 Feb 2002 18:42:29 -0800
Lines: 17
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3089

Matthew Jonathan Holliman wrote:
> 
> >I understand STCP will reassemble data in the advent that the packets
> >get dropped or arrive out of order.  But what about when packets arrive
> >in the order receiver expects? What I am confused about is what is said
> >on discription page 4 under overview section.  Thanks
> 
> Even if packets arrive in the order the receiver expects, the application
> shouldn't see packets, but rather a reliable byte stream.

I don't understand.  If next expect byte seq number is 10, receiver
receiver receives 100 bytes starting from seq number 10, it can deliver
the 100 bytes to application without violating reliable byte stream
definition.  Isn't it?  I did not see which application should not see
this packet and what time the application should see the packet.

Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Retransmission Question
Date: 3 Feb 2002 02:46:12 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine11.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3090


>STCP is a Go-back-N protocol.  So say I transmit 5 segments and I get ACKs
>back for the all the segments except for the first.  For my re-transmission
>I start transmitting at the initial sequence number and re-transmit the data
>covering all 5 segments.  For segments 2-5 does that re-transmission count
>toward the total of 6 retransmission per segment or is that count only
>updated for the first segment lost?  Thanks


Remember that the other side has ACKed with its expected sequence number.
So if any segment after that first is acknowledged, the first is
implicitly acknowledged also.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: clarification
Date: 3 Feb 2002 02:48:05 GMT
Lines: 17
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: elaine11.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3091


>> >I understand STCP will reassemble data in the advent that the packets
>> >get dropped or arrive out of order.  But what about when packets arrive
>> >in the order receiver expects? What I am confused about is what is said
>> >on discription page 4 under overview section.  Thanks
>> 
>> Even if packets arrive in the order the receiver expects, the application
>> shouldn't see packets, but rather a reliable byte stream.

>I don't understand.  If next expect byte seq number is 10, receiver
>receiver receives 100 bytes starting from seq number 10, it can deliver
>the 100 bytes to application without violating reliable byte stream
>definition.  Isn't it?  I did not see which application should not see
>this packet and what time the application should see the packet.

Yes, that's right.  Could you clarify what your question is?

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission Question
Date: 3 Feb 2002 02:54:03 GMT
Lines: 36
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine34.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3092

You will not get ACKs for the packets since the ACK always indicates the
next byte expected, which will be the dropped first packet (assuming
segments 1-5 are sequential). Hence, when an ACK for the first packet does
not appear within the timeout period, it is unclear which of the 5 packets
made it across. Thus, a simple mechanism around this is to retransmit all
packets.

There are optimizations which will note repeating ACKs that are requesting
the first packet. TCP can then short circuit the timeout period and resend
segment 1, assuming it has been lost. You do not have to implement this
for STCP.

Arun

Martin Casado  wrote:

: I really don't see the point of the Go-back-N protocol... why would you
: resend segments that have already been acknowledged?

:  Can we just send all bytes from N that haven't been acknowledged? This
:  makes quite a bit more sense to me.

:                     ~~m

:> STCP is a Go-back-N protocol.  So say I transmit 5 segments and I get ACKs
:> back for the all the segments except for the first.  For my re-transmission
:> I start transmitting at the initial sequence number and re-transmit the data
:> covering all 5 segments.  For segments 2-5 does that re-transmission count
:> toward the total of 6 retransmission per segment or is that count only
:> updated for the first segment lost?  Thanks
:>
:> --Russ
:>
:>
:>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine26.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: How to figure out if peer doesn't exist
Date: Sat, 2 Feb 2002 21:50:12 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine26.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3093

Your reply got lost in the flood.

When I do a switch errno after
receiving -1 on read, I get an "Unrecgnised error" - I mean that the errno
doesn't match any of

EBADF, EINTR, EIO, ENOMEM, ENOSR, ENOTSOCK, ESTALE, EWOULDBLOCK

which are the possible values of errno mentioned in documentation.

Dhawal Kumar

On 3 Feb 2002, Matthew Jonathan Holliman wrote:

> Dhawal Kumar  writes:
>
> >The only indication we get is error in network_recv when we are waiting
> >for a reply. But that can be a temporary failure. I looked at the errno
> >variable and couldn't find something very relevant. In fact when I tested
> >errno I found that it matched none of the values in documentation of
> >recvfrom which is called by network_recv. Should I take that to be an
> >indication?
>
>
> What error did you see from recvfrom()?
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Victor Wong" 
Newsgroups: su.class.cs244a
Subject: hw2.A test script
Date: Sat, 2 Feb 2002 22:33:10 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: summerbridge.stanford.edu
X-Trace: news.Stanford.EDU 1012717998 19265 128.12.82.21 (3 Feb 2002 06:33:18 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3094

Hi,

Is it possible for the TAs to leave the hw2.A test script in place for us to
test our hw2.B against? Since the test script for hw2.B is not cumulative
(ie it doesn't include the tests from hw2.A), we would not be able to detect
any bugs that we introduce into our basic functionality as a result of
adding code for part B. With the hw2.A test script, we can at least do some
sort of regression testing to make sure be didn't break anything from
before. Thanks!

Victor.


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga9.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: SYN retransmit
Date: Sat, 2 Feb 2002 23:00:48 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3095


This must be reallystupid, but do we retransmit SYN if we don't get a
SYN_ACK in
500ms (2 * initial estimatedRTT) ?

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine19.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: How to figure out if peer doesn't exist
Date: 3 Feb 2002 07:13:09 GMT
Lines: 11
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine19.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3096


>When I do a switch errno after
>receiving -1 on read, I get an "Unrecgnised error" - I mean that the errno
>doesn't match any of

>EBADF, EINTR, EIO, ENOMEM, ENOSR, ENOTSOCK, ESTALE, EWOULDBLOCK

>which are the possible values of errno mentioned in documentation.


Okay, but what's the value of errno?
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine19.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: SYN retransmit
Date: 3 Feb 2002 07:14:23 GMT
Lines: 7
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine19.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3097


>This must be reallystupid, but do we retransmit SYN if we don't get a
>SYN_ACK in
>500ms (2 * initial estimatedRTT) ?

Yep, SYN and FIN should be retransmitted if an ACK isn't received in time,
just like data segments.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Yashar Ganjali 
Newsgroups: su.class.cs244a
Subject: Retransmission.
Date: Sun, 03 Feb 2002 05:47:14 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: obi.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3098

Hi,

In a Go-back-N protocol, if the transport layer decides (for any reason) 
that the peer has not received the segment N, it retransmits all data 
starting at N. What happens if we have a segment, say X, with a sequence 
number less than N which has not been ACKed yet? Do we need to send the 
segment X too? Or we just send segments >= N and keep waiting for an ACK 
for X (and of course all other transmitted segments) and the timeout for X?

--Yashar

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: Re: hw2.A test script
Date: Sun, 3 Feb 2002 06:51:02 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3099

I second the opinion! It would indeed be useful to have the test script
for hw2.A around.

Thanks,
Ashmi

On Sat, 2 Feb 2002, Victor Wong wrote:

> Hi,
>
> Is it possible for the TAs to leave the hw2.A test script in place for us to
> test our hw2.B against? Since the test script for hw2.B is not cumulative
> (ie it doesn't include the tests from hw2.A), we would not be able to detect
> any bugs that we introduce into our basic functionality as a result of
> adding code for part B. With the hw2.A test script, we can at least do some
> sort of regression testing to make sure be didn't break anything from
> before. Thanks!
>
> Victor.
>
>
>

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: 3 Feb 2002 19:37:19 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3100

If Segment N has not been ACKed, and X < N, then X cannot have been ACKed 
eitehr as per the rules of (S)TCP. You can assume that timeouts are in 
order for the most part, so that if packet a is sent before packet b, then 
a will timeout before b. Thus X should timeout before N, and all packets 
>= X will be retransmitted upon timeout.

Arun

Yashar Ganjali  wrote:
: Hi,

: In a Go-back-N protocol, if the transport layer decides (for any reason) 
: that the peer has not received the segment N, it retransmits all data 
: starting at N. What happens if we have a segment, say X, with a sequence 
: number less than N which has not been ACKed yet? Do we need to send the 
: segment X too? Or we just send segments >= N and keep waiting for an ACK 
: for X (and of course all other transmitted segments) and the timeout for X?

: --Yashar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Sun, 3 Feb 2002 12:02:20 -0800
Lines: 40
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3101


I have a related question..

When timeout occurs, we update RTO..
Suppose X sent at 100ms, N sent at 300ms, Initial RTO=500ms..
Then without any ACK, X will timeout at 600ms and RTO becomes 1000ms
Next timeout of X will be at 1600ms,
but, how about N?  Will N timeout at 800ms or 1300ms?

And when & how shall we restore RTO to 2*RTT?
Maybe it's up to us to determine?

Hopil


"Arun Upadhyaya Kishan"  wrote in message

> If Segment N has not been ACKed, and X < N, then X cannot have been ACKed
> eitehr as per the rules of (S)TCP. You can assume that timeouts are in
> order for the most part, so that if packet a is sent before packet b, then
> a will timeout before b. Thus X should timeout before N, and all packets
> >= X will be retransmitted upon timeout.
>
> Arun
>
> Yashar Ganjali  wrote:
> : Hi,
>
> : In a Go-back-N protocol, if the transport layer decides (for any reason)
> : that the peer has not received the segment N, it retransmits all data
> : starting at N. What happens if we have a segment, say X, with a sequence
> : number less than N which has not been ACKed yet? Do we need to send the
> : segment X too? Or we just send segments >= N and keep waiting for an ACK
> : for X (and of course all other transmitted segments) and the timeout for
X?
>
> : --Yashar
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Sun, 03 Feb 2002 12:21:26 -0800
Lines: 32
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3102

So we can ignore then case, when i send 2 packets with seq numbers 100
and 200.  I then send a third packet with seq 300.  I then hear acks on
packets 100 and 200... and the estRTT is lowered such that the next
packet i send (packet with seq 400) will timeout before packet with seq
300.

So we just resend the entire window on a timeout?

Thanks
Clayton

Arun Upadhyaya Kishan wrote:
> 
> If Segment N has not been ACKed, and X < N, then X cannot have been ACKed
> eitehr as per the rules of (S)TCP. You can assume that timeouts are in
> order for the most part, so that if packet a is sent before packet b, then
> a will timeout before b. Thus X should timeout before N, and all packets
> >= X will be retransmitted upon timeout.
> 
> Arun
> 
> Yashar Ganjali  wrote:
> : Hi,
> 
> : In a Go-back-N protocol, if the transport layer decides (for any reason)
> : that the peer has not received the segment N, it retransmits all data
> : starting at N. What happens if we have a segment, say X, with a sequence
> : number less than N which has not been ACKed yet? Do we need to send the
> : segment X too? Or we just send segments >= N and keep waiting for an ACK
> : for X (and of course all other transmitted segments) and the timeout for X?
> 
> : --Yashar
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Retransmissions
Date: Sun, 3 Feb 2002 14:56:14 -0800
Lines: 61
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3103

>
> >  In the Go-Back-N protocol if a segment N is retransmitted,
> >  all data greater than N is also retransmitted.  Does that
> >  mean that the timers for all segments above N is reset and
> >  their timeout counter is increased by one? Thoughts? Thanks :-)
>
> >                        ~~m
>
>
> It's up to you.  However, updating the timeout counter probably unfairly
> penalizes segments retransmitted as part of Go-back-N--I only incremented
> it when the segment's timer itself expired.  (I found it necessary to reset
> the timer for segments retransmitted as part of Go-back-N; otherwise for
> large file transfers I'd frequently find my connection timing out).
>
>

I also find that updating the timeout counter on Go-Back-N causes my connection to
timeout frequently on large files usually around a couple hundred k or so.  If
I don't increament the timeout counter on Go-Back-N, I can transfer arbitrarily
large files (I've tested up to 20 megs) with both the server and the client in
unreliable mode.

Here is the problem, the testing script seems to assume that you are
incrementing the timeout counter for each retransmission, including go-back-n.
If I implement timeouts in Go-Back-N I pass test 2.k,

===============

    * Test 2.K: Tx fails to enforce 5 Retransmissions
  [out of 2.5]

[SUMMARY] OK
[RESULT]  5 retransmissions enforced (CORRECT!)

===============

However, the connection times out frequently on large files.  If I don't
increment the time-out counter on Go-Back-N... I fail the retransmission
test :-(

===============

    * Test 2.K: Tx fails to enforce 5 Retransmissions
  [out of 2.5]

[SUMMARY] NOT OK
[RESULT]  5 retransmissions not enforced(INCORRECT!)
got up to 10 transmissions

---

Has anyone been able to timeout on Go-Back-N and still be able to send arbitrarily
large files wihtout the connection timing out eventually?  Comments, thoughts would
be greatly appreciated. Thanks ;-)

                                ~~m




.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Error meaning
Date: Sun, 03 Feb 2002 15:57:29 -0800
Lines: 89
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3104

Is this a possible result of a Tx/Rx hanging and ultimately being 
killed? I think I'd run into this during my first time through the tests.

-Sandeep

Matthew Jonathan Holliman wrote:

> Moh'd Saleem Saleem Alrawashdeh  writes:
> 
> 
>>Sorry again to flood the list, but I read all the previous postings and
>>till now I cannot exactely understand the test script output. For example,
>>below is the test that says that my tx and rx aren't working with each
>>other. What is the meaning of 0 0 0 TestData2? What are these zeros?
>>
> 
> 
> The numbers correspond to the output of diff between the 'wc' result for
> (1) a file transferred using your implementation and (2) the reference
> original, and then likewise for the file checksums.
> 
> So in the example below, nothing was transferred, but 5098 bytes were
> expected to have been.
> 
> 
> 
>>===============
>>
> 
>>   * Test 1.J: If the Tx and Rx fails with our Rx and our Tx but works
>>with each other.
>> [out of 14]
>>
> 
> 
>>[SUMMARY] NOT OK
>>[RESULT]
>>===========================
>>===(T1.Ja) Byte count differences found (INCORRECT!)
>>Student's Tx against student's Rx (reliable mode)
>>
> 
>>---
>>
> 
>>---
>>Difference between expected result and your result:
>>3c3
>><       0       0       0 TESTDATA.2
>>---
>>
>>>    131     501    5098 TESTDATA.2
>>>
> 
>>---
>>
> 
>>===========================
>>===(T1.Jb) Checksum differences found (INCORRECT!)
>>Student's Tx against student's Rx (reliable mode)
>>
> 
>>---
>>
> 
>>---
>>Difference between expected result and your result:
>>3c3
>>< 4294967295 0 TESTDATA.2
>>---
>>
>>>2181702833 5098 TESTDATA.2
>>>
> 
>>---
>>
> 
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga16.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Overlapping Packets
Date: Sun, 3 Feb 2002 17:55:04 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3105


Can we expect overlapping packets as a receiver. ex:
- Packet 1 with seq num = 1,   length 200
- Packet 2 with seq num = 101, length = 200

If the answer is yes, which one should we consider "correct" for the
overlapping part (seq nums 101 to 200 in above example) - 1'st or 2'nd.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic25.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: About the print Congestion window.
Date: Sun, 3 Feb 2002 18:01:53 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic25.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3106

Hi
Can someone please clarify what is fid in the print congestion window. Is
it the udp socket or the tcp socket. As we have both the information in
the ctx structure we can use both but the code says that i can be called
from either network.c or mysock.c which is kind of confusing as each of
them has either tcp information or udp information. Please clarify. Also
is it MIN(20, .......) or MAX(20, .....) for updating the RTT.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga16.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: How to figure out if peer doesn't exist
Date: Sun, 3 Feb 2002 18:38:17 -0800
Lines: 19
Distribution: su
Message-ID: 
References:  
  
 
NNTP-Posting-Host: saga16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3107

errno = 146

Dhawal Kumar

On 3 Feb 2002, Matthew Jonathan Holliman wrote:

>
> >When I do a switch errno after
> >receiving -1 on read, I get an "Unrecgnised error" - I mean that the errno
> >doesn't match any of
>
> >EBADF, EINTR, EIO, ENOMEM, ENOSR, ENOTSOCK, ESTALE, EWOULDBLOCK
>
> >which are the possible values of errno mentioned in documentation.
>
>
> Okay, but what's the value of errno?
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Overlapping Packets
Date: Sun, 3 Feb 2002 22:48:53 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Dhawal Kumar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3108


Hi Dhawal,

If packet 2 came before packet 1, I would think that
when packet 1 arrives, 1 to 100 can be received and
the rest of it dropped.

If packet 1 comes first, drop all of packet 2.

Thanks, Arvind


On Sun, 3 Feb 2002, Dhawal Kumar wrote:

>
> Can we expect overlapping packets as a receiver. ex:
> - Packet 1 with seq num = 1,   length 200
> - Packet 2 with seq num = 101, length = 200
>
> If the answer is yes, which one should we consider "correct" for the
> overlapping part (seq nums 101 to 200 in above example) - 1'st or 2'nd.
>
> Dhawal Kumar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Romain Thibaux 
Newsgroups: su.class.cs244a
Subject: Failed to enforce 5 retransmissions
Date: Sun, 03 Feb 2002 22:58:41 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: thibaux.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1012806298 28847 128.12.191.77 (4 Feb 2002 07:04:58 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.4.2-2 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3109

I just wanted to know if other people than Francois and me have this
error and can't understand why.
It is most probably a bug in my program, but if many are in this case,
who knows...

     Romain


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Failed to enforce 5 retransmissions
Date: Sun, 3 Feb 2002 23:11:46 -0800
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3110


It works fine for me.  Initially I was failing that test because I was not
counting Go-Back-N resends towards the 5 timeouts, but the test script
expects that you do.  I do think this is a problem, because I find that
if you do count Go-Back-N retransmissions towards the 5, the connection
is more likely to time out.

                        ~~m

> I just wanted to know if other people than Francois and me have this
> error and can't understand why.
> It is most probably a bug in my program, but if many are in this case,
> who knows...
>
>      Romain
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: francois-marie lefevere 
Newsgroups: su.class.cs244a
Subject: test script results.
Date: Sun, 03 Feb 2002 23:30:32 -0800
Lines: 36
Distribution: su
Message-ID: 
NNTP-Posting-Host: nastia.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1012807833 29116 128.12.187.116 (4 Feb 2002 07:30:33 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.4.9-12 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3111

I got errors for SYN and FIN sequences but the script does not give me
any trace of what happened.
What does this mean ?
(If my program crashes before SYN, I can't understand that everything
else is right.)



    * Test 2.G: Failure to send SYN_ACK after a SYN
  [out of 5]

[SUMMARY] NOT OK
[RESULT]  did not send SYN_ACK after SYN

---


---
 (INCORRECT!)


===============

    * Test 2.H: Failure to send all packets and make sure they are ACKed
before the FIN.
  [out of 5]

[SUMMARY] NOT OK
[RESULT]  FIN sent before Acking all data

---


---
 (INCORRECT!)

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Failed to enforce 5 retransmissions
Date: Sun, 3 Feb 2002 23:29:13 -0800
Lines: 39
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3112


Hi..

I think we can see it from a different angle..
Even if a packet is retransmitted because of Go-Back-N, not because of its
own timeout,
it is still true that it is actually retransmitted and its probability of
reaching the destination actually rises..
I mean,  can we not think of the count as a "retransmit count" rather than a
"timeout count"?
Anyway, I just wish that some authoritative person gives us a definite
answer...

And, I also wonder whether the test script is also giving us NOT OK if our
code gives up before
making 5 retransmissions..

Hopil

"Martin Casado"  wrote in message

>
> It works fine for me.  Initially I was failing that test because I was not
> counting Go-Back-N resends towards the 5 timeouts, but the test script
> expects that you do.  I do think this is a problem, because I find that
> if you do count Go-Back-N retransmissions towards the 5, the connection
> is more likely to time out.
>
>                         ~~m
>
> > I just wanted to know if other people than Francois and me have this
> > error and can't understand why.
> > It is most probably a bug in my program, but if many are in this case,
> > who knows...
> >
> >      Romain



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: FIN questions
Date: Mon, 4 Feb 2002 00:17:55 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3113

For Part B,

The handout says 'The FIN packet is subject to the same rules of
retransmission as any other segment' but also that 'If the active end never
receives a FIN-ACK, as per the rules of transmission, it terminates anyway'.

Does this mean that we send the FIN packet up to a total of 6 times, and if
we never get a FIN-ACK, we just disconnect?

thanks,
Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: 4 Feb 2002 09:04:09 GMT
Lines: 23
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3114

"BAEHOPIL"  writes:


>I have a related question..

>When timeout occurs, we update RTO..
>Suppose X sent at 100ms, N sent at 300ms, Initial RTO=500ms..
>Then without any ACK, X will timeout at 600ms and RTO becomes 1000ms
>Next timeout of X will be at 1600ms,
>but, how about N?  Will N timeout at 800ms or 1300ms?

>And when & how shall we restore RTO to 2*RTT?
>Maybe it's up to us to determine?

It's probably more reasonable for N's timeout to be left unchanged
by the update in RTO.  (I think it's definitely easier to implement,
especially if you use exponential back-off).

Keep in mind that the RTO/ERTT are very coarse estimates, which include
some history.  So if the values change a bit following the reception
of an ACK, it's not necessarily particularly useful or meaningful to
update following unacked packets' timeouts/ERTTs given this extra
information.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: 4 Feb 2002 09:06:24 GMT
Lines: 11
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3115


>So we can ignore then case, when i send 2 packets with seq numbers 100
>and 200.  I then send a third packet with seq 300.  I then hear acks on
>packets 100 and 200... and the estRTT is lowered such that the next
>packet i send (packet with seq 400) will timeout before packet with seq
>300.

>So we just resend the entire window on a timeout?

In the example you mention, given the way the ERTT/timeouts work out, you
would.  (But note that this is not always the case).
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Retransmissions
Date: 4 Feb 2002 09:14:47 GMT
Lines: 16
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3116


>I also find that updating the timeout counter on Go-Back-N causes my connection to
>timeout frequently on large files usually around a couple hundred k or so.  If
>I don't increament the timeout counter on Go-Back-N, I can transfer arbitrarily
>large files (I've tested up to 20 megs) with both the server and the client in
>unreliable mode.

>Here is the problem, the testing script seems to assume that you are
>incrementing the timeout counter for each retransmission, including go-back-n.
>If I implement timeouts in Go-Back-N I pass test 2.k,


I suspect there's a bug (or at least a difference of opinion that must be
resolved) in the test script for this case, but I'll have to confirm with
Pablo and get back to you.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Error meaning
Date: 4 Feb 2002 09:16:31 GMT
Lines: 5
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3117


>Is this a possible result of a Tx/Rx hanging and ultimately being 
>killed? I think I'd run into this during my first time through the tests.

Yes, this is quite possible.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: test script results.
Date: Mon, 4 Feb 2002 01:21:08 -0800
Lines: 47
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012814469 29994 128.12.133.48 (4 Feb 2002 09:21:09 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3118

I had this exact same result at one point--passing everything but G & H--and
the problem turned out to be the code I'd written to allow
print_congestion_window_and_estimated_rtt() to be called from the parent
process. I undid all the changes related to that and the test worked again.

Jonathan

"francois-marie lefevere"  wrote in message

> I got errors for SYN and FIN sequences but the script does not give me
> any trace of what happened.
> What does this mean ?
> (If my program crashes before SYN, I can't understand that everything
> else is right.)
>
>
>
>     * Test 2.G: Failure to send SYN_ACK after a SYN
>   [out of 5]
>
> [SUMMARY] NOT OK
> [RESULT]  did not send SYN_ACK after SYN
>
> ---
>
>
> ---
>  (INCORRECT!)
>
>
> ===============
>
>     * Test 2.H: Failure to send all packets and make sure they are ACKed
> before the FIN.
>   [out of 5]
>
> [SUMMARY] NOT OK
> [RESULT]  FIN sent before Acking all data
>
> ---
>
>
> ---
>  (INCORRECT!)
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Overlapping Packets
Date: 4 Feb 2002 09:22:48 GMT
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3119

Dhawal Kumar  writes:


>Can we expect overlapping packets as a receiver. ex:
>- Packet 1 with seq num = 1,   length 200
>- Packet 2 with seq num = 101, length = 200

>If the answer is yes, which one should we consider "correct" for the
>overlapping part (seq nums 101 to 200 in above example) - 1'st or 2'nd.


The one that is correct is arbitrary--if they're different, the receiver
has no way of knowing which one to use... but this case should not happen.

Your receiver should be able to correctly handle such cases.  (I doubt
there's any particular effort required to support this, since it's basically
the functionality that's required of the receiver window anyway).
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: About the print Congestion window.
Date: 4 Feb 2002 09:30:24 GMT
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3120


>Hi
>Can someone please clarify what is fid in the print congestion window. Is
>it the udp socket or the tcp socket. As we have both the information in
>the ctx structure we can use both but the code says that i can be called
>from either network.c or mysock.c which is kind of confusing as each of
>them has either tcp information or udp information. Please clarify. Also
>is it MIN(20, .......) or MAX(20, .....) for updating the RTT.

It's sockfd.

One or two people had asked what the function should do.  To clarify:

It should print out just the STCP window size for CWND.  This is rather
meaningless now; it's a vestige of past years' assignments.

More important is that the function must print out your current estimated
RTT, which is used by the test script to verify that your code is updating
the RTT as specified by STCP.

There are a couple of dummy variables in the function, which you should
replace appropriately.

min vs. max was posted earlier and updated on the web page (if you
think about it, you should also realize the answer).
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: FIN questions
Date: 4 Feb 2002 09:35:33 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3121


>For Part B,

>The handout says 'The FIN packet is subject to the same rules of
>retransmission as any other segment' but also that 'If the active end never
>receives a FIN-ACK, as per the rules of transmission, it terminates anyway'.

>Does this mean that we send the FIN packet up to a total of 6 times, and if
>we never get a FIN-ACK, we just disconnect?

>thanks,
>Pete


Yes... you don't really have many choices in this situation.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: Retransmissions
Date: Mon, 4 Feb 2002 01:56:26 -0800
Lines: 95
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012816587 275 128.12.133.48 (4 Feb 2002 09:56:27 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3122

Turns out when you've got the server and client both in unreliable mode it's
pretty darn unreliable. The number of retransmissions should probably be set
for everything you retransmit, regardless of whether it was via go-back-N or
via its own timeout--the fact is it was resent, and it was never
acknowledged.

I suspect what's going on is a result of however you decided to transition
from 2*RTO back to 2*RTT after a successful send. If you do it too quickly,
you'll get mad timeouts. By a small change to my transition mechanism, I was
able to transfer a 42MB file in unreliable mode on both sides without a
problem. (It was slow because of all the retransmitting that was going on,
but it transferred.) Of course, changing your retransmission count has a
similar effect, since more retransmissions means more timeout cycles which
gives ACKs a longer time to get through. Now, I haven't passed that code
through the script yet to see if it likes how I calculate the RTTs...

Either way, your solution that times out a lot seems to adhere to the spec,
and thus I would bet it's OK. I ain't a grader though...

Jonathan

"Martin Casado"  wrote in message

> >
> > >  In the Go-Back-N protocol if a segment N is retransmitted,
> > >  all data greater than N is also retransmitted.  Does that
> > >  mean that the timers for all segments above N is reset and
> > >  their timeout counter is increased by one? Thoughts? Thanks :-)
> >
> > >                        ~~m
> >
> >
> > It's up to you.  However, updating the timeout counter probably unfairly
> > penalizes segments retransmitted as part of Go-back-N--I only
incremented
> > it when the segment's timer itself expired.  (I found it necessary to
reset
> > the timer for segments retransmitted as part of Go-back-N; otherwise for
> > large file transfers I'd frequently find my connection timing out).
> >
> >
>
> I also find that updating the timeout counter on Go-Back-N causes my
connection to
> timeout frequently on large files usually around a couple hundred k or so.
If
> I don't increament the timeout counter on Go-Back-N, I can transfer
arbitrarily
> large files (I've tested up to 20 megs) with both the server and the
client in
> unreliable mode.
>
> Here is the problem, the testing script seems to assume that you are
> incrementing the timeout counter for each retransmission, including
go-back-n.
> If I implement timeouts in Go-Back-N I pass test 2.k,
>
> ===============
>
>     * Test 2.K: Tx fails to enforce 5 Retransmissions
>   [out of 2.5]
>
> [SUMMARY] OK
> [RESULT]  5 retransmissions enforced (CORRECT!)
>
> ===============
>
> However, the connection times out frequently on large files.  If I don't
> increment the time-out counter on Go-Back-N... I fail the retransmission
> test :-(
>
> ===============
>
>     * Test 2.K: Tx fails to enforce 5 Retransmissions
>   [out of 2.5]
>
> [SUMMARY] NOT OK
> [RESULT]  5 retransmissions not enforced(INCORRECT!)
> got up to 10 transmissions
>
> ---
>
> Has anyone been able to timeout on Go-Back-N and still be able to send
arbitrarily
> large files wihtout the connection timing out eventually?  Comments,
thoughts would
> be greatly appreciated. Thanks ;-)
>
>                                 ~~m
>
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga0.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Mon, 4 Feb 2002 02:07:56 -0800
Lines: 21
Distribution: su
Message-ID: 
References:  
  
NNTP-Posting-Host: saga0.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3123

Does this mean that irrespective of what packet in window times out we
resend the whole window? Or is there a situation where we don't - If yes
can you give me an example?

Dhawal Kumar

On 4 Feb 2002, Matthew Jonathan Holliman wrote:

>
> >So we can ignore then case, when i send 2 packets with seq numbers 100
> >and 200.  I then send a third packet with seq 300.  I then hear acks on
> >packets 100 and 200... and the estRTT is lowered such that the next
> >packet i send (packet with seq 400) will timeout before packet with seq
> >300.
>
> >So we just resend the entire window on a timeout?
>
> In the example you mention, given the way the ERTT/timeouts work out, you
> would.  (But note that this is not always the case).
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: test script results.
Date: Mon, 4 Feb 2002 02:21:29 -0800
Lines: 60
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Jonathan Keljo 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3124


Thinking more about my hw2.A testing failures, I have to agree
with you.
I was seeing these failures as soon as i added the code for
printing estimated RTT :-(

I wonder if these failures are legitmate.


On Mon, 4 Feb 2002, Jonathan Keljo wrote:

> I had this exact same result at one point--passing everything but G & H--and
> the problem turned out to be the code I'd written to allow
> print_congestion_window_and_estimated_rtt() to be called from the parent
> process. I undid all the changes related to that and the test worked again.
>
> Jonathan
>
> "francois-marie lefevere"  wrote in message
> 
> > I got errors for SYN and FIN sequences but the script does not give me
> > any trace of what happened.
> > What does this mean ?
> > (If my program crashes before SYN, I can't understand that everything
> > else is right.)
> >
> >
> >
> >     * Test 2.G: Failure to send SYN_ACK after a SYN
> >   [out of 5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  did not send SYN_ACK after SYN
> >
> > ---
> >
> >
> > ---
> >  (INCORRECT!)
> >
> >
> > ===============
> >
> >     * Test 2.H: Failure to send all packets and make sure they are ACKed
> > before the FIN.
> >   [out of 5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  FIN sent before Acking all data
> >
> > ---
> >
> >
> > ---
> >  (INCORRECT!)
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: 4 Feb 2002 19:37:34 GMT
Lines: 27
Distribution: su
Message-ID: 
References:      
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3125

Dhawal Kumar  writes:

>Does this mean that irrespective of what packet in window times out we
>resend the whole window? Or is there a situation where we don't - If yes
>can you give me an example?

Not necessarily.  (If that were the case, time out handling would be quite
simple, and you could maintain only a single timeout for the entire window).

Consider this example:

The sender window contains just a single new segment to send, [1].  Assume
the ERTT = 100.

So you send packet 1, with the retransmission timeout RTO = 2*ERTT = 200.

Assume packet 1 times out, so you retransmit; assuming exponential backoff
is used, RTO = 400.

Perhaps packet 1 times out again, so you transmit; RTO = 800.

Now assume the application requests another segment, [2] to be written.
ERTT for this is still 100, since that isn't updated on packet retransmissions.
Thus RTO = 200 for this segment.  So in this case, [2] could time out
before [1], and since Go-back N transmits segments N onwards, only [2] would
be retransmitted on this time out.

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga18.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Mon, 4 Feb 2002 12:18:55 -0800
Lines: 43
Distribution: su
Message-ID: 
References:  
   
 
 
NNTP-Posting-Host: saga18.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3126

hi mathew,

Given your example, what will be the response of
print_congestion_window_and_estimatedRTT? now there are two estimatedRTTs
since the ACKs of both 1 and 2 are pending?

one more question;-
what happens if I get a corrupted segment - maybe an ACK or a data
segment? Do I just drop it - if it is an ACK, receiver thinks it sent the
ACK, but sender thinks the ACK didn't arrive!!

On 4 Feb 2002, Matthew Jonathan Holliman wrote:

> Consider this example:
>
> The sender window contains just a single new segment to send, [1].  Assume
> the ERTT = 100.
>
> So you send packet 1, with the retransmission timeout RTO = 2*ERTT = 200.
>
> Assume packet 1 times out, so you retransmit; assuming exponential backoff
> is used, RTO = 400.
>
> Perhaps packet 1 times out again, so you transmit; RTO = 800.
>
> Now assume the application requests another segment, [2] to be written.
> ERTT for this is still 100, since that isn't updated on packet retransmissions.
> Thus RTO = 200 for this segment.  So in this case, [2] could time out
> before [1], and since Go-back N transmits segments N onwards, only [2] would
> be retransmitted on this time out.
>
>

thanks
Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine36.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: 4 Feb 2002 21:25:42 GMT
Lines: 45
Distribution: su
Message-ID: 
References:           
NNTP-Posting-Host: elaine36.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3127


>Given your example, what will be the response of
>print_congestion_window_and_estimatedRTT? now there are two estimatedRTTs
>since the ACKs of both 1 and 2 are pending?

Remember that the ERTT is updated only once an ACK is received for a segment
that was not retransmitted.  (So while each of those segments has an implicit
ERTT, which is reflected in its timeout, your code maintains only a single
global RTT estimate).  In the example below, the ERTT is either unchanged,
if [2] times out, or changed on reception of [2]'s ACK.  print_congestion...()
should print out this value, i.e. whatever the current (most up-to-date)
RTT estimate is.


>one more question;-
>what happens if I get a corrupted segment - maybe an ACK or a data
>segment? Do I just drop it - if it is an ACK, receiver thinks it sent the
>ACK, but sender thinks the ACK didn't arrive!!


Since the specification states that STCP doesn't support piggybacking of ACKs
on data packets, you could just drop the packet.  (The other end is a
non-compliant implementation--why waste your time trying to interpret what
it's saying since it could be nonsense?)


>On 4 Feb 2002, Matthew Jonathan Holliman wrote:

>> Consider this example:
>>
>> The sender window contains just a single new segment to send, [1].  Assume
>> the ERTT = 100.
>>
>> So you send packet 1, with the retransmission timeout RTO = 2*ERTT = 200.
>>
>> Assume packet 1 times out, so you retransmit; assuming exponential backoff
>> is used, RTO = 400.
>>
>> Perhaps packet 1 times out again, so you transmit; RTO = 800.
>>
>> Now assume the application requests another segment, [2] to be written.
>> ERTT for this is still 100, since that isn't updated on packet retransmissions.
>> Thus RTO = 200 for this segment.  So in this case, [2] could time out
>> before [1], and since Go-back N transmits segments N onwards, only [2] would
>> be retransmitted on this time out.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: hw2.A test script
Date: Mon, 04 Feb 2002 13:49:32 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, en, fr, de
To: Nick McKeown 
Xref: nntp.stanford.edu su.class.cs244a:3128

The test script now performs tests for milestones A and B. You will 
receive a separate email for each test.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Failed to enforce 5 retransmissions
Date: Mon, 04 Feb 2002 13:57:08 -0800
Lines: 54
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3129

 From the assignment "... If [a segment] has been sent a total of 6 
times (once from the original send plus 5 retransmissions), then the 
network is assumed to have failed and the network terminated. Otherwise 
the segment is retransmitted and the timeout is reset again."

The test script will say NOT OK if you say that the network failed, and 
you did not transmit at least one packet 6 times, or you transmitted one 
packet more than 6 times.

Pablo

BAEHOPIL wrote:

> Hi..
> 
> I think we can see it from a different angle..
> Even if a packet is retransmitted because of Go-Back-N, not because of its
> own timeout,
> it is still true that it is actually retransmitted and its probability of
> reaching the destination actually rises..
> I mean,  can we not think of the count as a "retransmit count" rather than a
> "timeout count"?
> Anyway, I just wish that some authoritative person gives us a definite
> answer...
> 
> And, I also wonder whether the test script is also giving us NOT OK if our
> code gives up before
> making 5 retransmissions..
> 
> Hopil
> 
> "Martin Casado"  wrote in message
> 
> 
>>It works fine for me.  Initially I was failing that test because I was not
>>counting Go-Back-N resends towards the 5 timeouts, but the test script
>>expects that you do.  I do think this is a problem, because I find that
>>if you do count Go-Back-N retransmissions towards the 5, the connection
>>is more likely to time out.
>>
>>                        ~~m
>>
>>
>>>I just wanted to know if other people than Francois and me have this
>>>error and can't understand why.
>>>It is most probably a bug in my program, but if many are in this case,
>>>who knows...
>>>
>>>     Romain
>>>
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: About the print Congestion window.
Date: 4 Feb 2002 22:06:29 GMT
Lines: 35
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: myth8.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3130

By STCP window size, do you mean number of bytes waiting to 
be ACK'ed in the sending window, or number of bytes in the
sending window available to the application?

Thanks,

Derrick


Matthew Jonathan Holliman  wrote:

:>Hi
:>Can someone please clarify what is fid in the print congestion window. Is
:>it the udp socket or the tcp socket. As we have both the information in
:>the ctx structure we can use both but the code says that i can be called
:>from either network.c or mysock.c which is kind of confusing as each of
:>them has either tcp information or udp information. Please clarify. Also
:>is it MIN(20, .......) or MAX(20, .....) for updating the RTT.

: It's sockfd.

: One or two people had asked what the function should do.  To clarify:

: It should print out just the STCP window size for CWND.  This is rather
: meaningless now; it's a vestige of past years' assignments.

: More important is that the function must print out your current estimated
: RTT, which is used by the test script to verify that your code is updating
: the RTT as specified by STCP.

: There are a couple of dummy variables in the function, which you should
: replace appropriately.

: min vs. max was posted earlier and updated on the web page (if you
: think about it, you should also realize the answer).
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: test script results.
Date: Mon, 04 Feb 2002 14:11:52 -0800
Lines: 45
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011226
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3131

The test script has been modified, so that it gives you more information 
now on tests 2.G and 2.H.

Pablo

francois-marie lefevere wrote:

> I got errors for SYN and FIN sequences but the script does not give me
> any trace of what happened.
> What does this mean ?
> (If my program crashes before SYN, I can't understand that everything
> else is right.)
> 
> 
> 
>     * Test 2.G: Failure to send SYN_ACK after a SYN
>   [out of 5]
> 
> [SUMMARY] NOT OK
> [RESULT]  did not send SYN_ACK after SYN
> 
> ---
> 
> 
> ---
>  (INCORRECT!)
> 
> 
> ===============
> 
>     * Test 2.H: Failure to send all packets and make sure they are ACKed
> before the FIN.
>   [out of 5]
> 
> [SUMMARY] NOT OK
> [RESULT]  FIN sent before Acking all data
> 
> ---
> 
> 
> ---
>  (INCORRECT!)
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine29.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: About the print Congestion window.
Date: 4 Feb 2002 22:59:18 GMT
Lines: 8
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine29.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3132


>By STCP window size, do you mean number of bytes waiting to 
>be ACK'ed in the sending window, or number of bytes in the
>sending window available to the application?

Sorry for the confusion--it's the "full" window size, i.e. if no data were
waiting to be acknowledged.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine29.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: How to figure out if peer doesn't exist
Date: 4 Feb 2002 23:09:44 GMT
Lines: 23
Distribution: su
Message-ID: 
References:        
NNTP-Posting-Host: elaine29.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3133


>errno = 146

>Dhawal Kumar

>On 3 Feb 2002, Matthew Jonathan Holliman wrote:

>>
>> >When I do a switch errno after
>> >receiving -1 on read, I get an "Unrecgnised error" - I mean that the errno
>> >doesn't match any of
>>
>> >EBADF, EINTR, EIO, ENOMEM, ENOSR, ENOTSOCK, ESTALE, EWOULDBLOCK
>>
>> >which are the possible values of errno mentioned in documentation.
>>
>>
>> Okay, but what's the value of errno?
>>

I suppose the simple answer would be that it doesn't matter too much, as we
won't be testing this case; otherwise I'm not sure why recvfrom() would
return that error.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: SYN_ACK and FIN
Date: Mon, 4 Feb 2002 17:30:41 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3134


I got error in SYN_ACK and FIN behavior , which I think I have taken
care of in the way the script tells us.

But I test it correctly with my own Tx and Rx. Passive part did send
SYN_ACK. And I do let the sender wait until all data has been acked
before sending a FIN.

What's more, I am not quite sure how this will be relevant with
print_congestion_window_and_estiamtedRTT(). Basically, we just
need to implement this function but not call it in our code.
So it won;t affect the behavior of SYN and FIN.

Appreciate for any comments !

THanks,
-jinhui



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!rxuan
From: Ruozhong Xuan 
Newsgroups: su.class.cs244a
Subject: Re: Failed to enforce 5 retransmissions
Date: Mon, 4 Feb 2002 18:15:51 -0800
Lines: 113
Distribution: su
Message-ID: 
References:  
  
NNTP-Posting-Host: elaine37.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Pablo Molinero Fernandez 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3135

Hi Pablo:
    I am a little confused because I got the following result. For the
packet with seq=199, it does transmit 1 original time and 5 more
retransmissions before it sends the FIN and terminates. Could you tell me
where it could be wrong. Thanks.

--rxuan

===============

    * Test 2.K: Tx fails to enforce 5 Retransmissions
  [out of 2.5]

[SUMMARY] NOT OK
[RESULT]  5 retransmissions not enforced(INCORRECT!)
got up to 27 transmissions

---
Your code produced:
network_send: a SYN packet with seq=198
network_recv: a SYN-ACK packet with seq=198 and ack=199 and
timestamp=3476717994
network_recv: a DATA-ACK packet with ack=211
network_recv: a DATA packet with seq=199, len=536 and timestamp=3476719083
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=735, len=536 and timestamp=3476719084
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1271, len=536 and
timestamp=3476719084
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1807, len=536 and
timestamp=3476719084
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2343, len=536 and
timestamp=3476719085
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2879, len=392 and
timestamp=3476719085
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=199, len=536 and timestamp=3476719592
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=199, len=536 and timestamp=3476720602
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=199, len=536 and timestamp=3476721612
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=199, len=536 and timestamp=3476722622
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=199, len=536 and timestamp=3476723632
network_recv(2): Dropping data packet num 0
network_recv: a FIN packet with seq=3271 and timestamp=3476724642


===============


On Mon, 4 Feb 2002, Pablo Molinero Fernandez wrote:

>  From the assignment "... If [a segment] has been sent a total of 6
> times (once from the original send plus 5 retransmissions), then the
> network is assumed to have failed and the network terminated. Otherwise
> the segment is retransmitted and the timeout is reset again."
>
> The test script will say NOT OK if you say that the network failed, and
> you did not transmit at least one packet 6 times, or you transmitted one
> packet more than 6 times.
>
> Pablo
>
> BAEHOPIL wrote:
>
> > Hi..
> >
> > I think we can see it from a different angle..
> > Even if a packet is retransmitted because of Go-Back-N, not because of its
> > own timeout,
> > it is still true that it is actually retransmitted and its probability of
> > reaching the destination actually rises..
> > I mean,  can we not think of the count as a "retransmit count" rather than a
> > "timeout count"?
> > Anyway, I just wish that some authoritative person gives us a definite
> > answer...
> >
> > And, I also wonder whether the test script is also giving us NOT OK if our
> > code gives up before
> > making 5 retransmissions..
> >
> > Hopil
> >
> > "Martin Casado"  wrote in message
> > 
> >
> >>It works fine for me.  Initially I was failing that test because I was not
> >>counting Go-Back-N resends towards the 5 timeouts, but the test script
> >>expects that you do.  I do think this is a problem, because I find that
> >>if you do count Go-Back-N retransmissions towards the 5, the connection
> >>is more likely to time out.
> >>
> >>                        ~~m
> >>
> >>
> >>>I just wanted to know if other people than Francois and me have this
> >>>error and can't understand why.
> >>>It is most probably a bug in my program, but if many are in this case,
> >>>who knows...
> >>>
> >>>     Romain
> >>>
> >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Darren Lewis" 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Mon, 4 Feb 2002 18:55:34 -0800
Lines: 39
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: darren.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3136

Hi Matthew,
    If N's timeout is left unchanged, then aren't you left with a
conflicting situation? Specifically, once X times out, X and N will both be
sent again. If N is left to timeout at 800ms (and we assume both segments
got sent out again at 600ms when X timed out), then doesn't that mean that
the original N will timeout only 200ms after the retransmission was sent
out? Why wouldn't we want to reset the timeout on segment N to be the same
as the retransmission of segment X, since they're both getting sent out on
retransmissions at the same time? Thanks.

-Darren

"Matthew Jonathan Holliman"  wrote in message

> "BAEHOPIL"  writes:
>
>
> >I have a related question..
>
> >When timeout occurs, we update RTO..
> >Suppose X sent at 100ms, N sent at 300ms, Initial RTO=500ms..
> >Then without any ACK, X will timeout at 600ms and RTO becomes 1000ms
> >Next timeout of X will be at 1600ms,
> >but, how about N?  Will N timeout at 800ms or 1300ms?
>
> >And when & how shall we restore RTO to 2*RTT?
> >Maybe it's up to us to determine?
>
> It's probably more reasonable for N's timeout to be left unchanged
> by the update in RTO.  (I think it's definitely easier to implement,
> especially if you use exponential back-off).
>
> Keep in mind that the RTO/ERTT are very coarse estimates, which include
> some history.  So if the values change a bit following the reception
> of an ACK, it's not necessarily particularly useful or meaningful to
> update following unacked packets' timeouts/ERTTs given this extra
> information.


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Mon, 04 Feb 2002 19:40:24 -0800
Lines: 35
Distribution: su
Message-ID: 
References:       
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3137

Matthew Jonathan Holliman wrote:
> 
> Dhawal Kumar  writes:
> 
Doesnt your example here contradict the assumption that Arun said we
could make in his initial response.... namely that we can assume that
timeouts are in order.

Thanks,
Clayton

> >Does this mean that irrespective of what packet in window times out we
> >resend the whole window? Or is there a situation where we don't - If yes
> >can you give me an example?
> 
> Not necessarily.  (If that were the case, time out handling would be quite
> simple, and you could maintain only a single timeout for the entire window).
> 
> Consider this example:
> 
> The sender window contains just a single new segment to send, [1].  Assume
> the ERTT = 100.
> 
> So you send packet 1, with the retransmission timeout RTO = 2*ERTT = 200.
> 
> Assume packet 1 times out, so you retransmit; assuming exponential backoff
> is used, RTO = 400.
> 
> Perhaps packet 1 times out again, so you transmit; RTO = 800.
> 
> Now assume the application requests another segment, [2] to be written.
> ERTT for this is still 100, since that isn't updated on packet retransmissions.
> Thus RTO = 200 for this segment.  So in this case, [2] could time out
> before [1], and since Go-back N transmits segments N onwards, only [2] would
> be retransmitted on this time out.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Mon, 4 Feb 2002 20:13:59 -0800
Lines: 40
Distribution: su
Message-ID: 
References:  
   
 
  
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3138


Even without backoff, I don't see how segments can be gauranteed to timeout in
order (order by sequence that is) with a variable ERTT.

                        ~~m

> >
> Doesnt your example here contradict the assumption that Arun said we
> could make in his initial response.... namely that we can assume that
> timeouts are in order.
>
> Thanks,
> Clayton
>
> > >Does this mean that irrespective of what packet in window times out we
> > >resend the whole window? Or is there a situation where we don't - If yes
> > >can you give me an example?
> >
> > Not necessarily.  (If that were the case, time out handling would be quite
> > simple, and you could maintain only a single timeout for the entire window).
> >
> > Consider this example:
> >
> > The sender window contains just a single new segment to send, [1].  Assume
> > the ERTT = 100.
> >
> > So you send packet 1, with the retransmission timeout RTO = 2*ERTT = 200.
> >
> > Assume packet 1 times out, so you retransmit; assuming exponential backoff
> > is used, RTO = 400.
> >
> > Perhaps packet 1 times out again, so you transmit; RTO = 800.
> >
> > Now assume the application requests another segment, [2] to be written.
> > ERTT for this is still 100, since that isn't updated on packet retransmissions.
> > Thus RTO = 200 for this segment.  So in this case, [2] could time out
> > before [1], and since Go-back N transmits segments N onwards, only [2] would
> > be retransmitted on this time out.
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Overlapping Packets and Retransmission
Date: Mon, 04 Feb 2002 20:34:25 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.5) Gecko/20011012
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3139

Let's say I transmit a packet with seqno 1 and datalen 100.
Let's say I receive ACK 50.

After the timout for packet 1 triggers, I notice that I'm not fully 
acknowledged, nor fully unack'ed.  I suppose I retry starting from 50. 
Do I update my retry count in this scenario, or do I treat this send 
like a fresh new send?

-Sandeep

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: francois-marie lefevere 
Newsgroups: su.class.cs244a
Subject: Is the test script down ?
Date: Mon, 04 Feb 2002 20:38:59 -0800
Lines: 2
Distribution: su
Message-ID: 
NNTP-Posting-Host: nastia.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1012883941 8772 128.12.187.116 (5 Feb 2002 04:39:01 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.4.9-12 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3140



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Failed to enforce 5 retransmissions
Date: Mon, 4 Feb 2002 20:42:26 -0800
Lines: 89
Distribution: su
Message-ID: 
References:  
  
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3141

On Mon, 4 Feb 2002, Pablo Molinero Fernandez wrote:

>  From the assignment "... If [a segment] has been sent a total of 6
> times (once from the original send plus 5 retransmissions), then the
> network is assumed to have failed and the network terminated. Otherwise
> the segment is retransmitted and the timeout is reset again."
>
> The test script will say NOT OK if you say that the network failed, and
> you did not transmit at least one packet 6 times, or you transmitted one
> packet more than 6 times.
>
> Pablo

Here is a snippet from one of Matthew's earlier emails.

"..... updating the timeout counter probably unfairly penalizes segments
retransmitted as part of Go-back-N--I only incremented it when the segment's
timer itself expired.  (I found it necessary to reset the timer for segments
retransmitted as part of Go-back-N; otherwise for large file transfers I'd
frequently find my connection timing out)."

I've found the same behavior he has, reliability to very large files work
best with the timer bieng reset in in Go-Back-N but not to increment
the number of timeouts.

Assuming you have 6 packets in the timer queue in increasing sequence order
that were sent in the same burst (with very close timeout values), I think
there is a relatively high probability that the 6th packet will be sent
6 times (once for each Go-Back-N and once for itself), before the receiving
side actually reads the package. I'm curious if anyone else has experienced this.

Secondly, one of the tests is:

* Test 2.L: Tx fails to enforce 2k window size
[out of 5]

Which window is this exactly? Is this the 3k sender window?

Thanks :-)

                        ~~m


>
> BAEHOPIL wrote:
>
> > Hi..
> >
> > I think we can see it from a different angle..
> > Even if a packet is retransmitted because of Go-Back-N, not because of its
> > own timeout,
> > it is still true that it is actually retransmitted and its probability of
> > reaching the destination actually rises..
> > I mean,  can we not think of the count as a "retransmit count" rather than a
> > "timeout count"?
> > Anyway, I just wish that some authoritative person gives us a definite
> > answer...
> >
> > And, I also wonder whether the test script is also giving us NOT OK if our
> > code gives up before
> > making 5 retransmissions..
> >
> > Hopil
> >
> > "Martin Casado"  wrote in message
> > 
> >
> >>It works fine for me.  Initially I was failing that test because I was not
> >>counting Go-Back-N resends towards the 5 timeouts, but the test script
> >>expects that you do.  I do think this is a problem, because I find that
> >>if you do count Go-Back-N retransmissions towards the 5, the connection
> >>is more likely to time out.
> >>
> >>                        ~~m
> >>
> >>
> >>>I just wanted to know if other people than Francois and me have this
> >>>error and can't understand why.
> >>>It is most probably a bug in my program, but if many are in this case,
> >>>who knows...
> >>>
> >>>     Romain
> >>>
> >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic20.Stanford.EDU!vasco
From: Vassil Chatalbashev 
Newsgroups: su.class.cs244a
Subject: overlap of window boundary
Date: Mon, 4 Feb 2002 20:55:18 -0800
Lines: 22
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic20.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3142

If we receive a packet which crosses over the window
boundary, are we supposed to acknowledge it, so that
the next byte we expect is the one past the boundary?

If so, when we are transmitting a packet which starts
at byte X and has length L bytes, but receive an ack for
byte B, X < B < X+L, can we simply retransmit the same
packet(from X to X+L) again (i.e. treat it as lost)
or should we just retransmit starting from B. If so,
is that technically not a retransmission (since it is not the
same data we are sending,) so that we don't update
the retransmission count?


Thanks




***********************


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: About the print Congestion window.
Date: Mon, 4 Feb 2002 20:58:46 -0800
Lines: 13
Distribution: su
Message-ID: 
References: 
  
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3143


>
> >By STCP window size, do you mean number of bytes waiting to
> >be ACK'ed in the sending window, or number of bytes in the
> >sending window available to the application?
>
> Sorry for the confusion--it's the "full" window size, i.e. if no data were
> waiting to be acknowledged.

Isn't that a constant 3072? Or am I missing something...

            ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Mon, 4 Feb 2002 22:33:00 -0800
Lines: 68
Distribution: su
Message-ID: 
References:  
   
 
  
 
NNTP-Posting-Host: elaine9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3144

I think it would be much better if the TAs come up with 2-3 examples each
describing a different scenario so that we know what to do. Until a couple
of postings ago I thought that there needs to be just one ERTT but now
there are several and I think that all the question-answers regarding
retransmissions were based on one ERTT idea.

Let me ask the following:
(1) If the global RTT estimate due to ACK of one packet changes, do we
need to change the RTOs of packet already sent but not ACKed. If not then
a packet sent by appl just after the change in RTT may experience a very
different RTO from the one transmitted just before that.

(2) If RTO of a packet with seq num N changes due to timeout, do we adjust
the RTO for packets with seq num > N. If not then whats the point of Go
back N. I think Go back N is used because the sender thinks that all
packets > N are lost and if they all are lost RTOs of all of them must be
changed (whether its difficult or easy to implement is another matter)

Again I say, it would be better if TAs come up with some examples which
everyone can use as reference instead of replying to questions
individually.

Dhawal Kumar

On Mon, 4 Feb 2002, Martin Casado wrote:

>
> Even without backoff, I don't see how segments can be gauranteed to timeout in
> order (order by sequence that is) with a variable ERTT.
>
>                         ~~m
>
> > >
> > Doesnt your example here contradict the assumption that Arun said we
> > could make in his initial response.... namely that we can assume that
> > timeouts are in order.
> >
> > Thanks,
> > Clayton
> >
> > > >Does this mean that irrespective of what packet in window times out we
> > > >resend the whole window? Or is there a situation where we don't - If yes
> > > >can you give me an example?
> > >
> > > Not necessarily.  (If that were the case, time out handling would be quite
> > > simple, and you could maintain only a single timeout for the entire window).
> > >
> > > Consider this example:
> > >
> > > The sender window contains just a single new segment to send, [1].  Assume
> > > the ERTT = 100.
> > >
> > > So you send packet 1, with the retransmission timeout RTO = 2*ERTT = 200.
> > >
> > > Assume packet 1 times out, so you retransmit; assuming exponential backoff
> > > is used, RTO = 400.
> > >
> > > Perhaps packet 1 times out again, so you transmit; RTO = 800.
> > >
> > > Now assume the application requests another segment, [2] to be written.
> > > ERTT for this is still 100, since that isn't updated on packet retransmissions.
> > > Thus RTO = 200 for this segment.  So in this case, [2] could time out
> > > before [1], and since Go-back N transmits segments N onwards, only [2] would
> > > be retransmitted on this time out.
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Failed to enforce 5 retransmissions
Date: Mon, 4 Feb 2002 23:05:41 -0800
Lines: 74
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Martin Casado 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3145

Sorry for the confusion. It should be 3k. 2k window is used in the
assignment last year. Xin



> Secondly, one of the tests is:
> 
> * Test 2.L: Tx fails to enforce 2k window size
> [out of 5]
> 
> Which window is this exactly? Is this the 3k sender window?
> 
> Thanks :-)
> 
>                         ~~m
> 
> 
> >
> > BAEHOPIL wrote:
> >
> > > Hi..
> > >
> > > I think we can see it from a different angle..
> > > Even if a packet is retransmitted because of Go-Back-N, not because of its
> > > own timeout,
> > > it is still true that it is actually retransmitted and its probability of
> > > reaching the destination actually rises..
> > > I mean,  can we not think of the count as a "retransmit count" rather than a
> > > "timeout count"?
> > > Anyway, I just wish that some authoritative person gives us a definite
> > > answer...
> > >
> > > And, I also wonder whether the test script is also giving us NOT OK if our
> > > code gives up before
> > > making 5 retransmissions..
> > >
> > > Hopil
> > >
> > > "Martin Casado"  wrote in message
> > > 
> > >
> > >>It works fine for me.  Initially I was failing that test because I was not
> > >>counting Go-Back-N resends towards the 5 timeouts, but the test script
> > >>expects that you do.  I do think this is a problem, because I find that
> > >>if you do count Go-Back-N retransmissions towards the 5, the connection
> > >>is more likely to time out.
> > >>
> > >>                        ~~m
> > >>
> > >>
> > >>>I just wanted to know if other people than Francois and me have this
> > >>>error and can't understand why.
> > >>>It is most probably a bug in my program, but if many are in this case,
> > >>>who knows...
> > >>>
> > >>>     Romain
> > >>>
> > >
> > >
> > >
> >
> >
> 
> 


-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: test script restarted
Date: Mon, 4 Feb 2002 23:05:34 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3146

Hi,

Today around 2:30pm, we run out of disk space and the test script could 
not compile. We have cleared the files that caused the problem and 
restarted the test script. If you submitted your code after 2:30pm, you 
should be receiving your results shortly.

Pablo

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: About the print Congestion window.
Date: Mon, 4 Feb 2002 23:07:25 -0800
Lines: 27
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Martin Casado 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3147

Yes. It is a constant 3072. Xin

On Mon, 4 Feb 2002, Martin Casado wrote:

> 
> >
> > >By STCP window size, do you mean number of bytes waiting to
> > >be ACK'ed in the sending window, or number of bytes in the
> > >sending window available to the application?
> >
> > Sorry for the confusion--it's the "full" window size, i.e. if no data were
> > waiting to be acknowledged.
> 
> Isn't that a constant 3072? Or am I missing something...
> 
>             ~~m
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Failed to enforce 5 retransmissions
Date: Mon, 4 Feb 2002 23:17:33 -0800
Lines: 78
Distribution: su
Message-ID: 
References:  
  
 
 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3148


The label in the report has been changed to say "* Test 2.L: Tx 
fails to enforce correct window size". The testing was still 
checking that the window was smaller than 3k, so the label did not affect 
the tests.

Pablo

On Mon, 4 Feb 2002, Xin Wang wrote:

> Sorry for the confusion. It should be 3k. 2k window is used in the
> assignment last year. Xin
> 
> 
> 
> > Secondly, one of the tests is:
> > 
> > * Test 2.L: Tx fails to enforce 2k window size
> > [out of 5]
> > 
> > Which window is this exactly? Is this the 3k sender window?
> > 
> > Thanks :-)
> > 
> >                         ~~m
> > 
> > 
> > >
> > > BAEHOPIL wrote:
> > >
> > > > Hi..
> > > >
> > > > I think we can see it from a different angle..
> > > > Even if a packet is retransmitted because of Go-Back-N, not because of its
> > > > own timeout,
> > > > it is still true that it is actually retransmitted and its probability of
> > > > reaching the destination actually rises..
> > > > I mean,  can we not think of the count as a "retransmit count" rather than a
> > > > "timeout count"?
> > > > Anyway, I just wish that some authoritative person gives us a definite
> > > > answer...
> > > >
> > > > And, I also wonder whether the test script is also giving us NOT OK if our
> > > > code gives up before
> > > > making 5 retransmissions..
> > > >
> > > > Hopil
> > > >
> > > > "Martin Casado"  wrote in message
> > > > 
> > > >
> > > >>It works fine for me.  Initially I was failing that test because I was not
> > > >>counting Go-Back-N resends towards the 5 timeouts, but the test script
> > > >>expects that you do.  I do think this is a problem, because I find that
> > > >>if you do count Go-Back-N retransmissions towards the 5, the connection
> > > >>is more likely to time out.
> > > >>
> > > >>                        ~~m
> > > >>
> > > >>
> > > >>>I just wanted to know if other people than Francois and me have this
> > > >>>error and can't understand why.
> > > >>>It is most probably a bug in my program, but if many are in this case,
> > > >>>who knows...
> > > >>>
> > > >>>     Romain
> > > >>>
> > > >
> > > >
> > > >
> > >
> > >
> > 
> > 
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: overlap of window boundary
Date: Mon, 4 Feb 2002 23:31:12 -0800
Lines: 39
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Vassil Chatalbashev 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3149

If the received data cross either end of the current receiver window, the
data should be split into two parts and each part is processed
appropriately. Xin


 On Mon, 4 Feb 2002, Vassil Chatalbashev wrote:

> If we receive a packet which crosses over the window
> boundary, are we supposed to acknowledge it, so that
> the next byte we expect is the one past the boundary?
> 
> If so, when we are transmitting a packet which starts
> at byte X and has length L bytes, but receive an ack for
> byte B, X < B < X+L, can we simply retransmit the same
> packet(from X to X+L) again (i.e. treat it as lost)
> or should we just retransmit starting from B. If so,
> is that technically not a retransmission (since it is not the
> same data we are sending,) so that we don't update
> the retransmission count?
> 
> 
> Thanks
> 
> 
> 
> 
> ***********************
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Tue, 05 Feb 2002 00:35:14 -0800
Lines: 49
Distribution: su
Message-ID: 
References:  
	   
	 
	   
NNTP-Posting-Host: elaine12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3150

This is why I said "assume" :)... can someone please clarify whether we
can assume this in our implementation or not.  Even if you do assume,
you sort have to work around the assumption still (ie checking for
timeouts after you have acked packets).

Thanks
Clayton


Martin Casado wrote:
> 
> Even without backoff, I don't see how segments can be gauranteed to timeout in
> order (order by sequence that is) with a variable ERTT.
> 
>                         ~~m
> 
> > >
> > Doesnt your example here contradict the assumption that Arun said we
> > could make in his initial response.... namely that we can assume that
> > timeouts are in order.
> >
> > Thanks,
> > Clayton
> >
> > > >Does this mean that irrespective of what packet in window times out we
> > > >resend the whole window? Or is there a situation where we don't - If yes
> > > >can you give me an example?
> > >
> > > Not necessarily.  (If that were the case, time out handling would be quite
> > > simple, and you could maintain only a single timeout for the entire window).
> > >
> > > Consider this example:
> > >
> > > The sender window contains just a single new segment to send, [1].  Assume
> > > the ERTT = 100.
> > >
> > > So you send packet 1, with the retransmission timeout RTO = 2*ERTT = 200.
> > >
> > > Assume packet 1 times out, so you retransmit; assuming exponential backoff
> > > is used, RTO = 400.
> > >
> > > Perhaps packet 1 times out again, so you transmit; RTO = 800.
> > >
> > > Now assume the application requests another segment, [2] to be written.
> > > ERTT for this is still 100, since that isn't updated on packet retransmissions.
> > > Thus RTO = 200 for this segment.  So in this case, [2] could time out
> > > before [1], and since Go-back N transmits segments N onwards, only [2] would
> > > be retransmitted on this time out.
> >
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: About the print Congestion window.
Date: Tue, 5 Feb 2002 00:47:40 -0800
Lines: 32
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012898848 10661 128.12.133.48 (5 Feb 2002 08:47:28 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3151


"Matthew Jonathan Holliman"  wrote in message

>
> >Hi
> >Can someone please clarify what is fid in the print congestion window. Is
> >it the udp socket or the tcp socket. As we have both the information in
> >the ctx structure we can use both but the code says that i can be called
> >from either network.c or mysock.c which is kind of confusing as each of
> >them has either tcp information or udp information. Please clarify. Also
> >is it MIN(20, .......) or MAX(20, .....) for updating the RTT.
>
> It's sockfd.

The mysock.c code shows print_congestion... being called from within
mywrite(), where the only available socket value is actually the parent's
end of the local data pipe (data_sd[0] in transport_init).  (Someone back a
ways in the parent process' call stack has sockfd--they used it to call
transport_init--but it's not passed to mywrite.) The network.c code has it
being called from network_send, where the socket it has is the UDP socket
(comm_sd in transport_init, sockfd in the child's context struct).

Should we follow the code and have the function interpret fid as the local
data socket when called from the parent process, but the peer socket when
called from the child? Or assume it's never  called from  the parent and
just treat it as sockfd?

Thanks,

Jonathan


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jing Jiang" 
Newsgroups: su.class.cs244a
Subject: Go-back-N protocol and grouping data
Date: Tue, 5 Feb 2002 04:27:42 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: jjiang.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2314.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
Xref: nntp.stanford.edu su.class.cs244a:3152

when packet with seq N times out, we retransmit data starting from N. do we
send data in segments of the original sizes, or do we group data into new
segments with maximum segment size? i assume we should group data into
bigger segments. but if so, how do we keep the timeout count of each
segment? e.g., we first send packet A of size 300. say A times out twice
(and thus is retransmitted twice). then we receive data from appl and send
packet B of size 300. then B times out (before A because A has a bigger
RTO), and so we retransmit B. when A times out the fourth time, we need to
retransmit both packet A and B, which are 600 bytes of data. do we first
send a segment with size 536 and timeout count 4, and then send the rest of
the data in another segment with timeout count 2? please clarify. thanks!


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic21.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Sending Fin after the Max Retransmission
Date: Tue, 5 Feb 2002 08:39:35 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To:    
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3153

Hi,
I have one question regarding the MAX timeout. When the Maximum timeout
happens are we suppose to send the FIN packet after that or just close the
connection. Also after we send the FIN packet for this case (in case we
do) then are we suppose to wait for the finack for this fin packet. Now if
we are suppose to wait for the fin ack for this fin packet then are we
suppose to retransmit the FIN packet in case this fin times out.
Thanks
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Network timeouts in *reliable* mode
Date: 5 Feb 2002 20:02:15 GMT
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth8.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3154

I'm trying to transfer a 56MB file in *reliable* mode, and I
always hit the 5 retransmission limit (at different times: 1MB,
7MB, ...). Is this expected behavior, even in reliable mode, or
am I doing something wrong with my timeouts?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Xiaoyan Cheng 
Newsgroups: su.class.cs244a
Subject: timeout
Date: Tue, 05 Feb 2002 12:53:20 -0800
Lines: 13
Distribution: su
Message-ID: 
Reply-To: 
NNTP-Posting-Host: pup3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1012942403 15384 171.64.66.153 (5 Feb 2002 20:53:23 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.78 [en] (X11; U; Linux 2.4.9-13 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3155

Do we need to remember the timeout value for EACH packet sent out or
just a global timeout applied to every packet? in Karn's Algorithm, when
we do exponential backoff, do we twice the timeout value just for that
retransmitted packet or everybody else?

I think the latter(global timeout) is more reasonable, but I just got
confused by the earlier messages from the newsgroup. Could anyone
clarify this?

thanks.



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: timeout
Date: 5 Feb 2002 21:14:24 GMT
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth8.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3156

I used to think we only double the timeout of the retransmitted packet,
but it almost doesn't make sense. Say you have 6 packets, A through F:

A   B   C   D   E   F
100 100 100 100 100 100  <-- timeouts

At 100ms, A times out, and all packets are retransmitted. If only A is
doubled, then we have:

A   B   C   D   E   F
200 100 100 100 100 100

Then B will timeout and everything retransmitted, then C, and so forth. By
the time we get to F, it will have been transmitted 6 times, and we would
conclude that there was a network error. But we haven't waited the 200ms
for A.

So I agree with Xiaoyan that it's more reasonable to double everyone's
timeout, but what is the *correct* way to do it?



Xiaoyan Cheng  wrote:
: Do we need to remember the timeout value for EACH packet sent out or
: just a global timeout applied to every packet? in Karn's Algorithm, when
: we do exponential backoff, do we twice the timeout value just for that
: retransmitted packet or everybody else?

: I think the latter(global timeout) is more reasonable, but I just got
: confused by the earlier messages from the newsgroup. Could anyone
: clarify this?

: thanks.



.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: retransmission and timeout
Date: Tue, 5 Feb 2002 16:00:40 -0800
Lines: 45
Distribution: su
Message-ID: 
References:          
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Dhawal Kumar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3157

Hi,

The goal of the assignment is to provide a reliable delivery mechanism.
You also need to make your implementation robust, e.g., your STCP
implementation should be able to transmit large files without connection
timeout. As for how to handle retransmission and timeout, there could be
multiple ways. As long as your implementation can provide a robust
reliable transportation, any reasonable strategy is acceptable.

I list my suggestions as below:

1. One ERTT is fine. 

2. If ERTT changes due to ACK of one packet, we don't need to
change the RTOs of packet already sent but not ACKed. This is because
RTO/ERTT are coarse estimates, considering this extra information is not
very necessary.

3. If RTO of a packet with seq num N changes due to timeout, we need to
modify the RTO for packets with seq num > N. I think this is a better way
because otherwise, the connection could be timeout when transfering large
files.

4. The timout counter should be updated for go-back-N, which is consistent
with the requirement of the test script. Some students said that if the
counter was updated, then the connection will timeout for large files. I
think this problem could be solved by resetting RTOs as said in 3. This
suggestion is different from Matthew's. As mentioned in Matthew's previous
post, he will check it with Pablo and get back to the newsgroup.

Again, there could be multiple ways to implement the retransmission and
timeout. Any reasonable way is acceptable.

Best,

Xin

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: SYN_ACK and FIN
Date: Tue, 5 Feb 2002 16:14:18 -0800
Lines: 43
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Jinhui Pan 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3158

How does your passive end exit? After the passive end receives FIN, the
passive end should send back FIN-ACK immediately and exit.

print_congestion_window_and_estiamtedRTT() just prints some
information. It should be irrelevent to your problem.

Best,

Xin


On Mon, 4 Feb 2002, Jinhui Pan wrote:

> 
> I got error in SYN_ACK and FIN behavior , which I think I have taken
> care of in the way the script tells us.
> 
> But I test it correctly with my own Tx and Rx. Passive part did send
> SYN_ACK. And I do let the sender wait until all data has been acked
> before sending a FIN.
> 
> What's more, I am not quite sure how this will be relevant with
> print_congestion_window_and_estiamtedRTT(). Basically, we just
> need to implement this function but not call it in our code.
> So it won;t affect the behavior of SYN and FIN.
> 
> Appreciate for any comments !
> 
> THanks,
> -jinhui
> 
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: Re: SYN_ACK and FIN
Date: Tue, 5 Feb 2002 16:24:35 -0800
Lines: 62
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Xin Wang 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3159



Yes, I handle it in this way. And I got FIN_ACK when I test with my
own Tx and Rx. But I do not know how the test script test this case
and why it failed with test.

It is very difficult to debug on this because the test script does
not give any information on how it fails and we can not reproduce the
bug.

Could the test scipt give more information that we can get some hint
on what is the bug and where does it fail? That will be of great help!

Could anyone give some suggestion on how to debug and reproduce the
the bug that the test script may produce.

Thank a lot!!!


-jinhui




On Tue, 5 Feb 2002, Xin Wang wrote:

> How does your passive end exit? After the passive end receives FIN, the
> passive end should send back FIN-ACK immediately and exit.
>
> print_congestion_window_and_estiamtedRTT() just prints some
> information. It should be irrelevent to your problem.
>
> Best,
>
> Xin
>
>
> On Mon, 4 Feb 2002, Jinhui Pan wrote:
>
> >
> > I got error in SYN_ACK and FIN behavior , which I think I have taken
> > care of in the way the script tells us.
> >
> > But I test it correctly with my own Tx and Rx. Passive part did send
> > SYN_ACK. And I do let the sender wait until all data has been acked
> > before sending a FIN.
> >
> > What's more, I am not quite sure how this will be relevant with
> > print_congestion_window_and_estiamtedRTT(). Basically, we just
> > need to implement this function but not call it in our code.
> > So it won;t affect the behavior of SYN and FIN.
> >
> > Appreciate for any comments !
> >
> > THanks,
> > -jinhui
> >
> >
> >
> >
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Error in test script?
Date: 6 Feb 2002 02:25:25 GMT
Lines: 117
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth8.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3160

Test 2.K says 5 retransmissions were not enforced, getting up to 15
transmissions. It looks to me like all the data packets were
transmitted exactly 6 times (1 + 5 retransmissions).

Is there a problem with the test script?


- Derrick


Excerpt from test results:

===============

    * Test 2.K: Tx fails to enforce 5 Retransmissions
  [out of 2.5]


[SUMMARY] NOT OK
[RESULT]  5 retransmissions not enforced(INCORRECT!)
got up to 15 transmissions


---
Your code produced:
network_send: a SYN packet with seq=197
network_recv: a SYN-ACK packet with seq=197 and ack=198 and timestamp=3644374300
network_recv: a DATA-ACK packet with ack=210
network_recv: a DATA packet with seq=198, len=20 and timestamp=3644374704
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=218, len=536 and timestamp=3644374704
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=754, len=536 and timestamp=3644374705
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644374705
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644374705
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644374705
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644374705
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=198, len=20 and timestamp=3644375212
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=218, len=536 and timestamp=3644375212
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=754, len=536 and timestamp=3644375212
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644375212
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644375212
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644375213
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644375213
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=198, len=20 and timestamp=3644376222
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=218, len=536 and timestamp=3644376222
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=754, len=536 and timestamp=3644376222
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644376222
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644376222
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644376223
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644376223
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=198, len=20 and timestamp=3644378232
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=218, len=536 and timestamp=3644378232
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=754, len=536 and timestamp=3644378232
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644378232
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644378233
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644378233
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644378233
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=198, len=20 and timestamp=3644382242
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=218, len=536 and timestamp=3644382242
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=754, len=536 and timestamp=3644382242
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644382243
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644382243
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644382243
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644382243
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=198, len=20 and timestamp=3644390252
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=218, len=536 and timestamp=3644390253
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=754, len=536 and timestamp=3644390253
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644390253
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644390253
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644390253
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644390254
network_recv(2): Dropping data packet num 0



===============

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Sending Fin after the Max Retransmission
Date: Tue, 5 Feb 2002 18:41:54 -0800
Lines: 34
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
cc:    
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3161


Hi Shankar,

If u meant MAX timeout is 2 x ERTT,

the active end should resend the FIN 5 more times before
it finally gives up and exits anyway.

For the passive end that receives FIN, it sends FIN ACK
and exits immediately. So if this FIN ACK was lost, the
active end will not get another FIN ACK. The active end
is retransmitting FIN hoping that the earlier FINs were
lost.

Please feel free to correct :-)

Thanks, Arvind


On Tue, 5 Feb 2002, Shankar Agarwal wrote:

> Hi,
> I have one question regarding the MAX timeout. When the Maximum timeout
> happens are we suppose to send the FIN packet after that or just close the
> connection. Also after we send the FIN packet for this case (in case we
> do) then are we suppose to wait for the finack for this fin packet. Now if
> we are suppose to wait for the fin ack for this fin packet then are we
> suppose to retransmit the FIN packet in case this fin times out.
> Thanks
> Shankar
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: SYN_ACK and FIN
Date: Tue, 5 Feb 2002 18:56:16 -0800
Lines: 75
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012964177 19026 128.12.133.48 (6 Feb 2002 02:56:17 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3162

Print_congestion... should indeed be irrelevant, however it's easy to write
code in that function that leads to segfaults. In this situation, the test
script will pass you on everything but D G H and I, and give you no output
with which to debug. Make sure you've tested print_congestion... on your
own, calling from both network and mysock.

Jonathan


"Jinhui Pan"  wrote in message

>
>
> Yes, I handle it in this way. And I got FIN_ACK when I test with my
> own Tx and Rx. But I do not know how the test script test this case
> and why it failed with test.
>
> It is very difficult to debug on this because the test script does
> not give any information on how it fails and we can not reproduce the
> bug.
>
> Could the test scipt give more information that we can get some hint
> on what is the bug and where does it fail? That will be of great help!
>
> Could anyone give some suggestion on how to debug and reproduce the
> the bug that the test script may produce.
>
> Thank a lot!!!
>
>
> -jinhui
>
>
>
>
> On Tue, 5 Feb 2002, Xin Wang wrote:
>
> > How does your passive end exit? After the passive end receives FIN, the
> > passive end should send back FIN-ACK immediately and exit.
> >
> > print_congestion_window_and_estiamtedRTT() just prints some
> > information. It should be irrelevent to your problem.
> >
> > Best,
> >
> > Xin
> >
> >
> > On Mon, 4 Feb 2002, Jinhui Pan wrote:
> >
> > >
> > > I got error in SYN_ACK and FIN behavior , which I think I have taken
> > > care of in the way the script tells us.
> > >
> > > But I test it correctly with my own Tx and Rx. Passive part did send
> > > SYN_ACK. And I do let the sender wait until all data has been acked
> > > before sending a FIN.
> > >
> > > What's more, I am not quite sure how this will be relevant with
> > > print_congestion_window_and_estiamtedRTT(). Basically, we just
> > > need to implement this function but not call it in our code.
> > > So it won;t affect the behavior of SYN and FIN.
> > >
> > > Appreciate for any comments !
> > >
> > > THanks,
> > > -jinhui
> > >
> > >
> > >
> > >
> >
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: A Request to the TAs
Date: Tue, 5 Feb 2002 19:10:18 -0800
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1012965018 19178 128.12.133.48 (6 Feb 2002 03:10:18 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3163

I think everyone involved would have it easier if the TAs could get
together, decide on and post a unified statement of what the parameters are
for timeouts and retransmissions in this project. It would help to know
three things:

1) What elements of the implementation we must do in a particular manner.
(E.g., use Karn-Partridge, keep timeouts on a per-packet basis, resend only
>X rather than all unACKed)
2) What elements of the implementation we are free to design ourselves.
(E.g., transitioning between timeouts determined by backoff and timeouts
determined by RTT samples.)
3) What behavior characteristics are required. (E.g., transferring large
files without killing the connection because of excessive resends.)

There have been conflicting answers from different TAs, both on the
newsgroup and offline, and a unified clarification of the constraints of the
assignment would be tremendously helpful, and would probably lower the
traffic and confusion on this newsgroup!

My $0.02,

Jonathan


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Go-back-N protocol and grouping data
Date: Tue, 5 Feb 2002 19:13:15 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Jing Jiang 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3164

You don't need to group data into new segments with maximum segment
size. Xin


On Tue, 5 Feb 2002, Jing Jiang wrote:

> when packet with seq N times out, we retransmit data starting from N. do we
> send data in segments of the original sizes, or do we group data into new
> segments with maximum segment size? i assume we should group data into
> bigger segments. but if so, how do we keep the timeout count of each
> segment? e.g., we first send packet A of size 300. say A times out twice
> (and thus is retransmitted twice). then we receive data from appl and send
> packet B of size 300. then B times out (before A because A has a bigger
> RTO), and so we retransmit B. when A times out the fourth time, we need to
> retransmit both packet A and B, which are 600 bytes of data. do we first
> send a segment with size 536 and timeout count 4, and then send the rest of
> the data in another segment with timeout count 2? please clarify. thanks!
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Sending Fin after the Max Retransmission
Date: Tue, 5 Feb 2002 19:25:13 -0800
Lines: 51
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Arvind Bhat 
cc: Shankar Agarwal   
    
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3165

Please refer to "Network Termination" section in the handout and Q6 of
FAQ about how to handle FIN, FIN ACK and connection termination. Xin


On Tue, 5 Feb 2002, Arvind Bhat wrote:

> 
> Hi Shankar,
> 
> If u meant MAX timeout is 2 x ERTT,
> 
> the active end should resend the FIN 5 more times before
> it finally gives up and exits anyway.
> 
> For the passive end that receives FIN, it sends FIN ACK
> and exits immediately. So if this FIN ACK was lost, the
> active end will not get another FIN ACK. The active end
> is retransmitting FIN hoping that the earlier FINs were
> lost.
> 
> Please feel free to correct :-)
> 
> Thanks, Arvind
> 
> 
> On Tue, 5 Feb 2002, Shankar Agarwal wrote:
> 
> > Hi,
> > I have one question regarding the MAX timeout. When the Maximum timeout
> > happens are we suppose to send the FIN packet after that or just close the
> > connection. Also after we send the FIN packet for this case (in case we
> > do) then are we suppose to wait for the finack for this fin packet. Now if
> > we are suppose to wait for the fin ack for this fin packet then are we
> > suppose to retransmit the FIN packet in case this fin times out.
> > Thanks
> > Shankar
> >
> >
> >
> 
> 


-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: SYN retransmit
Date: 6 Feb 2002 04:42:03 GMT
Lines: 21
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine12.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3166

Matthew Jonathan Holliman  wrote:

>>This must be reallystupid, but do we retransmit SYN if we don't get a
>>SYN_ACK in
>>500ms (2 * initial estimatedRTT) ?

> Yep, SYN and FIN should be retransmitted if an ACK isn't received in time,
> just like data segments.

	What do we do if the SYN_ACK gets lost? It states clearly on the 
handout under "Network Initiation" that when the passive end sends a 
SYN_ACK, the connection is established from the point of view of the 
passive side. But when the SYN_ACK is lost, the active side could still 
potentially send SYN packets. Do we have to look out for this case as 
well?

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Overlapping Packets and Retransmission
Date: Tue, 5 Feb 2002 23:01:22 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Sandeep Tamhankar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3167

I think you should update the retransmission count only for the part which
are retransmitted. Xin


On Mon, 4 Feb 2002, Sandeep Tamhankar wrote:

> Let's say I transmit a packet with seqno 1 and datalen 100.
> Let's say I receive ACK 50.
> 
> After the timout for packet 1 triggers, I notice that I'm not fully 
> acknowledged, nor fully unack'ed.  I suppose I retry starting from 50. 
> Do I update my retry count in this scenario, or do I treat this send 
> like a fresh new send?
> 
> -Sandeep
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Network timeouts in *reliable* mode
Date: Tue, 5 Feb 2002 23:03:32 -0800
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Derrick Wen-Shiuan Tong 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3168

You are doing something wrong with your timeout. Xin


On 5 Feb 2002, Derrick Wen-Shiuan Tong wrote:

> I'm trying to transfer a 56MB file in *reliable* mode, and I
> always hit the 5 retransmission limit (at different times: 1MB,
> 7MB, ...). Is this expected behavior, even in reliable mode, or
> am I doing something wrong with my timeouts?
> 
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Error in test script?
Date: Tue, 5 Feb 2002 23:10:35 -0800
Lines: 135
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Derrick Wen-Shiuan Tong 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3169

Please refer to ealier posts about the retransmission. The retransmission
should be exactly 6 times, including both retransmissions caused
by its own timeout and those caused by go-back-N. Xin



On 6 Feb 2002, Derrick Wen-Shiuan Tong wrote:

> Test 2.K says 5 retransmissions were not enforced, getting up to 15
> transmissions. It looks to me like all the data packets were
> transmitted exactly 6 times (1 + 5 retransmissions).
> 
> Is there a problem with the test script?
> 
> 
> - Derrick
> 
> 
> Excerpt from test results:
> 
> ===============
> 
>     * Test 2.K: Tx fails to enforce 5 Retransmissions
>   [out of 2.5]
> 
> 
> [SUMMARY] NOT OK
> [RESULT]  5 retransmissions not enforced(INCORRECT!)
> got up to 15 transmissions
> 
> 
> ---
> Your code produced:
> network_send: a SYN packet with seq=197
> network_recv: a SYN-ACK packet with seq=197 and ack=198 and timestamp=3644374300
> network_recv: a DATA-ACK packet with ack=210
> network_recv: a DATA packet with seq=198, len=20 and timestamp=3644374704
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=218, len=536 and timestamp=3644374704
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=754, len=536 and timestamp=3644374705
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644374705
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644374705
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644374705
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644374705
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=198, len=20 and timestamp=3644375212
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=218, len=536 and timestamp=3644375212
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=754, len=536 and timestamp=3644375212
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644375212
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644375212
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644375213
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644375213
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=198, len=20 and timestamp=3644376222
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=218, len=536 and timestamp=3644376222
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=754, len=536 and timestamp=3644376222
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644376222
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644376222
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644376223
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644376223
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=198, len=20 and timestamp=3644378232
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=218, len=536 and timestamp=3644378232
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=754, len=536 and timestamp=3644378232
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644378232
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644378233
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644378233
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644378233
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=198, len=20 and timestamp=3644382242
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=218, len=536 and timestamp=3644382242
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=754, len=536 and timestamp=3644382242
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644382243
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644382243
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644382243
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644382243
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=198, len=20 and timestamp=3644390252
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=218, len=536 and timestamp=3644390253
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=754, len=536 and timestamp=3644390253
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1290, len=536 and timestamp=3644390253
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=1826, len=536 and timestamp=3644390253
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2362, len=536 and timestamp=3644390253
> network_recv(2): Dropping data packet num 0
> network_recv: a DATA packet with seq=2898, len=372 and timestamp=3644390254
> network_recv(2): Dropping data packet num 0
> 
> 
> 
> ===============
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Question Regarding ACKs - a simple one
Date: Tue, 5 Feb 2002 23:22:18 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3170

Suppose a sender has 3 packets A, B, C to transmit and the receiver gets
B, C and then A. When the receiver gets B and C it would send an ACK
corresponding to sequence number of A since it expects A. Now when
receiver gets A should it send 3 ACKs or 1 ACK corresponding to seq_num of last
byte of C + 1.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tim Chao 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Wed, 6 Feb 2002 08:39:05 +0000 (UTC)
Lines: 9
Distribution: su
Message-ID: 
References:         
NNTP-Posting-Host: slimtc.stanford.edu
X-Trace: news.Stanford.EDU 1012984745 21423 128.12.65.49 (6 Feb 2002 08:39:05 GMT)
X-Complaints-To: 
User-Agent: Xnews/4.11.09
Xref: nntp.stanford.edu su.class.cs244a:3171

Sorry to post another question about retransmissions, but I need 
clarification on this point.

I send three segments 10, 20, and 30. Let's say the segment with number 10 
times out and it's time to Go-Back-N. So now I'm going to retransmit 
everything starting from 10. But segments 20 and 30 have timeouts 
associated with them - so when I retransmit the data associated with them, 
am I going to invalidate those timeouts? I think that's the most logical 
thing to do...
.

Path: shelby.stanford.edu!nntp.stanford.edu!fable9.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: Re: SYN retransmit
Date: Wed, 6 Feb 2002 01:46:06 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: fable9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Victor Tung 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3172

Yup, it can get lost and if that is the case:
 passive side is in "ESTABLISHED" state.
 active side is in "WAIT" state.

So, if a packet(with SYN) after this point appears on the passive side
again, it needs to be acknowledged. The passive side should try its best
to get SYN/ACK through to the active side to pull it up to "ESTABLISHED"
state as well.

The rules for retransmission as far as active side is concerned, I guess,
should remain the same.

Hope this helps,
-gaurav


>
> 	What do we do if the SYN_ACK gets lost? It states clearly on the
> handout under "Network Initiation" that when the passive end sends a
> SYN_ACK, the connection is established from the point of view of the
> passive side. But when the SYN_ACK is lost, the active side could still
> potentially send SYN packets. Do we have to look out for this case as
> well?
>
> --
> Victor Tung			| 
> ------------------------------------------------------------------------------
> Metaphysics is the science of proving what we don't understand.
>                                 -- Josh Billings (Henry Wheeler Shaw)
>

*************************************************************************
Gaurav Garg                          * Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             *       Escondido Village, Stanford
Department of Electrical Engineering *       CA-94305
Stanford University                  * Ph:650-498-1208
*************************************************************************

-Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Enforce 5 Retransmissions
Date: Wed, 6 Feb 2002 02:02:36 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3173

What should happen when 5 retransmissions have occured (6 transmission
total)? Should the program just quit, or should it try to send a FIN packet
and wait for a FIN_ACK back? The assignment sheet is not very clear on that.

Also, do we need to keep track of how many duplicate ACKs we have sent? In
other words, do we need to enforce 5 retransmission on ACKs packet? SYN_ACK
packet? FIN_ACK packet? I know we definitely need to enforce 5
retransmission data packets, SYN packet and FIN packet. Imagine that the
receiver never receives a FIN, how is it suposed to terminate because the
sender has already terminated because of the 5 retransmission rules.

-Henry


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Question Regarding ACKs - a simple one
Date: Wed, 6 Feb 2002 02:38:42 -0800
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Dhawal Kumar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3174


just one ACK with th_ack = C_last + 1

from ACK Packets section, bullet 5.

Thanks, Arvind

On Tue, 5 Feb 2002, Dhawal Kumar wrote:

> Suppose a sender has 3 packets A, B, C to transmit and the receiver gets
> B, C and then A. When the receiver gets B and C it would send an ACK
> corresponding to sequence number of A since it expects A. Now when
> receiver gets A should it send 3 ACKs or 1 ACK corresponding to seq_num of last
> byte of C + 1.
>
> Dhawal Kumar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Enforce 5 Retransmissions
Date: Wed, 6 Feb 2002 02:53:02 -0800
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Henry Fu 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3175


Hi Henry,

On Wed, 6 Feb 2002, Henry Fu wrote:

> What should happen when 5 retransmissions have occured (6 transmission
> total)? Should the program just quit, or should it try to send a FIN packet
> and wait for a FIN_ACK back? The assignment sheet is not very clear on that.

FIN is to be sent only after we are sure that all data has been
transferred. Since that is not the case, I think it should quit
with an error message saying timed out or something.

>
> Also, do we need to keep track of how many duplicate ACKs we have sent? In
> other words, do we need to enforce 5 retransmission on ACKs packet? SYN_ACK
> packet? FIN_ACK packet? I know we definitely need to enforce 5
> retransmission data packets, SYN packet and FIN packet. Imagine that the
> receiver never receives a FIN, how is it suposed to terminate because the
> sender has already terminated because of the 5 retransmission rules.
>

ACK/SYN_ACK/FIN_ACKs should not be retransmitted in the sense of data
packets. But are supposed to be sent as a response, like say if 5 SYN
packets show up, we should send 5 SYN_ACKs back :-)

I do not think the doc says how to handle the Receiver not receving
FIN case.... Probably a timed out exit when no activity is observed
could be a solution.

> -Henry
>
>

Thanks, Arvind

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic12.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: About the retransmission.
Date: Wed, 6 Feb 2002 08:56:51 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic12.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3176

Hi,
Has the issue been resolved about the retransmissions. What i remember
from past mails that TA's will soon post a mail about it. But i have not
received any mail about that. Moreover i am not sure why but i am seeing
83 retranmission statement in the test script which is not possible at
all. Infact the data output from the script never shows 83 packets
received from the sender.
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Are the test scripts ok ?
Date: Wed, 6 Feb 2002 09:37:32 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3177


I submitted tests last night and still not seeing the results.

Thanks, Arvind

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Are the test scripts ok ?
Date: Wed, 06 Feb 2002 10:52:17 -0800
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.8) Gecko/20020205
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3178

Unfortunately the kerberos tickets expired last night at midnight, and 
they were renewed, and the server restarted, today at 10:20am.

Pablo

Arvind Bhat wrote:
> I submitted tests last night and still not seeing the results.
> 
> Thanks, Arvind
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: purify warning
Date: Wed, 6 Feb 2002 11:53:23 -0800
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3179


Hi,

Is there a way to get around this ?

Thanks, Arvind


-------------------------------------------------------------------------
Purify slave: Warning: The version of /usr/lib/libc.so.1
        on the current machine does not match the version on the machine
        where the application was instrumented.
        This is not a supported configuration.
        The program may crash or produce incorrect results.


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: About the print Congestion window.
Date: Wed, 6 Feb 2002 13:44:53 -0800
Lines: 50
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Jonathan Keljo 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3180


wondering if you got a clarification on this....

How can we get the ERTT from parent process ?

I thought it was something know only to the child
which acted as the transport layer.

What am i missing here :-(


On Tue, 5 Feb 2002, Jonathan Keljo wrote:

>
> "Matthew Jonathan Holliman"  wrote in message
> 
> >
> > >Hi
> > >Can someone please clarify what is fid in the print congestion window. Is
> > >it the udp socket or the tcp socket. As we have both the information in
> > >the ctx structure we can use both but the code says that i can be called
> > >from either network.c or mysock.c which is kind of confusing as each of
> > >them has either tcp information or udp information. Please clarify. Also
> > >is it MIN(20, .......) or MAX(20, .....) for updating the RTT.
> >
> > It's sockfd.
>
> The mysock.c code shows print_congestion... being called from within
> mywrite(), where the only available socket value is actually the parent's
> end of the local data pipe (data_sd[0] in transport_init).  (Someone back a
> ways in the parent process' call stack has sockfd--they used it to call
> transport_init--but it's not passed to mywrite.) The network.c code has it
> being called from network_send, where the socket it has is the UDP socket
> (comm_sd in transport_init, sockfd in the child's context struct).
>
> Should we follow the code and have the function interpret fid as the local
> data socket when called from the parent process, but the peer socket when
> called from the child? Or assume it's never  called from  the parent and
> just treat it as sockfd?
>
> Thanks,
>
> Jonathan
>
>
>




.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: purify warning
Date: Wed, 6 Feb 2002 20:07:58 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013054879 1074 128.12.133.48 (7 Feb 2002 04:07:59 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3181

Remake all.purify on the current machine.

Jonathan

"Arvind Bhat"  wrote in message

>
> Hi,
>
> Is there a way to get around this ?
>
> Thanks, Arvind
>
>
> -------------------------------------------------------------------------
> Purify slave: Warning: The version of /usr/lib/libc.so.1
>         on the current machine does not match the version on the machine
>         where the application was instrumented.
>         This is not a supported configuration.
>         The program may crash or produce incorrect results.
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!cardinal4.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: About Sliding Window
Date: Wed, 6 Feb 2002 21:27:18 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: cardinal4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3182

the assignment description says: "Note that received data may cross either
end of the current receiver window"
But when I consider it, I get confused. How can the received data cross
the either end of the current receiver?
Since "The first byte of all windows is always the last acknowledged byte
of data", so the lower boundary of the sender window is equal to or
lower(when it hasn't received the ACK) than the lower boundary of the
receiver window. Both window have fixed size 3072, so the upper
boundary of the sender window is equal to or lower than the upper boundary
of the receiver window, too.
And we "Do not send any data outside of your sender window. "
So, my question is how can a packet cross the upper boundary of the
receiver window be sent?
I can't figure out how can a packet cross the lower boundary of the
receiver window, either?
So, can somebody give me an example for both cases?
Thank you very much.


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: timeout on local socket
Date: Wed, 06 Feb 2002 22:04:11 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic21.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3183

Isn't it true that our timeouts should only be used to wait on network
socket? I've been using select() to wait on both the local socket and
the network socket, but now that I've introduced timeouts, this is a
bug. The reason is that if any data is left on the data socket the
select() will never time out because there is stuff waiting for me on
the data socket.

My solution to this would be to do separate select()'s on the local
socket and the network socket, doing an immediate poll on the local
socket and using the timeout for the network one. But if I'm in the
control loop, this means I lose responsiveness to any new data on the
local socket. (I only check the local socket once, then wait on the
network, and then loop again)

Any suggestions on what should be done?
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: purify warning
Date: Wed, 6 Feb 2002 22:24:26 -0800
Lines: 39
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3184


 I'm having the same problem as Arvind's..
 But deleting every *.o file and remaking all.purify doesn't work for me..
 And probably as the result of the warning, I'm getting no purify info
about child process... I tried this on epic and saga, seperately..

 Any suggestions?

 Hopil

"Jonathan Keljo"  wrote in message

> Remake all.purify on the current machine.
>
> Jonathan
>
> "Arvind Bhat"  wrote in message
> 
> >
> > Hi,
> >
> > Is there a way to get around this ?
> >
> > Thanks, Arvind
> >
> >
>
> -------------------------------------------------------------------------
> > Purify slave: Warning: The version of /usr/lib/libc.so.1
> >         on the current machine does not match the version on the machine
> >         where the application was instrumented.
> >         This is not a supported configuration.
> >         The program may crash or produce incorrect results.
> >
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga5.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Is the test script working?
Date: Wed, 6 Feb 2002 22:42:32 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3185

Hi,

I am waiting for around 15-20 minutes. Is the scripts working?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: print congestion window and estimated RTT
Date: Wed, 6 Feb 2002 23:20:52 -0800
Lines: 129
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_018A_01C1AF64.EB6C2C90"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3186

This is a multi-part message in MIME format.

------=_NextPart_000_018A_01C1AF64.EB6C2C90
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


I want to repeat Matthew's question...
Making eRTT available to the parent looks like an unimportant but tricky =
task...
I want to avoid it if PERMISSIBLE.. Please clarify...
And, if we have to, shall we interpret fid as local data socket as =
mysock.c code says?

Thanks,
Hopil

"Matthew Jonathan Holliman"  wrote in message

>
> >Hi
> >Can someone please clarify what is fid in the print congestion =
window. Is
> >it the udp socket or the tcp socket. As we have both the information =
in
> >the ctx structure we can use both but the code says that i can be =
called
> >from either network.c or mysock.c which is kind of confusing as each =
of
> >them has either tcp information or udp information. Please clarify. =
Also
> >is it MIN(20, .......) or MAX(20, .....) for updating the RTT.
>
> It's sockfd.

The mysock.c code shows print_congestion... being called from within
mywrite(), where the only available socket value is actually the =
parent's
end of the local data pipe (data_sd[0] in transport_init).  (Someone =
back a
ways in the parent process' call stack has sockfd--they used it to call
transport_init--but it's not passed to mywrite.) The network.c code has =
it
being called from network_send, where the socket it has is the UDP =
socket
(comm_sd in transport_init, sockfd in the child's context struct).

Should we follow the code and have the function interpret fid as the =
local
data socket when called from the parent process, but the peer socket =
when
called from the child? Or assume it's never  called from  the parent and
just treat it as sockfd?

Thanks,

Jonathan


------=_NextPart_000_018A_01C1AF64.EB6C2C90
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>I want to repeat Matthew's =
question...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Making eRTT available to the =
parent looks like=20
an unimportant </FONT><FONT face=3DArial size=3D2>but tricky=20
task...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I want to avoid it if PERMISSIBLE.. =
Please=20
clarify...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>And, if we have to, shall we interpret =
fid as local=20
data socket as mysock.c code says?</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Thanks,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
<DIV><FONT face=3DArial size=3D2><BR>"Matthew Jonathan Holliman" <<A=20
 =
wrote in=20
message<BR><A=20

tanford.EDU</A>...<BR>><BR>>=20
>Hi<BR>> >Can someone please clarify what is fid in the print=20
congestion window. Is<BR>> >it the udp socket or the tcp socket. =
As we=20
have both the information in<BR>> >the ctx structure we can use =
both but=20
the code says that i can be called<BR>> >from either network.c or =
mysock.c=20
which is kind of confusing as each of<BR>> >them has either tcp=20
information or udp information. Please clarify. Also<BR>> >is it =
MIN(20,=20
........) or MAX(20, .....) for updating the RTT.<BR>><BR>> It's=20
sockfd.<BR><BR>The mysock.c code shows print_congestion... being called =
from=20
within<BR>mywrite(), where the only available socket value is actually =
the=20
parent's<BR>end of the local data pipe (data_sd[0] in =
transport_init). =20
(Someone back a<BR>ways in the parent process' call stack has =
sockfd--they used=20
it to call<BR>transport_init--but it's not passed to mywrite.) The =
network.c=20
code has it<BR>being called from network_send, where the socket it has =
is the=20
UDP socket<BR>(comm_sd in transport_init, sockfd in the child's context=20
struct).<BR><BR>Should we follow the code and have the function =
interpret fid as=20
the local<BR>data socket when called from the parent process, but the =
peer=20
socket when<BR>called from the child? Or assume it's never  called=20
from  the parent and<BR>just treat it as=20
sockfd?<BR><BR>Thanks,<BR><BR>Jonathan<BR></FONT></DIV></BODY></HTML>

------=_NextPart_000_018A_01C1AF64.EB6C2C90--

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Test 2.K
Date: Wed, 6 Feb 2002 23:45:53 -0800
Lines: 54
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3187

Hi,

I am keep failing the Test 2.k (Test 2.K: Tx fails to enforce 5
Retransmissions). I revised my code and I am sure that I send only 5
retransmission. I want just someone to tell me what is the meaning of the
error, specifically, what is the mean of the "got up to 30 transmissions",
and also what I can learn from the network_send/netwrok_recv log that I
got. (Can I know from it how many retransmission may code is actually
doing? and if the problem that I send more or less than 5
retransmission).

I like to mention that it will be more beneficial if these tests can
handle us back our debugging log, so we can figure out where is the
problem.

Thanks,

Moh'd

* Test 2.K: Tx fails to enforce 5 Retransmissions
[out of 2.5]

[SUMMARY] NOT OK
[RESULT]  5 retransmissions not enforced(INCORRECT!)
got up to 30 transmissions

---
Your code produced:
network_send: a SYN packet with seq=198
network_recv: a SYN-ACK packet with seq=198 and ack=199 and
timestamp=3749515829
network_send: a DATA packet with seq=199, len=12
network_send: a DATA packet with seq=199, len=12
network_recv: a DATA-ACK packet with ack=211
network_recv: a DATA packet with seq=199, len=536 and timestamp=3749516391
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=735, len=536 and timestamp=3749516391
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1271, len=536 and
timestamp=3749516391
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1807, len=536 and
timestamp=3749516391
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2343, len=536 and
timestamp=3749516392
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2879, len=392 and
timestamp=3749516392
network_recv(2): Dropping data packet num 0




.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Function name mismatch for print_congestion_...
Date: Thu, 7 Feb 2002 00:14:23 -0800
Lines: 53
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_01EA_01C1AF6C.6542A600"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3188

This is a multi-part message in MIME format.

------=_NextPart_000_01EA_01C1AF6C.6542A600
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

 mysock.c and network.c is calling
     print_congestion_window_and_timeout()

while transport.c has=20
    print_congestion_window_and_estimatedRTT()

A trivial typo, but please make sure which one we should stick to,
so that it doesn't cause any problem for testing..

Hopil


------=_NextPart_000_01EA_01C1AF6C.6542A600
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2> mysock.c and network.c is=20
calling</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>   =20
 print_congestion_window_and_timeout()</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>while transport.c has </FONT></DIV>
<DIV><FONT face=3DArial size=3D2>   =20
print_congestion_window_and_estimatedRTT()</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>A trivial typo, but please make sure =
which one we=20
should stick to,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>so that it doesn't cause any =
problem for=20
testing..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_01EA_01C1AF6C.6542A600--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: timeout on local socket
Date: 7 Feb 2002 08:51:53 GMT
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3189

Tim,

If you cannot read any more data because you're sending window is
full, don't set the bit for the local data socket in your fd_set.
That way, select will only block on the network socket, and 
you will get your timeouts.

- Derrick


Timothy Tay Chao  wrote:
: Isn't it true that our timeouts should only be used to wait on network
: socket? I've been using select() to wait on both the local socket and
: the network socket, but now that I've introduced timeouts, this is a
: bug. The reason is that if any data is left on the data socket the
: select() will never time out because there is stuff waiting for me on
: the data socket.

: My solution to this would be to do separate select()'s on the local
: socket and the network socket, doing an immediate poll on the local
: socket and using the timeout for the network one. But if I'm in the
: control loop, this means I lose responsiveness to any new data on the
: local socket. (I only check the local socket once, then wait on the
: network, and then loop again)

: Any suggestions on what should be done?
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Test 1.C problems
Date: 7 Feb 2002 09:07:55 GMT
Lines: 62
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth8.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3190


Anyone else have a problem with Test 1.C?
It's their Tx, student Rx. I assume the diff means
that the expected output has those files/checksums,
and my output has nothing ... I suspect that my Receiver
is just blocking and getting killed. Is there something
weird about the way their Transmitter works? I pass all
of the other tests ...

Thanks,

Derrick



===============


    * Test 1.C: Receiver fails to work in reliable mode with our Transmitter
        [out of 10]


[SUMMARY] NOT OK
[RESULT]  
===========================
===(T1.Ca) Byte count differences found (INCORRECT!)
Our Tx against student's Rx (reliable mode)


---


---
Difference between expected result and your result:
1a2,3
>    1812   10049  299328 TESTDATA.1
>     131     501    5098 TESTDATA.2


---


===========================
===(T1.Cb) Checksum differences found (INCORRECT!)
Our Tx against student's Rx (reliable mode)


---


---
Difference between expected result and your result:
1a2,3
> 4265638025 299328 TESTDATA.1
> 2181702833 5098 TESTDATA.2


---



===============
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: estimatedRTT quickly declines to 20
Date: Thu, 07 Feb 2002 01:13:41 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3191

Although my tests look good, I'm watching my estimatedRTT quickly
decline to 20ms when transferring large files...

Is this what others are observing as well?

Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: estimatedRTT quickly declines to 20
Date: 7 Feb 2002 09:32:47 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3192


I'm getting that as well, but only when I turn off all of
my debugging output. I get sample RTT's of mostly 1 or 2 ms.

Should we be adjusting our estimated RTT's by taking into account
how often we timeout, or just use the successful ACK's of packets
which were sent once?


Luis Robles  wrote:
: Although my tests look good, I'm watching my estimatedRTT quickly
: decline to 20ms when transferring large files...

: Is this what others are observing as well?

: Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: client.c reading 0 bytes and exiting
Date: Thu, 7 Feb 2002 01:51:14 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3193

Although I am never closing local_data_sd and never writing 0 bytes to
local_data_sd, client exits while its in the middle of a file transfer
because it reads 0 from its socket sd. Any suggestions as to how can I
figure out whats going on local_data_sd <--> client - I mean is there a
way I can check the traffic going between the two ends of the pipe.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: retransmit exponential timeouts.
Date: Thu, 7 Feb 2002 01:52:14 -0800
Lines: 29
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3194


Hi,

from the previous posts, i could understand that we should be
using exponential backoff for retransmissions. If say t1, t2,
t3, t4 and t5 are the timeout values for each retransmission,
are we going to use (A) or (B) ?

A) t1 = 2 * ERTT,
   t2 = 4 * ERTT,
   t3 = 8 * ERTT,
   t4 = 16 * ERTT,
   t5 = 32 * ERTT

OR

B) t1 = 2 * ERTT,
   t2 = 2 * t1,
   t3 = 4 * t1,
   t4 = 8 * t1,
   t5 = 16 * t1

Or is the timeout value calculated with some other method than
these ?

Please let me know...

Thanks, Arvind

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: print_congestion_window_and_estimatedRTT
Date: Thu, 07 Feb 2002 02:01:47 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3195

I'm also getting some strange behavior correlated
w/ my print_congestion_window function,

I'm simply printing the constant window size of 3072
and the estimatedRTT from the context,

But leaving that small part of code uncommented
causes me to fail the 5 retransmission test...

If I comment it out, bingo, everything
seems dandy according to the tests...

As stated before, the fid passed in refers to
the sockfd... 

But what should be done w/ the fid?

Can't we get the current context and its
associated estimatedRTT from the curr_ctx global?

Cheers,
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: retransmit exponential timeouts.
Date: 7 Feb 2002 10:06:45 GMT
Lines: 39
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3196

I believe the Karn/Partridge algorithm specifies:

t1 = 2 * ERTT
t2 = 2 * t1
t3 = 2 * t2
t4 = 2 * t3
t5 = 2 * t4


Arvind Bhat  wrote:

: Hi,

: from the previous posts, i could understand that we should be
: using exponential backoff for retransmissions. If say t1, t2,
: t3, t4 and t5 are the timeout values for each retransmission,
: are we going to use (A) or (B) ?

: A) t1 = 2 * ERTT,
:    t2 = 4 * ERTT,
:    t3 = 8 * ERTT,
:    t4 = 16 * ERTT,
:    t5 = 32 * ERTT

: OR

: B) t1 = 2 * ERTT,
:    t2 = 2 * t1,
:    t3 = 4 * t1,
:    t4 = 8 * t1,
:    t5 = 16 * t1

: Or is the timeout value calculated with some other method than
: these ?

: Please let me know...

: Thanks, Arvind

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: print_congestion_window_and_estimatedRTT
Date: 7 Feb 2002 10:11:27 GMT
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga18.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3197

curr_ctx is set in the child process, so when
print_congestion_and_estimatedRTT is called from
the parent process, dereferencing curr_ctx will
probably cause a seg fault. This may explain the
test failures. As Jonathan suggested, you should
test the print_congestion_and_estimatedRTT from
both mysock.c and network.c, where they are
commented out.



Luis Robles  wrote:
: I'm also getting some strange behavior correlated
: w/ my print_congestion_window function,

: I'm simply printing the constant window size of 3072
: and the estimatedRTT from the context,

: But leaving that small part of code uncommented
: causes me to fail the 5 retransmission test...

: If I comment it out, bingo, everything
: seems dandy according to the tests...

: As stated before, the fid passed in refers to
: the sockfd... 

: But what should be done w/ the fid?

: Can't we get the current context and its
: associated estimatedRTT from the curr_ctx global?

: Cheers,
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sutin Chen 
Newsgroups: su.class.cs244a
Subject: hw2.a testscript
Date: Thu, 07 Feb 2002 03:18:36 -0800
Lines: 3
Distribution: su
Message-ID: 
NNTP-Posting-Host: talon1.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3198

hi - could you remove the limit for submissions for hw2a?  i'd like to
check my hw2b with it, but can't because it says i've exceeded my max
submissions
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: SYN_ACK and FIN
Date: Thu, 07 Feb 2002 08:07:50 -0800
Lines: 97
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3199

Thanks a lot, Jonathan.  I'd been racking my brain about what I could be 
doing wrong with SYN/SYN-ACK behavior all day yesterday.  I actually 
managed to fix a few potential bugs, but they didn't help solve this 
particular issue.

It never occurred to me that I could've done something silly in 
print_congestion_window.  This really saved me.

-Sandeep

Jonathan Keljo wrote:

> Print_congestion... should indeed be irrelevant, however it's easy to write
> code in that function that leads to segfaults. In this situation, the test
> script will pass you on everything but D G H and I, and give you no output
> with which to debug. Make sure you've tested print_congestion... on your
> own, calling from both network and mysock.
> 
> Jonathan
> 
> 
> "Jinhui Pan"  wrote in message
> 
> 
>>
>>Yes, I handle it in this way. And I got FIN_ACK when I test with my
>>own Tx and Rx. But I do not know how the test script test this case
>>and why it failed with test.
>>
>>It is very difficult to debug on this because the test script does
>>not give any information on how it fails and we can not reproduce the
>>bug.
>>
>>Could the test scipt give more information that we can get some hint
>>on what is the bug and where does it fail? That will be of great help!
>>
>>Could anyone give some suggestion on how to debug and reproduce the
>>the bug that the test script may produce.
>>
>>Thank a lot!!!
>>
>>
>>-jinhui
>>
>>
>>
>>
>>On Tue, 5 Feb 2002, Xin Wang wrote:
>>
>>
>>>How does your passive end exit? After the passive end receives FIN, the
>>>passive end should send back FIN-ACK immediately and exit.
>>>
>>>print_congestion_window_and_estiamtedRTT() just prints some
>>>information. It should be irrelevent to your problem.
>>>
>>>Best,
>>>
>>>Xin
>>>
>>>
>>>On Mon, 4 Feb 2002, Jinhui Pan wrote:
>>>
>>>
>>>>I got error in SYN_ACK and FIN behavior , which I think I have taken
>>>>care of in the way the script tells us.
>>>>
>>>>But I test it correctly with my own Tx and Rx. Passive part did send
>>>>SYN_ACK. And I do let the sender wait until all data has been acked
>>>>before sending a FIN.
>>>>
>>>>What's more, I am not quite sure how this will be relevant with
>>>>print_congestion_window_and_estiamtedRTT(). Basically, we just
>>>>need to implement this function but not call it in our code.
>>>>So it won;t affect the behavior of SYN and FIN.
>>>>
>>>>Appreciate for any comments !
>>>>
>>>>THanks,
>>>>-jinhui
>>>>
>>>>
>>>>
>>>>
>>>>
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Question Regarding ACKs - a simple one
Date: Thu, 7 Feb 2002 08:14:26 -0800
Lines: 32
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3200

Dhawal,

1 ACK is correct. You don't send acks for B and C (as you havent
received A yet) but once you get A you ack until the byte after C.
From the specs:

--
Data is acknowledged by setting the ACK bit in the flags field in the
packet header. If this bit is set, then the ack_seq field contains the
sequence number of the next byte of data the receiver expects (i.e. one
past the last byte of data already received). This is true no matter what
data is being acknowledged
--

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Tue, 5 Feb 2002, Dhawal Kumar wrote:

> Suppose a sender has 3 packets A, B, C to transmit and the receiver gets
> B, C and then A. When the receiver gets B and C it would send an ACK
> corresponding to sequence number of A since it expects A. Now when
> receiver gets A should it send 3 ACKs or 1 ACK corresponding to seq_num of last
> byte of C + 1.
> 
> Dhawal Kumar
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Retransmission.
Date: Thu, 7 Feb 2002 08:29:13 -0800
Lines: 30
Distribution: su
Message-ID: 
References:          
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3201

Tim,

basically yes. When you do Go-Back-N you should (packages
here are the same thing you call segments):
- retransmit all unack'd packages in the queue
- reset the time out counters for all packets (actually you 
  don't really need one time out counter for each packet,
  a global one will do the job).
- increase retransmit count for all packets, if it hits 7
  give up and go home.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 6 Feb 2002, Tim Chao wrote:

> Sorry to post another question about retransmissions, but I need 
> clarification on this point.
> 
> I send three segments 10, 20, and 30. Let's say the segment with number 10 
> times out and it's time to Go-Back-N. So now I'm going to retransmit 
> everything starting from 10. But segments 20 and 30 have timeouts 
> associated with them - so when I retransmit the data associated with them, 
> am I going to invalidate those timeouts? I think that's the most logical 
> thing to do...
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: SampleRTT Calculations
Date: 7 Feb 2002 16:31:37 GMT
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine17.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3202

I have a question about Sample RTT calculations. I'm in need of a bit of a 
clarification on the scenario when one ACK is received for multiple 
packets. Suppose the sender sends packets with sequence number 100, 200, 
and 300; then the receiver gets 100 first, and sends an ACK For 101, but 
it then receives 300, and then 200 causing it to only send an ACK of 301. 
Upon receiving this ACK, do we count this as 2 samples in our sample RTT 
calculations (for packets 200 AND 300), or just one sample. I guess it 
makes sense to count it as 2 since as the sender, if we counted only one, 
I can see even more ambiguity as far as deciding which packet to calculate 
that 1 sample RTT with. 

Anyway, thanks in advance for anyone helping me clarify this.

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: SYN retransmit
Date: Thu, 7 Feb 2002 08:33:34 -0800
Lines: 57
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3203

Victor,

that's correct, if you are passive and in ESTABLISHED state
you do have to deal witht he possibility that your SYN/ACK did 
not o thorugh and you have to retransmit it if you get another
SYN.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 6 Feb 2002, Gaurav Garg wrote:

> Yup, it can get lost and if that is the case:
>  passive side is in "ESTABLISHED" state.
>  active side is in "WAIT" state.
> 
> So, if a packet(with SYN) after this point appears on the passive side
> again, it needs to be acknowledged. The passive side should try its best
> to get SYN/ACK through to the active side to pull it up to "ESTABLISHED"
> state as well.
> 
> The rules for retransmission as far as active side is concerned, I guess,
> should remain the same.
> 
> Hope this helps,
> -gaurav
> 
> 
> >
> > 	What do we do if the SYN_ACK gets lost? It states clearly on the
> > handout under "Network Initiation" that when the passive end sends a
> > SYN_ACK, the connection is established from the point of view of the
> > passive side. But when the SYN_ACK is lost, the active side could still
> > potentially send SYN packets. Do we have to look out for this case as
> > well?
> >
> > --
> > Victor Tung			| 
> > ------------------------------------------------------------------------------
> > Metaphysics is the science of proving what we don't understand.
> >                                 -- Josh Billings (Henry Wheeler Shaw)
> >
> 
> *************************************************************************
> Gaurav Garg                          * Contact Info: 20A Comstock Circle
> 1st Yr, Graduate Student             *       Escondido Village, Stanford
> Department of Electrical Engineering *       CA-94305
> Stanford University                  * Ph:650-498-1208
> *************************************************************************
> 
> -Never ruin an apology with an excuse
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Enforce 5 Retransmissions
Date: Thu, 7 Feb 2002 09:14:54 -0800
Lines: 56
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3204

Arvind, Henry,

Arvind is correct on all counts. FIN is only send if you
actively tear down a connection. If you time out you just fail
quietly and so will the other side. SYN-ACKs and FIN-ACKs are
never retransmitted but if you get a new SYN or FIN you should
send another reply. If all your FIN's get lost you simply time
out.

  Guido


---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 6 Feb 2002, Arvind Bhat wrote:

> 
> Hi Henry,
> 
> On Wed, 6 Feb 2002, Henry Fu wrote:
> 
> > What should happen when 5 retransmissions have occured (6 transmission
> > total)? Should the program just quit, or should it try to send a FIN packet
> > and wait for a FIN_ACK back? The assignment sheet is not very clear on that.
> 
> FIN is to be sent only after we are sure that all data has been
> transferred. Since that is not the case, I think it should quit
> with an error message saying timed out or something.
> 
> >
> > Also, do we need to keep track of how many duplicate ACKs we have sent? In
> > other words, do we need to enforce 5 retransmission on ACKs packet? SYN_ACK
> > packet? FIN_ACK packet? I know we definitely need to enforce 5
> > retransmission data packets, SYN packet and FIN packet. Imagine that the
> > receiver never receives a FIN, how is it suposed to terminate because the
> > sender has already terminated because of the 5 retransmission rules.
> >
> 
> ACK/SYN_ACK/FIN_ACKs should not be retransmitted in the sense of data
> packets. But are supposed to be sent as a response, like say if 5 SYN
> packets show up, we should send 5 SYN_ACKs back :-)
> 
> I do not think the doc says how to handle the Receiver not receving
> FIN case.... Probably a timed out exit when no activity is observed
> could be a solution.
> 
> > -Henry
> >
> >
> 
> Thanks, Arvind
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: About the retransmission.
Date: Thu, 7 Feb 2002 09:16:31 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3205

Shankar,

did the follow-up to the last thread answer your open
questions? If not, re-post!

  Guido


---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 6 Feb 2002, Shankar Agarwal wrote:

> Hi,
> Has the issue been resolved about the retransmissions. What i remember
> from past mails that TA's will soon post a mail about it. But i have not
> received any mail about that. Moreover i am not sure why but i am seeing
> 83 retranmission statement in the test script which is not possible at
> all. Infact the data output from the script never shows 83 packets
> received from the sender.
> Shankar
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: purify warning
Date: Thu, 7 Feb 2002 09:46:39 -0800
Lines: 59
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3206

Hopil,

no idea what this could be. I guess you are sure that there 
is no old file lurking around somewhere (also make sure to
delete the executable and any .??* files).

We can have a look at this in office hours (6pm in Sweet Hall
as the deadline is tomorrow I might start them a little
earlier).

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Wed, 6 Feb 2002, BAEHOPIL wrote:

> 
>  I'm having the same problem as Arvind's..
>  But deleting every *.o file and remaking all.purify doesn't work for me..
>  And probably as the result of the warning, I'm getting no purify info
> about child process... I tried this on epic and saga, seperately..
> 
>  Any suggestions?
> 
>  Hopil
> 
> "Jonathan Keljo"  wrote in message
> 
> > Remake all.purify on the current machine.
> >
> > Jonathan
> >
> > "Arvind Bhat"  wrote in message
> > 
> > >
> > > Hi,
> > >
> > > Is there a way to get around this ?
> > >
> > > Thanks, Arvind
> > >
> > >
> >
> > -------------------------------------------------------------------------
> > > Purify slave: Warning: The version of /usr/lib/libc.so.1
> > >         on the current machine does not match the version on the machine
> > >         where the application was instrumented.
> > >         This is not a supported configuration.
> > >         The program may crash or produce incorrect results.
> > >
> > >
> >
> >
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine27.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: About the retransmission.
Date: Thu, 7 Feb 2002 10:05:32 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Guido Appenzeller 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3207

Hi Guido,
I am still not clear about it. I have one basic question about this
retransmission bussiness. When we say Maximum of 6 retranmisssion we mean
6 transimission in total or 6 transmissions for a particular packet. I
think its for a particular packet. But then i am not sending any one
particular packet for more then 6 times then how am i getting 83
retransmissions.
Thanks
Shankar

 On Thu, 7 Feb 2002, Guido Appenzeller wrote:

> Shankar,
>
> did the follow-up to the last thread answer your open
> questions? If not, re-post!
>
>   Guido
>
>
> ---------------------------------------------------------------
> Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
>  - office: 650 7253545  cell: 650 7042781
>
> On Wed, 6 Feb 2002, Shankar Agarwal wrote:
>
> > Hi,
> > Has the issue been resolved about the retransmissions. What i remember
> > from past mails that TA's will soon post a mail about it. But i have not
> > received any mail about that. Moreover i am not sure why but i am seeing
> > 83 retranmission statement in the test script which is not possible at
> > all. Infact the data output from the script never shows 83 packets
> > received from the sender.
> > Shankar
> >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: is the test script  down again?
Date: Thu, 7 Feb 2002 10:21:28 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3208

hi,
i sent a test about 3 hours ago... haven't heard anything yet?
thanks
ashmi

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: hw2.a testscript
Date: Thu, 7 Feb 2002 10:32:09 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013106729 6444 128.12.133.48 (7 Feb 2002 18:32:09 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3209

Read back a bit...the 2B script now runs both.

Jonathan

"Sutin Chen"  wrote in message

> hi - could you remove the limit for submissions for hw2a?  i'd like to
> check my hw2b with it, but can't because it says i've exceeded my max
> submissions


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: print_congestion_window_and_estimatedRTT
Date: Thu, 7 Feb 2002 11:45:02 -0800
Lines: 49
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3210

Hi everyone,

Using curr_ctx is definitely not a good idea. We might have an
issue here with how print_congestion_window is specified. I
am looking into this and will post again today.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On 7 Feb 2002, Derrick Wen-Shiuan Tong wrote:

> curr_ctx is set in the child process, so when
> print_congestion_and_estimatedRTT is called from
> the parent process, dereferencing curr_ctx will
> probably cause a seg fault. This may explain the
> test failures. As Jonathan suggested, you should
> test the print_congestion_and_estimatedRTT from
> both mysock.c and network.c, where they are
> commented out.
> 
> 
> 
> Luis Robles  wrote:
> : I'm also getting some strange behavior correlated
> : w/ my print_congestion_window function,
> 
> : I'm simply printing the constant window size of 3072
> : and the estimatedRTT from the context,
> 
> : But leaving that small part of code uncommented
> : causes me to fail the 5 retransmission test...
> 
> : If I comment it out, bingo, everything
> : seems dandy according to the tests...
> 
> : As stated before, the fid passed in refers to
> : the sockfd... 
> 
> : But what should be done w/ the fid?
> 
> : Can't we get the current context and its
> : associated estimatedRTT from the curr_ctx global?
> 
> : Cheers,
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Running on Linux
Date: Thu, 07 Feb 2002 13:45:33 -0800
Lines: 37
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3211

FYI:

I decided to try running the transport layer on Linux as a final test 
before submitting.  I needed to make the following changes in order to 
make it build properly:

Makefile.linux: add -D_BSD_SOURCE -DBSD to CFLAGS
client.c: There's an #ifdef SOLARIS near the top of the file; add the 
following #else clause:

#else
#include <unistd.h>

Things should build after that.  You might get some warnings like the 
following:

warning: extra tokens at end of #endif directive

This is because the header files end if stuff like

#endif foo

when it should be

#endif /* foo */

My home gcc complained about it, but firebird didn't.  Go figure.

-Sandeep

-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: is the test script  down again?
Date: Thu, 07 Feb 2002 13:45:57 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.8) Gecko/20020205
X-Accept-Language: es, en-us, en, fr, de
To: Ashmi 
Xref: nntp.stanford.edu su.class.cs244a:3212

The script is working.

Pablo

Ashmi wrote:
> hi,
> i sent a test about 3 hours ago... haven't heard anything yet?
> thanks
> ashmi
> 
> ###############################################################################
> 
> 				ASHMI CHOKSHI
> Graduate Student				    141L Escondido Village
> Dept. of Computer Science			    Stanford University
> Stanford University				    Stanford Ca 94305
> Stanford Ca 94305				    (650)498-1103
> 
> ###############################################################################
> 
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Yichen Xie" 
Newsgroups: su.class.cs244a
Subject: Re: print_congestion_window_and_estimatedRTT
Date: Thu, 7 Feb 2002 13:47:19 -0800
Lines: 59
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: dn800cb605.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3213

But how could the parent process know anything about the current RTT? The
only way I can think of involves some messy IPC (another socket, or shared
memory) between the parent and the child. Storing RTT in the global context
array won't work, 'coz the child only updates its own copy of the global
context.

"Guido Appenzeller"  wrote in message

> Hi everyone,
>
> Using curr_ctx is definitely not a good idea. We might have an
> issue here with how print_congestion_window is specified. I
> am looking into this and will post again today.
>
>   Guido
>
> ---------------------------------------------------------------
> Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
>  - office: 650 7253545  cell: 650 7042781
>
> On 7 Feb 2002, Derrick Wen-Shiuan Tong wrote:
>
> > curr_ctx is set in the child process, so when
> > print_congestion_and_estimatedRTT is called from
> > the parent process, dereferencing curr_ctx will
> > probably cause a seg fault. This may explain the
> > test failures. As Jonathan suggested, you should
> > test the print_congestion_and_estimatedRTT from
> > both mysock.c and network.c, where they are
> > commented out.
> >
> >
> >
> > Luis Robles  wrote:
> > : I'm also getting some strange behavior correlated
> > : w/ my print_congestion_window function,
> >
> > : I'm simply printing the constant window size of 3072
> > : and the estimatedRTT from the context,
> >
> > : But leaving that small part of code uncommented
> > : causes me to fail the 5 retransmission test...
> >
> > : If I comment it out, bingo, everything
> > : seems dandy according to the tests...
> >
> > : As stated before, the fid passed in refers to
> > : the sockfd...
> >
> > : But what should be done w/ the fid?
> >
> > : Can't we get the current context and its
> > : associated estimatedRTT from the curr_ctx global?
> >
> > : Cheers,
> >
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Guido Appenzeller" 
Newsgroups: su.class.cs244a
Subject: Important Print_congestion_window clarifications...
Date: Thu, 7 Feb 2002 13:51:53 -0800
Lines: 25
Distribution: su
Message-ID: 
NNTP-Posting-Host: dnab4046fa.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3214

In response to the various postings, here are some important clarifications
on Print_congestion_window (p_c_w):

Contrary to the comment in the source, p_c_w() will only be called from the
*client* part of transport.c, never from the server part. This means it can
never be called directly from mysocket.c (ignore the comment in mywrite()).
There is no need to transition any data from the server part to the client
part. Accessing anything defined in the server part of transport.c can crash
your program as after the fork they have different data spaces.

p_c_w() can be callled from network.c. The socket descriptor will always
point to the sockfd from the network layer (the UDP socket). Some people are
wondering what to do with the socket descriptor. Depending on how you
implement the system you don't actually need to use it at all, as each
Socket will fork off a new thread (which has a seperate data space). A
global variable is an option here.

As some of you have pointed out printing out the window size actually is not
very exciting but it is easy  to implement :-)

If there are any more questions please follow up here in the newsgroup.

  Guido


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: test script discrepancies
Date: Thu, 7 Feb 2002 14:17:04 -0800
Lines: 45
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
cc: Matthew Jonathan Holliman 
Xref: nntp.stanford.edu su.class.cs244a:3215


Hi,

I am seeing that for the same source, not even a single line changed,
the test scripts gave two different results on different submissions.

It will be great if someone can explain these failures. Am I the only
one seeing the discrepancies in test script runs ?

Thanks, Arvind



First Time
===============

    * Test 2.K: Tx fails to enforce 5 Retransmissions
  [out of 2.5]

[SUMMARY] NOT OK
[RESULT]  5 retransmissions not enforced(INCORRECT!)
got up to 18 transmissions

---
Your code produced:
network_send: a SYN packet with seq=229
network_recv: a SYN-ACK packet with seq=178 and ack=230 and timestamp=3795526038
network_recv: a DATA-ACK packet with ack=242
network_recv: a DATA packet with seq=179, len=20 and timestamp=3795526443



Second Time
===============

    * Test 2.K: Tx fails to enforce 5 Retransmissions
  [out of 2.5]

[SUMMARY] OK
[RESULT]  5 retransmissions enforced (CORRECT!)





.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: test script discrepancies
Date: 7 Feb 2002 23:45:16 GMT
Lines: 52
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth8.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3216

It happens for me too, on that same test.
Funny thing is, the output never shows that any packet
is transmitted more than 6 times, although [RESULT]
says that it exceeds it.


Arvind Bhat  wrote:

: Hi,

: I am seeing that for the same source, not even a single line changed,
: the test scripts gave two different results on different submissions.

: It will be great if someone can explain these failures. Am I the only
: one seeing the discrepancies in test script runs ?

: Thanks, Arvind



: First Time
: ===============

:     * Test 2.K: Tx fails to enforce 5 Retransmissions
:   [out of 2.5]

: [SUMMARY] NOT OK
: [RESULT]  5 retransmissions not enforced(INCORRECT!)
: got up to 18 transmissions

: ---
: Your code produced:
: network_send: a SYN packet with seq=229
: network_recv: a SYN-ACK packet with seq=178 and ack=230 and timestamp=3795526038
: network_recv: a DATA-ACK packet with ack=242
: network_recv: a DATA packet with seq=179, len=20 and timestamp=3795526443



: Second Time
: ===============

:     * Test 2.K: Tx fails to enforce 5 Retransmissions
:   [out of 2.5]

: [SUMMARY] OK
: [RESULT]  5 retransmissions enforced (CORRECT!)





.

Path: shelby.stanford.edu!nntp.stanford.edu!cardinal5.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: Can I use signal to implement timeout?
Date: Thu, 7 Feb 2002 16:22:56 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: cardinal5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3217

I use signal(SIGALRM,...) to implement timeout. But when a signal comes,
if the client is waiting for read something, the read call will return and
read 0 byte, and then the client will exist.
How to figure out this problem? Or it is a wrong way to implement the
timeout by using signal?
Thanks.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Shyam Maddali 
Newsgroups: su.class.cs244a
Subject: FIN , FIN_ACK questions
Date: Thu, 07 Feb 2002 17:35:42 -0800
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic23.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3218

Hi,

    when i run the test_scipt on my code , it gives   errors on the
FIN/FIN_ACK test cases.
Last time i posted a question , Arun replied and i believe i implemented
that behaviour.
In My Local Test Environment i see that FIN and FIN_ACk are exchanged
correcly.

But i still have the problem with this incorrect FIN/FIN_ACK behaviour.

Also i have a question on what to do in this case
--
 client sends the ACK to the last packet and gets a Zero Byte read on
the local data
socket.  Client thinks that all data is ACKed and Sends FIN Packet.
  But if the ACK to the Last packet is lost, to the  server it seems
that client is 
misbehaving. isnt it ?

any clarifications are appreciated.

~
Lakshman Shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Jones 
Newsgroups: su.class.cs244a
Subject: Re: FIN , FIN_ACK questions
Date: Thu, 07 Feb 2002 17:39:13 -0800
Lines: 38
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: harrypotter.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.7 [en] (Win95; I)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3219

Last night I checked my code, everything worked, and I submitted.  Today I
was told that a new test had been added to the script, and so I ran it just
to make sure everything was okay.  I now have problems with test 2.H...
where as before I did not.  How has the script been changed?

Also, 2.I suddenly is implemented. I think that calculating RTT is somewhat
implementation specific (given the guidelines of the assignment), and so I
am wondering what decisions the test script makes.

Clayton

Lakshman Shyam Maddali wrote:

> Hi,
>
>     when i run the test_scipt on my code , it gives   errors on the
> FIN/FIN_ACK test cases.
> Last time i posted a question , Arun replied and i believe i implemented
> that behaviour.
> In My Local Test Environment i see that FIN and FIN_ACk are exchanged
> correcly.
>
> But i still have the problem with this incorrect FIN/FIN_ACK behaviour.
>
> Also i have a question on what to do in this case
> --
>  client sends the ACK to the last packet and gets a Zero Byte read on
> the local data
> socket.  Client thinks that all data is ACKed and Sends FIN Packet.
>   But if the ACK to the Last packet is lost, to the  server it seems
> that client is
> misbehaving. isnt it ?
>
> any clarifications are appreciated.
>
> ~
> Lakshman Shyam Maddali

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Test 2.I Suddenly Implemented?
Date: Thu, 7 Feb 2002 17:54:56 -0800
Lines: 28
Distribution: su
Message-ID: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013133296 10039 128.12.133.48 (8 Feb 2002 01:54:56 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3220

I just ran the 2.B test script and noticed that test 2.I (which prior to
today returned NOT TESTED even if all other tests passed) is suddenly
implemented and returning results.

First, this is a low blow--springing a new test on us when we've only got a
handful of test runs left. I would have preferred that you keep the test to
yourselves and use it for grading, rather than ratchet up our stress levels
one more notch at the last second.

Second, I don't believe it's possible for this test to even be useful. The
calculation of round-trip times has been left sufficiently to our
imaginations that expecting us to magically come up with the same solution
as the reference implementation is unrealistic. Do we measure the RTT on a
per-packet basis or a per-byte basis? Do we measure from the send time of
the first part of the sequence? The last? An average? You can make a case
for any of these decisions, and none of them is supported in particular by
the handout, the FAQ, or the newsgroup, so a strictly comparison-based test
is virtually useless.

The model of the test script so far has been "if you fail the test, your
implementation must be incorrect in some way." This new test breaks that. I
would request that it either be removed from the publicly-available script
or that some clarification be presented as to how worried we should be about
not passing it.

Jonathan


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: client.c reading 0 bytes and exiting
Date: Thu, 7 Feb 2002 18:02:13 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3221

Since I didn't get any response I am writing again. I have been unable to
figure out why the client reads 0 byte packet from TCP socket when I don't
send it one. I have also checked that I am not closing local_data_sd. A
double check I did is: When the client reads 0 and calls transport_close, we
close local_data_sd. At this call I check the return value of close - its
0 i.e SUCCESS. I then try closing it again and get a return value of -1
from close i.e. FAILURE. Had I closed it inadvertently anywhere, the first
call to close should have returned me -1.

Now as far as writing 0 bytes is concerned, there's only one function in
my entire code where I send data to application and there I check whether
the buffer I am passing to write() is NULL and datalen <= 0.

Does any of the TAs know of another possibility - I am not using signals
anywhere.

FYI, the same program works fine in reliable mode. Its the -U switch which
breaks it.

Dhawal Kumar

On Thu, 7 Feb 2002, Dhawal Kumar wrote:

> Although I am never closing local_data_sd and never writing 0 bytes to
> local_data_sd, client exits while its in the middle of a file transfer
> because it reads 0 from its socket sd. Any suggestions as to how can I
> figure out whats going on local_data_sd <--> client - I mean is there a
> way I can check the traffic going between the two ends of the pipe.
>
> Dhawal Kumar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: FIN , FIN_ACK questions
Date: Thu, 7 Feb 2002 17:59:03 -0800
Lines: 29
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3222


Hi..

Me,too...
I have been having no problem with 2.H (or either 1.H)...
but, suddenly I'm having NOT OK for both of them...
And, now, only 2 tests left.. quite stressful..
I wish to get some clarification on this..

Hopil

"Clayton Jones"  wrote in message

> Last night I checked my code, everything worked, and I submitted.  Today I
> was told that a new test had been added to the script, and so I ran it
just
> to make sure everything was okay.  I now have problems with test 2.H...
> where as before I did not.  How has the script been changed?
>
> Also, 2.I suddenly is implemented. I think that calculating RTT is
somewhat
> implementation specific (given the guidelines of the assignment), and so I
> am wondering what decisions the test script makes.
>
> Clayton
>



.

Path: shelby.stanford.edu!nntp.stanford.edu!saga5.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: Test 2.I Suddenly Implemented?
Date: Thu, 7 Feb 2002 18:06:15 -0800
Lines: 65
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3223

Hi,

I am also lost on how we can pass this test? The way to evaluate the ERTT
is not strictly defined ( like when you recieve an ack, will you use the
first Uack segment or the last one to calcuate the ERTT? ). When I got the
result of the test, it was:

got:
250 250 250 250 250 250 250 250 250 250 250 250 210 178 153 133 117 104 93
93 75 75 60 60 48
expected:
250 250 250 250 250 250 250 250 200 200 160 160 128 128 102 102 81 81 65
65 52 52 41 41 33

Since the ERTT is not something that is critical for correctness, what I
am see if that my values are converging in the same way at the expected
ones, and therefore, my implementation should be reasonably accepted.
Especially, after another few segments, I am sure both my code and the
test code will have 20msec as the ERTT.

I hope that you can clarify if it is required from us to match this test
exactely, and if we should, please tell us how your test calculate the
ERTT so that we can match it.

Also, can you give us at least 2 extra tests, since this test was added in
the critical last day.

Thanks,

Moh'd


 On Thu, 7 Feb 2002, Jonathan Keljo wrote:

> I just ran the 2.B test script and noticed that test 2.I (which prior to
> today returned NOT TESTED even if all other tests passed) is suddenly
> implemented and returning results.
>
> First, this is a low blow--springing a new test on us when we've only got a
> handful of test runs left. I would have preferred that you keep the test to
> yourselves and use it for grading, rather than ratchet up our stress levels
> one more notch at the last second.
>
> Second, I don't believe it's possible for this test to even be useful. The
> calculation of round-trip times has been left sufficiently to our
> imaginations that expecting us to magically come up with the same solution
> as the reference implementation is unrealistic. Do we measure the RTT on a
> per-packet basis or a per-byte basis? Do we measure from the send time of
> the first part of the sequence? The last? An average? You can make a case
> for any of these decisions, and none of them is supported in particular by
> the handout, the FAQ, or the newsgroup, so a strictly comparison-based test
> is virtually useless.
>
> The model of the test script so far has been "if you fail the test, your
> implementation must be incorrect in some way." This new test breaks that. I
> would request that it either be removed from the publicly-available script
> or that some clarification be presented as to how worried we should be about
> not passing it.
>
> Jonathan
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga5.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: FIN , FIN_ACK questions
Date: Thu, 7 Feb 2002 18:10:03 -0800
Lines: 50
Distribution: su
Message-ID: 
References:  
 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3224

Hi,

Me too :(
What a strange world we live in !!!

I tested my code in the previous days more than 5 times, and I always
passed 2.H. However, today, I did two tests till now and they all failed
it.

Can anyone tell me why this happens?

Thanks,

Moh'd


 On Thu, 7 Feb 2002, BAEHOPIL wrote:

>
> Hi..
>
> Me,too...
> I have been having no problem with 2.H (or either 1.H)...
> but, suddenly I'm having NOT OK for both of them...
> And, now, only 2 tests left.. quite stressful..
> I wish to get some clarification on this..
>
> Hopil
>
> "Clayton Jones"  wrote in message
> 
> > Last night I checked my code, everything worked, and I submitted.  Today I
> > was told that a new test had been added to the script, and so I ran it
> just
> > to make sure everything was okay.  I now have problems with test 2.H...
> > where as before I did not.  How has the script been changed?
> >
> > Also, 2.I suddenly is implemented. I think that calculating RTT is
> somewhat
> > implementation specific (given the guidelines of the assignment), and so I
> > am wondering what decisions the test script makes.
> >
> > Clayton
> >
>
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Yichen Xie" 
Newsgroups: su.class.cs244a
Subject: Re: FIN , FIN_ACK questions
Date: Thu, 7 Feb 2002 18:31:50 -0800
Lines: 59
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: dn800cb605.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3225

Me too am getting the FIN/FINACK problem, and I can't seem to find any clue
about how this could happen. The assert()'s I had in my program explicitly
forbids this, and they were never triggered through normal testing. So I'd
be eternally grateful if someone could look into the script (maybe at least
give us a clue why the script would judge I had a problem).

Also I'm having problems with the new RTT test. It seems the "expected"
version updates the RTT a little sooner than mine does. But after scanning
through the debug information generated by my program, I was quite sure that
RTT is updated as soon as the first qualifying ACK comes in. So again, I'd
be doubly grateful if someone could look into the script (or at least tell
us the criteria of sucess/failure).

Thanks very much!

"Clayton Jones"  wrote in message

> Last night I checked my code, everything worked, and I submitted.  Today I
> was told that a new test had been added to the script, and so I ran it
just
> to make sure everything was okay.  I now have problems with test 2.H...
> where as before I did not.  How has the script been changed?
>
> Also, 2.I suddenly is implemented. I think that calculating RTT is
somewhat
> implementation specific (given the guidelines of the assignment), and so I
> am wondering what decisions the test script makes.
>
> Clayton
>
> Lakshman Shyam Maddali wrote:
>
> > Hi,
> >
> >     when i run the test_scipt on my code , it gives   errors on the
> > FIN/FIN_ACK test cases.
> > Last time i posted a question , Arun replied and i believe i implemented
> > that behaviour.
> > In My Local Test Environment i see that FIN and FIN_ACk are exchanged
> > correcly.
> >
> > But i still have the problem with this incorrect FIN/FIN_ACK behaviour.
> >
> > Also i have a question on what to do in this case
> > --
> >  client sends the ACK to the last packet and gets a Zero Byte read on
> > the local data
> > socket.  Client thinks that all data is ACKed and Sends FIN Packet.
> >   But if the ACK to the Last packet is lost, to the  server it seems
> > that client is
> > misbehaving. isnt it ?
> >
> > any clarifications are appreciated.
> >
> > ~
> > Lakshman Shyam Maddali
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga5.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: Important Print_congestion_window clarifications...
Date: Thu, 7 Feb 2002 18:33:30 -0800
Lines: 77
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3226

Hi,

I still have some question after this clarification:

1. Can we use now curr_ctx in the function? Since in network.c, we are in
the same process of the transport layer (child process)?

2. Also, why we need this function exactely? Is it *only* needed for test
2.I, which compares our RTT with the expected value?

3. After writing this function using curr_ctx->ERTT, I am getting a
strange failure in Test 2.K: Tx fails to enforce 5 Retransmissions. It
tells me the message that I will paste below.

4. Finally, I posted this question twice before, and I didn't get an
answer. I am getting in test 2.I ( 5 retransmissions not enforced ) the
following: "got up to 1 transmissions", and before I the message
"got up to 32 transmissions". Can anyone please tell me what this message
means?

The message that I got when I added the curr_ctx->ERTT in the
print_congestion_window_and_estimatedRTT is:


---
Server output:connected to 171.64.15.39 at port 45785
client: TESTDATA.2


---

---
Client output:TIMEOUT! killed program:
'/afs/ir/class/cs244a/submissions/restricted/hw2.B/xwang00/mrawashd//compile.sun/client.retrans
-p
/afs/ir/class/cs244a/submissions/restricted/hw2.B/xwang00/mrawashd//compile.sun/S1
-U -f TESTDATA.2 foo '  [pid: 4561]


---

Thanks,

Moh'd



On Thu, 7 Feb 2002, Guido Appenzeller wrote:

> In response to the various postings, here are some important clarifications
> on Print_congestion_window (p_c_w):
>
> Contrary to the comment in the source, p_c_w() will only be called from the
> *client* part of transport.c, never from the server part. This means it can
> never be called directly from mysocket.c (ignore the comment in mywrite()).
> There is no need to transition any data from the server part to the client
> part. Accessing anything defined in the server part of transport.c can crash
> your program as after the fork they have different data spaces.
>
> p_c_w() can be callled from network.c. The socket descriptor will always
> point to the sockfd from the network layer (the UDP socket). Some people are
> wondering what to do with the socket descriptor. Depending on how you
> implement the system you don't actually need to use it at all, as each
> Socket will fork off a new thread (which has a seperate data space). A
> global variable is an option here.
>
> As some of you have pointed out printing out the window size actually is not
> very exciting but it is easy  to implement :-)
>
> If there are any more questions please follow up here in the newsgroup.
>
>   Guido
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: test results
Date: Thu, 7 Feb 2002 18:46:21 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3227


hi
My test results report a byte count difference in reliable mode, but I
don't see any such problem when I run it. Since these tests are under
milestone A, I am curious if it is neglecting the time-out cases, i.e on a
timeout(and hence closed connection), the byte counts will definately be
different.
Any comments on this...please!!!

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: test script discrepancies
Date: Thu, 7 Feb 2002 18:53:01 -0800
Lines: 77
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3228


As several people have pointed out, until this afternoon the test script
was not counting the retransmissions properly, usually giving a higher 
number. 

I think I have found what was causing the problem (a local hash that was
not being initialized to zero by perl). Now I have forced the hash to be
reset, and the results should be accurate. If you still find that the
script does not count properly, please let your TA know, and send them the
output for test 2.K, with the time when the test was performed.

One of the reasons to reproduce the packet exchange is to have a double
check on the script (Who would blindly trust a computer, BTW?). You 
probably have found that more information is given to you know, including 
(in some case) the output of the client and server.


Pablo



On 7 Feb 2002, Derrick Wen-Shiuan Tong wrote:

> It happens for me too, on that same test.
> Funny thing is, the output never shows that any packet
> is transmitted more than 6 times, although [RESULT]
> says that it exceeds it.
> 
> 
> Arvind Bhat  wrote:
> 
> : Hi,
> 
> : I am seeing that for the same source, not even a single line changed,
> : the test scripts gave two different results on different submissions.
> 
> : It will be great if someone can explain these failures. Am I the only
> : one seeing the discrepancies in test script runs ?
> 
> : Thanks, Arvind
> 
> 
> 
> : First Time
> : ===============
> 
> :     * Test 2.K: Tx fails to enforce 5 Retransmissions
> :   [out of 2.5]
> 
> : [SUMMARY] NOT OK
> : [RESULT]  5 retransmissions not enforced(INCORRECT!)
> : got up to 18 transmissions
> 
> : ---
> : Your code produced:
> : network_send: a SYN packet with seq=229
> : network_recv: a SYN-ACK packet with seq=178 and ack=230 and timestamp=3795526038
> : network_recv: a DATA-ACK packet with ack=242
> : network_recv: a DATA packet with seq=179, len=20 and timestamp=3795526443
> 
> 
> 
> : Second Time
> : ===============
> 
> :     * Test 2.K: Tx fails to enforce 5 Retransmissions
> :   [out of 2.5]
> 
> : [SUMMARY] OK
> : [RESULT]  5 retransmissions enforced (CORRECT!)
> 
> 
> 
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: FIN , FIN_ACK questions
Date: Thu, 7 Feb 2002 19:01:20 -0800
Lines: 67
Distribution: su
Message-ID: 
References:  
 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3229


I will also jump into this boat :-)

On Thu, 7 Feb 2002, Yichen Xie wrote:

> Me too am getting the FIN/FINACK problem, and I can't seem to find any clue
> about how this could happen. The assert()'s I had in my program explicitly
> forbids this, and they were never triggered through normal testing. So I'd
> be eternally grateful if someone could look into the script (maybe at least
> give us a clue why the script would judge I had a problem).
>
> Also I'm having problems with the new RTT test. It seems the "expected"
> version updates the RTT a little sooner than mine does. But after scanning
> through the debug information generated by my program, I was quite sure that
> RTT is updated as soon as the first qualifying ACK comes in. So again, I'd
> be doubly grateful if someone could look into the script (or at least tell
> us the criteria of sucess/failure).
>
> Thanks very much!
>
> "Clayton Jones"  wrote in message
> 
> > Last night I checked my code, everything worked, and I submitted.  Today I
> > was told that a new test had been added to the script, and so I ran it
> just
> > to make sure everything was okay.  I now have problems with test 2.H...
> > where as before I did not.  How has the script been changed?
> >
> > Also, 2.I suddenly is implemented. I think that calculating RTT is
> somewhat
> > implementation specific (given the guidelines of the assignment), and so I
> > am wondering what decisions the test script makes.
> >
> > Clayton
> >
> > Lakshman Shyam Maddali wrote:
> >
> > > Hi,
> > >
> > >     when i run the test_scipt on my code , it gives   errors on the
> > > FIN/FIN_ACK test cases.
> > > Last time i posted a question , Arun replied and i believe i implemented
> > > that behaviour.
> > > In My Local Test Environment i see that FIN and FIN_ACk are exchanged
> > > correcly.
> > >
> > > But i still have the problem with this incorrect FIN/FIN_ACK behaviour.
> > >
> > > Also i have a question on what to do in this case
> > > --
> > >  client sends the ACK to the last packet and gets a Zero Byte read on
> > > the local data
> > > socket.  Client thinks that all data is ACKed and Sends FIN Packet.
> > >   But if the ACK to the Last packet is lost, to the  server it seems
> > > that client is
> > > misbehaving. isnt it ?
> > >
> > > any clarifications are appreciated.
> > >
> > > ~
> > > Lakshman Shyam Maddali
> >
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Re: test results
Date: Thu, 7 Feb 2002 19:03:41 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3230

On Thu, 7 Feb 2002, Abhishek Das wrote:

> hi
> My test results report a byte count difference in reliable mode, but I
> don't see any such problem when I run it. Since these tests are under
> milestone A, I am curious if it is neglecting the time-out cases, i.e on a
> timeout(and hence closed connection), the byte counts will definately be
> different.
> Any comments on this...please!!!

Same here. As soon as my program started passing most of 2.B, it started
failing most of 2.A. This would be funny if the deadline weren't
tomorrow...

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: About the retransmission.
Date: 8 Feb 2002 03:05:47 GMT
Lines: 48
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3231

Yes, this is for a particular packet. Make sure your RTOs are being 
calculated correctly. You should resend everything, incrementing its 
retransmission count as you resend it. When something is about to be sent 
for the 7th time, this is when you should terminate the connection.

Arun 

Shankar Agarwal  wrote:
: Hi Guido,
: I am still not clear about it. I have one basic question about this
: retransmission bussiness. When we say Maximum of 6 retranmisssion we mean
: 6 transimission in total or 6 transmissions for a particular packet. I
: think its for a particular packet. But then i am not sending any one
: particular packet for more then 6 times then how am i getting 83
: retransmissions.
: Thanks
: Shankar

:  On Thu, 7 Feb 2002, Guido Appenzeller wrote:

:> Shankar,
:>
:> did the follow-up to the last thread answer your open
:> questions? If not, re-post!
:>
:>   Guido
:>
:>
:> ---------------------------------------------------------------
:> Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
:>  - office: 650 7253545  cell: 650 7042781
:>
:> On Wed, 6 Feb 2002, Shankar Agarwal wrote:
:>
:> > Hi,
:> > Has the issue been resolved about the retransmissions. What i remember
:> > from past mails that TA's will soon post a mail about it. But i have not
:> > received any mail about that. Moreover i am not sure why but i am seeing
:> > 83 retranmission statement in the test script which is not possible at
:> > all. Infact the data output from the script never shows 83 packets
:> > received from the sender.
:> > Shankar
:> >
:> >
:> >
:>
:>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic8.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: core dump on select function
Date: 8 Feb 2002 03:10:31 GMT
Organization: Stanford University, CA 94305, USA
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic8.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3232


Hi

I am having trouble with select. When in -U mode if an ACK from the server
gets dropped the client will continue to resend and the server will
continue to ACK but the client never gets the ACK, instead after the
client resends twice the select dumps core. Does any one have any ideas
why this would happen? Purify doesn't point to a problem.

Thank you,

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: About Sliding Window
Date: 8 Feb 2002 03:12:13 GMT
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3233

Consider the following case: suppose packets 1-10 are sent, and packet 1
is lost, and happens to be at the start of the receive window. Now suppose 
the total size is such that packet 10 crosses the end of the window, i.e., 
only part of it is stored. The rest of it is discarded. When packet 1 is 
resent the hole is closed and the window advanced. Now, the ACK is sent 
for the first missing byte of packet 10. Now, if you resend all of packet 
10, rather than just the missing part, this packet will cross the lower 
end of the receive window.

Let me know if you see any problems with this scenario.

Arun

Fang Sun  wrote:
: the assignment description says: "Note that received data may cross either
: end of the current receiver window"
: But when I consider it, I get confused. How can the received data cross
: the either end of the current receiver?
: Since "The first byte of all windows is always the last acknowledged byte
: of data", so the lower boundary of the sender window is equal to or
: lower(when it hasn't received the ACK) than the lower boundary of the
: receiver window. Both window have fixed size 3072, so the upper
: boundary of the sender window is equal to or lower than the upper boundary
: of the receiver window, too.
: And we "Do not send any data outside of your sender window. "
: So, my question is how can a packet cross the upper boundary of the
: receiver window be sent?
: I can't figure out how can a packet cross the lower boundary of the
: receiver window, either?
: So, can somebody give me an example for both cases?
: Thank you very much.


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: estimatedRTT quickly declines to 20
Date: 8 Feb 2002 03:15:00 GMT
Lines: 21
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3234

Yes, the sample RTT should only be for packets that were sent once. 

Arun

Derrick Wen-Shiuan Tong  wrote:

: I'm getting that as well, but only when I turn off all of
: my debugging output. I get sample RTT's of mostly 1 or 2 ms.

: Should we be adjusting our estimated RTT's by taking into account
: how often we timeout, or just use the successful ACK's of packets
: which were sent once?


: Luis Robles  wrote:
: : Although my tests look good, I'm watching my estimatedRTT quickly
: : decline to 20ms when transferring large files...

: : Is this what others are observing as well?

: : Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: estimatedRTT
Date: Thu, 7 Feb 2002 19:20:34 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3235

hi
if we get the ACK of 3-4 segments together,
say 1, 2, 3 and 4.
do we take the last segment i.e 4 's sampleRTT
into consideration, or the first one i.e 1's in calculating estimatedRTT?

thanks

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: client.c reading 0 bytes and exiting
Date: 8 Feb 2002 03:23:37 GMT
Lines: 40
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3236

Without looking at the code it is certainly difficult to say. You are 
saying that a select() triggers for the local_data_sd, though a 
subsequent read indicates the parent has closed the socket?

Arun

 Dhawal Kumar  wrote:
: Since I didn't get any response I am writing again. I have been unable to
: figure out why the client reads 0 byte packet from TCP socket when I don't
: send it one. I have also checked that I am not closing local_data_sd. A
: double check I did is: When the client reads 0 and calls transport_close, we
: close local_data_sd. At this call I check the return value of close - its
: 0 i.e SUCCESS. I then try closing it again and get a return value of -1
: from close i.e. FAILURE. Had I closed it inadvertently anywhere, the first
: call to close should have returned me -1.

: Now as far as writing 0 bytes is concerned, there's only one function in
: my entire code where I send data to application and there I check whether
: the buffer I am passing to write() is NULL and datalen <= 0.

: Does any of the TAs know of another possibility - I am not using signals
: anywhere.

: FYI, the same program works fine in reliable mode. Its the -U switch which
: breaks it.

: Dhawal Kumar

: On Thu, 7 Feb 2002, Dhawal Kumar wrote:

:> Although I am never closing local_data_sd and never writing 0 bytes to
:> local_data_sd, client exits while its in the middle of a file transfer
:> because it reads 0 from its socket sd. Any suggestions as to how can I
:> figure out whats going on local_data_sd <--> client - I mean is there a
:> way I can check the traffic going between the two ends of the pipe.
:>
:> Dhawal Kumar
:>
:>

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Known issues with the current testing script...
Date: Thu, 7 Feb 2002 19:28:32 -0800
Lines: 39
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3237

Hi everyone,

a number of you have complained that allthough their code 
seems to be correct, it fails some tests on the testing
scripts. Generally, this is absolutely possible. The testing
script is meant as a help to test your code. As for some 
parts of the problem there is more than one way to implement
it, the testing script is not 100% reliable. In 90% of the
cases it is, but sometimes it is not. For the final grading
we will at each person's code individually and not purely
rely on the script's output. Then again if the script output
is correct, you have a very good chance of getting full credit.

That being said, there are currently a few known/suspected
bugs with the script. We are working on them but it is
possible that they won't be resolved before the deadline:

1) If you received an error that the number of retransmissions
   is incorrect, check when the test was submitted. If it is 
   before 2pm today, it could be that it now works (we found 
   and fixed a bug).
2) If you get an error in 2.H saying "FIN sent before acking
   all data" this could be an error. Double check that if the
   app closes the socket you wait for all data to be acked 
   before you send the FIN.
3) If the RTT values in test 2.I don't exactly correspond,
   that is fine. If they increase, don't change, drop below
   20 or do other funny things you do have a problem though.
  
Good luck coding, once it is done the worst part of the 
class should lie behind you!

  Guido


---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: SampleRTT Calculations
Date: 8 Feb 2002 03:29:08 GMT
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3238

It is fine to use this ACK to obtain multiple sample RTTs, as long as 
none of the packets have been retransmitted.

Arun

Victor Tung  wrote:
: I have a question about Sample RTT calculations. I'm in need of a bit of a 
: clarification on the scenario when one ACK is received for multiple 
: packets. Suppose the sender sends packets with sequence number 100, 200, 
: and 300; then the receiver gets 100 first, and sends an ACK For 101, but 
: it then receives 300, and then 200 causing it to only send an ACK of 301. 
: Upon receiving this ACK, do we count this as 2 samples in our sample RTT 
: calculations (for packets 200 AND 300), or just one sample. I guess it 
: makes sense to count it as 2 since as the sender, if we counted only one, 
: I can see even more ambiguity as far as deciding which packet to calculate 
: that 1 sample RTT with. 

: Anyway, thanks in advance for anyone helping me clarify this.

: -- 
: Victor Tung			| 
: ------------------------------------------------------------------------------
: Metaphysics is the science of proving what we don't understand.
:                                 -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Can I use signal to implement timeout?
Date: 8 Feb 2002 03:32:18 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3239

You can implement timeouts using the last argument to the select call. 
This will avoid the problem you have mentioned, though you should be 
careful that your timeouts will all be serviced even if there is pending 
activity on all sockets.

Arun

 Fang Sun  wrote:
: I use signal(SIGALRM,...) to implement timeout. But when a signal comes,
: if the client is waiting for read something, the read call will return and
: read 0 byte, and then the client will exist.
: How to figure out this problem? Or it is a wrong way to implement the
: timeout by using signal?
: Thanks.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: test results
Date: 8 Feb 2002 03:37:02 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3240

Certainly we will look into this. However, I think it is safe to say that 
as long as you submission meets the requirements of HW2.B, you will be OK 
(assuming your previous submission met the requirements of HW2.A, that 
is).

Arun

Abhishek Das  wrote:

: hi
: My test results report a byte count difference in reliable mode, but I
: don't see any such problem when I run it. Since these tests are under
: milestone A, I am curious if it is neglecting the time-out cases, i.e on a
: timeout(and hence closed connection), the byte counts will definately be
: different.
: Any comments on this...please!!!

: thanks
: abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Re: About Sliding Window
Date: Thu, 7 Feb 2002 19:38:25 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3241

On 8 Feb 2002, Arun Upadhyaya Kishan wrote:

> Consider the following case: suppose packets 1-10 are sent, and packet 1
> is lost, and happens to be at the start of the receive window. Now suppose
> the total size is such that packet 10 crosses the end of the window, i.e.,
> only part of it is stored. The rest of it is discarded. When packet 1 is
> resent the hole is closed and the window advanced. Now, the ACK is sent
> for the first missing byte of packet 10. Now, if you resend all of packet
> 10, rather than just the missing part, this packet will cross the lower
> end of the receive window.
>
> Let me know if you see any problems with this scenario.

Given the protocol, packet 10 (coming from sender window of size 3072) can
never cross the receiver window upper boundary (of size 3072). In fact,
the sender window sometimes "lags behind" the receiver window, but is
never ahead. This problem could only occur if, for retransmission, you
create new packets. Thus, if you send bytes 10-19, 20-29, and 30+ in
separate packets, 10-19 times out, and then you resend the data as two
packets - 10-19 and 20-30+ - the receiver could get 10-19 and 20-29 and
ACK the middle of the new packet 20-30+. But we are not recreating packets
anyway. So, my humble opinion is that this whole packet chopping hubbub
served to confuse more people than it should have (myself included).

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: estimatedRTT
Date: 8 Feb 2002 03:39:00 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3242

It is fine to use the ACK to calculate the sampleRTT four times, as noted 
before, as long as none of the four packets has been retransmitted.

Arun

Abhishek Das  wrote:
: hi
: if we get the ACK of 3-4 segments together,
: say 1, 2, 3 and 4.
: do we take the last segment i.e 4 's sampleRTT
: into consideration, or the first one i.e 1's in calculating estimatedRTT?

: thanks

: Abhishek Das
: Graduate Research Assistant
: Computer Systems Lab
: Stanford University

: Address:-
: Escondido Village 33B
: Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine11.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: client.c reading 0 bytes and exiting
Date: Thu, 7 Feb 2002 19:40:17 -0800
Lines: 54
Distribution: su
Message-ID: 
References: 
 
 
NNTP-Posting-Host: elaine11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3243

No. My client side application (application layer) is reading 0 bytes from
the socket it has opened for reading the file from server although I never
close the other end point of the connection (which is local_data_sd in the
transport layer) in child process or send a data of length 0 over this
connection.

If you can please take a look at my code, that would be great.

Dhawal Kumar

On 8 Feb 2002, Arun Upadhyaya Kishan wrote:

> Without looking at the code it is certainly difficult to say. You are
> saying that a select() triggers for the local_data_sd, though a
> subsequent read indicates the parent has closed the socket?
>
> Arun
>
>  Dhawal Kumar  wrote:
> : Since I didn't get any response I am writing again. I have been unable to
> : figure out why the client reads 0 byte packet from TCP socket when I don't
> : send it one. I have also checked that I am not closing local_data_sd. A
> : double check I did is: When the client reads 0 and calls transport_close, we
> : close local_data_sd. At this call I check the return value of close - its
> : 0 i.e SUCCESS. I then try closing it again and get a return value of -1
> : from close i.e. FAILURE. Had I closed it inadvertently anywhere, the first
> : call to close should have returned me -1.
>
> : Now as far as writing 0 bytes is concerned, there's only one function in
> : my entire code where I send data to application and there I check whether
> : the buffer I am passing to write() is NULL and datalen <= 0.
>
> : Does any of the TAs know of another possibility - I am not using signals
> : anywhere.
>
> : FYI, the same program works fine in reliable mode. Its the -U switch which
> : breaks it.
>
> : Dhawal Kumar
>
> : On Thu, 7 Feb 2002, Dhawal Kumar wrote:
>
> :> Although I am never closing local_data_sd and never writing 0 bytes to
> :> local_data_sd, client exits while its in the middle of a file transfer
> :> because it reads 0 from its socket sd. Any suggestions as to how can I
> :> figure out whats going on local_data_sd <--> client - I mean is there a
> :> way I can check the traffic going between the two ends of the pipe.
> :>
> :> Dhawal Kumar
> :>
> :>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: About Sliding Window
Date: 8 Feb 2002 03:43:56 GMT
Lines: 41
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine36.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3244

Some students have implemented coaelescing of packets as you have noted,
so the scenario below is possible. In general, with windows set to the 
same size this cannot happen, as you have noted, as long as the sender is 
careful to not overrun the receiver's window. 

Arun

Nickolay Boytchev Stanev 
wrote: : On 8 Feb 2002, Arun Upadhyaya Kishan wrote:

:> Consider the following case: suppose packets 1-10 are sent, and packet 1
:> is lost, and happens to be at the start of the receive window. Now suppose
:> the total size is such that packet 10 crosses the end of the window, i.e.,
:> only part of it is stored. The rest of it is discarded. When packet 1 is
:> resent the hole is closed and the window advanced. Now, the ACK is sent
:> for the first missing byte of packet 10. Now, if you resend all of packet
:> 10, rather than just the missing part, this packet will cross the lower
:> end of the receive window.
:>
:> Let me know if you see any problems with this scenario.

: Given the protocol, packet 10 (coming from sender window of size 3072) can
: never cross the receiver window upper boundary (of size 3072). In fact,
: the sender window sometimes "lags behind" the receiver window, but is
: never ahead. This problem could only occur if, for retransmission, you
: create new packets. Thus, if you send bytes 10-19, 20-29, and 30+ in
: separate packets, 10-19 times out, and then you resend the data as two
: packets - 10-19 and 20-30+ - the receiver could get 10-19 and 20-29 and
: ACK the middle of the new packet 20-30+. But we are not recreating packets
: anyway. So, my humble opinion is that this whole packet chopping hubbub
: served to confuse more people than it should have (myself included).

: 							NStanev
: -- 
: ####################################################
: # Nickolay Stanev                                  #
: # E-mail:                   #
: # Cell: (650)-269-4756                             #
: # URL: http://www.stanford.edu/~nbstanev/          #
: ####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Exponential Backoff, client got killed by test script
Date: Thu, 7 Feb 2002 20:23:11 -0800
Lines: 98
Distribution: su
Message-ID: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3245

It looks like the testing script is killing the client or server after a fix
time, which is good. But for those of us who have implemented exponential
back off, the testing script is killing my client before 5 retransmissions
occur. So is there a way to get around this? Or we don't need to implement
exponential backoff at all?

Here is my test output for 2.K:

* Test 2.K: Tx fails to enforce 5 Retransmissions
  [out of 2.5]

[SUMMARY] NOT OK
[RESULT]  5 retransmissions not enforced(INCORRECT!)
got up to 4 transmissions

---
Your code produced:
network_send: a SYN packet with seq=106
network_recv: a SYN-ACK packet with seq=56 and ack=107 and
timestamp=3823996026
network_recv: a DATA-ACK packet with ack=119
network_recv: a DATA packet with seq=57, len=20 and timestamp=3823996528
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=77, len=536 and timestamp=3823996528
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=613, len=536 and timestamp=3823996528
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1149, len=536 and timestamp=3823996529
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1685, len=536 and timestamp=3823996529
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2221, len=536 and timestamp=3823996529
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2757, len=372 and timestamp=3823996529
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=57, len=20 and timestamp=3823997427
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=77, len=536 and timestamp=3823997427
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=613, len=536 and timestamp=3823997427
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1149, len=536 and timestamp=3823997427
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1685, len=536 and timestamp=3823997427
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2221, len=536 and timestamp=3823997428
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2757, len=372 and timestamp=3823997428
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=57, len=20 and timestamp=3824001437
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=77, len=536 and timestamp=3824001437
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=613, len=536 and timestamp=3824001437
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1149, len=536 and timestamp=3824001437
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1685, len=536 and timestamp=3824001438
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2221, len=536 and timestamp=3824001438
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2757, len=372 and timestamp=3824001438
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=57, len=20 and timestamp=3824009447
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=77, len=536 and timestamp=3824009447
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=613, len=536 and timestamp=3824009448
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1149, len=536 and timestamp=3824009448
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=1685, len=536 and timestamp=3824009448
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2221, len=536 and timestamp=3824009448
network_recv(2): Dropping data packet num 0
network_recv: a DATA packet with seq=2757, len=372 and timestamp=3824009448
network_recv(2): Dropping data packet num 0

---
Server output:connected to 171.64.15.39 at port 47332
client: TESTDATA.2


---

---
Client output:TIMEOUT! killed program:
'/afs/ir/class/cs244a/submissions/restricted/hw2.B/xwang00/hwfu//compile.sun
/client.retrans -p
/afs/ir/class/cs244a/submissions/restricted/hw2.B/xwang00/hwfu//compile.sun/
S1 -U -f TESTDATA.2 foo '  [pid: 4601]


---




.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Exponential Backoff, client got killed by test script
Date: 8 Feb 2002 04:40:25 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3246


>It looks like the testing script is killing the client or server after a fix
>time, which is good. But for those of us who have implemented exponential
>back off, the testing script is killing my client before 5 retransmissions
>occur. So is there a way to get around this? Or we don't need to implement
>exponential backoff at all?

In fact, I have to amend my earlier comments, in which I wrote that I
found exponential backoff necessary.  Those were true, but the comments
were based on my previous experience, in which there was no minimum RTT.
Since we're testing on a lightly loaded local network, you'll see that
the actual RTTs are on the order of 1 ms or so.  If you used these numbers
as your actual ERTT, you'd probably find that exp. backoff is necessary
to avoid spurious timeouts.

*But*, in this version of the assignment, ERTT is clamped to 20 ms--this is a
very generous amount of time in our setting.  So exponential backoff shouldn't
be required given this--I'd be surprised if you saw any difference in
your program's functionality if you disabled it.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: core dump on select function
Date: 8 Feb 2002 04:41:34 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3247


>I am having trouble with select. When in -U mode if an ACK from the server
>gets dropped the client will continue to resend and the server will
>continue to ACK but the client never gets the ACK, instead after the
>client resends twice the select dumps core. Does any one have any ideas
>why this would happen? Purify doesn't point to a problem.

Are your fd_set and timeout pointers valid?  Is nfds set correctly?

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Test 2.I Suddenly Implemented?
Date: 8 Feb 2002 04:45:46 GMT
Lines: 39
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3248

"Jonathan Keljo"  writes:

>I just ran the 2.B test script and noticed that test 2.I (which prior to
>today returned NOT TESTED even if all other tests passed) is suddenly
>implemented and returning results.

>First, this is a low blow--springing a new test on us when we've only got a
>handful of test runs left. I would have preferred that you keep the test to
>yourselves and use it for grading, rather than ratchet up our stress levels
>one more notch at the last second.

>Second, I don't believe it's possible for this test to even be useful. The
>calculation of round-trip times has been left sufficiently to our
>imaginations that expecting us to magically come up with the same solution
>as the reference implementation is unrealistic. Do we measure the RTT on a
>per-packet basis or a per-byte basis? Do we measure from the send time of
>the first part of the sequence? The last? An average? You can make a case
>for any of these decisions, and none of them is supported in particular by
>the handout, the FAQ, or the newsgroup, so a strictly comparison-based test
>is virtually useless.

>The model of the test script so far has been "if you fail the test, your
>implementation must be incorrect in some way." This new test breaks that. I
>would request that it either be removed from the publicly-available script
>or that some clarification be presented as to how worried we should be about
>not passing it.


Not really--the model is rather "if you pass the test, you're almost certainly
doing fine, and if you don't, we have to look at it manually."

Don't stress out about it.  (The tests are intended to help, not cause you
grief...)  As you'll have observed, the spec (like those in the real-world)
is quite ambiguous about how to calculate these kinds of things, and
everyone reading this probably has his/her own interpretation and/or
implementation.  (All we can look at during grading is whether you followed
the basic guidlines given in the assignment.  We aren't expecting to be able
to compare number-for-number output and have them match exactly).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Can I use signal to implement timeout?
Date: 8 Feb 2002 04:49:26 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3249

Fang Sun  writes:

>I use signal(SIGALRM,...) to implement timeout. But when a signal comes,
>if the client is waiting for read something, the read call will return and
>read 0 byte, and then the client will exist.
>How to figure out this problem? Or it is a wrong way to implement the
>timeout by using signal?
>Thanks.


I agree with Arun's solution, but just wanted to add that you should keep in
mind that read() (and many other blocking system calls) can return -1 with
errno set to EINTR if a signal is received.  (Often, whenever you have such
a call, you should explicitly handle errno == EINTR by retrying the call,
assuming that these are the semantics you were looking for).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Important Print_congestion_window clarifications...
Date: 8 Feb 2002 04:52:20 GMT
Lines: 31
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3250


>1. Can we use now curr_ctx in the function? Since in network.c, we are in
>the same process of the transport layer (child process)?

You can implement it however you want; I used curr_ctx in mine when called
from the child process.

>2. Also, why we need this function exactely? Is it *only* needed for test
>2.I, which compares our RTT with the expected value?

Yes, it's used so we can verify that your RTT is behaving (perhaps roughly) as
expected.

>3. After writing this function using curr_ctx->ERTT, I am getting a
>strange failure in Test 2.K: Tx fails to enforce 5 Retransmissions. It
>tells me the message that I will paste below.

Please make sure you aren't making any kind of silly mistake when you
modified the function; a number of people have reported other random errors
popping up when they modified p_c_w().

>4. Finally, I posted this question twice before, and I didn't get an
>answer. I am getting in test 2.I ( 5 retransmissions not enforced ) the
>following: "got up to 1 transmissions", and before I the message
>"got up to 32 transmissions". Can anyone please tell me what this message
>means?

The question has been asked (and I think answered) in several threads--
if this is an old message, it may well pertain to the bug that Pablo had
posted about.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: retransmit exponential timeouts.
Date: 8 Feb 2002 05:06:22 GMT
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3251


>from the previous posts, i could understand that we should be
>using exponential backoff for retransmissions. If say t1, t2,
>t3, t4 and t5 are the timeout values for each retransmission,
>are we going to use (A) or (B) ?

>A) t1 = 2 * ERTT,
>   t2 = 4 * ERTT,
>   t3 = 8 * ERTT,
>   t4 = 16 * ERTT,
>   t5 = 32 * ERTT

>OR

>B) t1 = 2 * ERTT,
>   t2 = 2 * t1,
>   t3 = 4 * t1,
>   t4 = 8 * t1,
>   t5 = 16 * t1

The book actually includes a brief discussion about these choices.

However, if you're referring to my previous post from a few days ago,
it turns out that clamping the ERTT to a minimum of 20 ms should pretty
much obviate the need for exponential backoff (or the hack where segments
that are retransmitted as part of Go-back N are not counted, too, for
that matter), so you should even be able to get by without.  (I updated
my implementation to include the 20 ms minimum ERTT, and could disable
exp. backoff without incurring spurious timeouts).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine14.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: A Request to the TAs
Date: 8 Feb 2002 05:11:35 GMT
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine14.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3252


>I think everyone involved would have it easier if the TAs could get
>together, decide on and post a unified statement of what the parameters are
>for timeouts and retransmissions in this project. It would help to know
>three things:

>1) What elements of the implementation we must do in a particular manner.
>(E.g., use Karn-Partridge, keep timeouts on a per-packet basis, resend only
>>X rather than all unACKed)
>2) What elements of the implementation we are free to design ourselves.
>(E.g., transitioning between timeouts determined by backoff and timeouts
>determined by RTT samples.)
>3) What behavior characteristics are required. (E.g., transferring large
>files without killing the connection because of excessive resends.)

Sorry for the confusion--as you can tell, everyone has a different
interpretation of the requirements, and more than one answer can be valid.

The short answer is that we're interested in seeing reliable delivery from
your implementation, in the spirit of the STCP specification.  If your
output matches the test script, you're doing fine.  If it doesn't, that
doesn't necessarily mean that your implementation is wrong, just that
we have to look at it manually.

I think Guido or Xin had posted a summary elsewhere on the newsgroup, which
may have been lost in the clutter of posts but basically described what
the assignment is looking for.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: FIN , FIN_ACK questions
Date: 8 Feb 2002 05:38:07 GMT
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine41.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3253

Lakshman,

The active end of network termination has to ensure that all of the
*data* that it has sent is ACK'ed. It does not have to verify that
the other side has received all of its ACK's--there's no way you
can know that.

- Derrick


Lakshman Shyam Maddali  wrote:
: Hi,

:     when i run the test_scipt on my code , it gives   errors on the
: FIN/FIN_ACK test cases.
: Last time i posted a question , Arun replied and i believe i implemented
: that behaviour.
: In My Local Test Environment i see that FIN and FIN_ACk are exchanged
: correcly.

: But i still have the problem with this incorrect FIN/FIN_ACK behaviour.

: Also i have a question on what to do in this case
: --
:  client sends the ACK to the last packet and gets a Zero Byte read on
: the local data
: socket.  Client thinks that all data is ACKed and Sends FIN Packet.
:   But if the ACK to the Last packet is lost, to the  server it seems
: that client is 
: misbehaving. isnt it ?

: any clarifications are appreciated.

: ~
: Lakshman Shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic5.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Test results: Script working???
Date: 8 Feb 2002 06:16:30 GMT
Organization: Stanford University, CA 94305, USA
Lines: 33
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic5.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3254


I am failing on test 2.G and 2.k where I was passing before the results I
get are 

>     * Test 2.K: Tx fails to enforce 5 Retransmissions
>   [out of 2.5]
> 
> [SUMMARY] NOT OK
> [RESULT]  5 retransmissions not enforced(INCORRECT!)
> got up to 0 transmissions
> 
> ---
> Your code produced:
> 
> ---
> Server output:
> 
> ---
> 
> ---
> Client output:myconnect: mismatch in peer names  ("0.0.0.0:48518" != "0.0.0.0:0")
> myconnect: Bad file number
> recv: Error 0
> 
> 
> ---
> 

Is this somthing I did?

Thanks,

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Romain Thibaux 
Newsgroups: su.class.cs244a
Subject: Re: Important Print_congestion_window clarifications...
Date: Thu, 07 Feb 2002 22:16:33 -0800
Lines: 5
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: thibaux.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1013149389 12357 128.12.191.77 (8 Feb 2002 06:23:09 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.4.2-2 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3255

Hey, which window is it by the way ? I assumed it was the out window because it
was simpler for me but...

    Romain

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Re: Important Print_congestion_window clarifications...
Date: Thu, 7 Feb 2002 22:27:04 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
 
  
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3256

On Thu, 7 Feb 2002, Romain Thibaux wrote:

> Hey, which window is it by the way ? I assumed it was the out window because it
> was simpler for me but...

You can just print 3072. The CWND is not really part of this assignment...

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: Re: core dump on select function
Date: Thu, 7 Feb 2002 22:42:40 -0800
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3257

I was getting this error b/c I was checking to see if a socket which had
been set to -1 was in my FD_SET.

You might be doing that too...

Pete
"Mark Joseph Dolan"  wrote in message

>
> Hi
>
> I am having trouble with select. When in -U mode if an ACK from the server
> gets dropped the client will continue to resend and the server will
> continue to ACK but the client never gets the ACK, instead after the
> client resends twice the select dumps core. Does any one have any ideas
> why this would happen? Purify doesn't point to a problem.
>
> Thank you,
>
> -mark


.

Path: shelby.stanford.edu!nntp.stanford.edu!manzanares.Stanford.EDU!molinero
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Test results: Script working???
Date: Fri, 8 Feb 2002 00:11:03 -0800
Lines: 43
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3258


Your parent and your child processes though they were not connected to the 
same peer, this is why your client died before transmittinganything.

Pablo

On 8 Feb 2002, Mark Joseph Dolan wrote:

> 
> I am failing on test 2.G and 2.k where I was passing before the results I
> get are 
> 
> >     * Test 2.K: Tx fails to enforce 5 Retransmissions
> >   [out of 2.5]
> > 
> > [SUMMARY] NOT OK
> > [RESULT]  5 retransmissions not enforced(INCORRECT!)
> > got up to 0 transmissions
> > 
> > ---
> > Your code produced:
> > 
> > ---
> > Server output:
> > 
> > ---
> > 
> > ---
> > Client output:myconnect: mismatch in peer names  ("0.0.0.0:48518" != "0.0.0.0:0")
> > myconnect: Bad file number
> > recv: Error 0
> > 
> > 
> > ---
> > 
> 
> Is this somthing I did?
> 
> Thanks,
> 
> -mark
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: FIN , FIN_ACK questions
Date: Fri, 8 Feb 2002 01:31:07 -0800
Lines: 42
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3259


There were many people who received unexpected NOT OK for 2.H..
Guido said it would probably be real error on our part, not test script's
error..
But, somehow, I could never find what's wrong with my code..
At least, my code has only two lines that can send TH_FIN,
either of which can only be activated when Tx queue is empty.. ( surrounded
by simple if statement )
Would anyone, who found out what's wrong with his/her code, give me a hint?
I'm passing all other tests..

Thanks
Hopil

"Lakshman Shyam Maddali"  wrote in message

> Hi,
>
>     when i run the test_scipt on my code , it gives   errors on the
> FIN/FIN_ACK test cases.
> Last time i posted a question , Arun replied and i believe i implemented
> that behaviour.
> In My Local Test Environment i see that FIN and FIN_ACk are exchanged
> correcly.
>
> But i still have the problem with this incorrect FIN/FIN_ACK behaviour.
>
> Also i have a question on what to do in this case
> --
>  client sends the ACK to the last packet and gets a Zero Byte read on
> the local data
> socket.  Client thinks that all data is ACKed and Sends FIN Packet.
>   But if the ACK to the Last packet is lost, to the  server it seems
> that client is
> misbehaving. isnt it ?
>
> any clarifications are appreciated.
>
> ~
> Lakshman Shyam Maddali


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: test script down?
Date: Fri, 8 Feb 2002 04:00:29 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3260


I submitted my code for testing around 40 minutes back. No reply as yet.

-anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Failing hw2.A script
Date: 8 Feb 2002 12:28:10 GMT
Organization: Stanford University, CA 94305, USA
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3261

Hi,

I pass the B script (except for 2.I :-( ) but I am failing 1.C for
script A. The handout says the propgrams must ALSO work in unreliable mode
so will we be tested against the first script as well? 

Any ideas why it would pass the unreliable script and not the reliable?
Are they run against different code?

Thanks,

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sutin Chen 
Newsgroups: su.class.cs244a
Subject: test script issues
Date: Fri, 08 Feb 2002 08:15:54 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga18.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3263

i got an e-mail saying my test was started 45 min ago, but still haven't
gotten any results.

can anyone do anything about this? thanks!
.

Path: shelby.stanford.edu!nntp.stanford.edu!fable17.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: About milestone 3
Date: Fri, 8 Feb 2002 10:37:18 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable17.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3264

Hi,

I have some questions on milestone C:

(1) If the supplied ftpclient(object file)  runs well with STCP in
milestone B, we actually do not need to make change to our own source
code of ftpclient and submit it, right?

(2) The delieverable files include: help.o and main.o for ftpclient,
but I only fine libftpcopy.a, libftpd.a and libsupport.a in the
provided dir. Are they the same thing?

Thanks a lot!!!!


Best,
-jinhui

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine2.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Important Print_congestion_window clarifications...
Date: 8 Feb 2002 19:01:03 GMT
Lines: 10
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine2.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3265

Romain Thibaux  writes:

>Hey, which window is it by the way ? I assumed it was the out window because it
>was simpler for me but...


To be honest, it doesn't matter--as was posted previously, the function is
somewhat of a vestige of years past, when a simple congestion window was
really implemented.  For this assignment, the congestion window is fixed to
3072 bytes, i.e. the full (best case) STCP window size.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Re: About milestone 3
Date: Fri, 8 Feb 2002 11:22:17 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3266

More question for milestone 3,

The supplied object code does not support the -U option, is there something
I did wrong? Also, it is not doing the right thing as it copies empty
directory as well. So if we use the supplied object code to submit for
milestone 3, will we get penalize for not passing the special case in PA1?

Thanks,
Henry


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: test script issues
Date: Fri, 08 Feb 2002 11:25:14 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.8) Gecko/20020205
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3267

Acording to our logs, you got a report at 7:42am today. Is this the one 
that you were waiting for?

The test script seems to have been running smoothly tonight. Sometimes 
if there are a lot of students, it can take very long to return a report.

Pablo


Sutin Chen wrote:
> i got an e-mail saying my test was started 45 min ago, but still haven't
> gotten any results.
> 
> can anyone do anything about this? thanks!
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga8.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: test script issues
Date: Fri, 8 Feb 2002 11:54:09 -0800
Lines: 27
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3268

The script cannot compile my code ( which is compiling, and I tested it
before). Can someone check for this?

Thanks,

Moh'd
On Fri, 8 Feb 2002, Pablo Molinero Fernandez wrote:

> Acording to our logs, you got a report at 7:42am today. Is this the one
> that you were waiting for?
>
> The test script seems to have been running smoothly tonight. Sometimes
> if there are a lot of students, it can take very long to return a report.
>
> Pablo
>
>
> Sutin Chen wrote:
> > i got an e-mail saying my test was started 45 min ago, but still haven't
> > gotten any results.
> >
> > can anyone do anything about this? thanks!
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!fable9.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: Test scripts fail to compile code??
Date: Fri, 8 Feb 2002 11:56:10 -0800
Lines: 50
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3269


Please lookinto this?? I think something wrong with test script
again...:(


    * Test 2.A: Fails to compile
  [out of 40]

[SUMMARY] NOT OK
[RESULT]  ERROR:  could not compile (512).
-------
umask 022
gcc -g -ansi -DSOLARIS  -c client.c
gcc -g -ansi -DSOLARIS  -c transport.c
gcc -g -ansi -DSOLARIS  -c network.c
gcc -g -ansi -DSOLARIS  -c mysock.c
gcc -g -ansi -DSOLARIS  -o client client.o transport.o network.o mysock.o
-lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c server.c
gcc -o server server.o transport.o network.o mysock.o -lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c finserver.c
gcc -g -ansi -DSOLARIS  -DGRADING -c network_grading.c
gcc -o finserver finserver.o transport.o network_grading.o mysock.o
-lsocket
-lnsl -lm
gcc -g -ansi -DSOLARIS  -c finclient.c
gcc -o finclient finclient.o transport.o network.o mysock.o -lsocket -lnsl
-lm
gcc -g -ansi -DSOLARIS  -c testclient.c
gcc -g -ansi -DSOLARIS  -DFIXED_INITNUM -c transport.c -o
transport_fixed_init.o
gcc -o testclient testclient.o transport_fixed_init.o network.o mysock.o
-lsocket -lnsl -lm
chmod -f 755 client server client.window badserver finserver \
finclient testclient testserver badclient client.retrans server.no_rtt \
ftpd ftpcopy 2>/dev/null
gmake: *** [change_mod] Error 1


Gaurav

*************************************************************************
Gaurav Garg                          * Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             *       Escondido Village, Stanford
Department of Electrical Engineering *       CA-94305
Stanford University                  * Ph:650-498-1208
*************************************************************************

-Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: test script issues
Date: Fri, 8 Feb 2002 11:57:30 -0800
Lines: 60
Distribution: su
Message-ID: 
References:  
 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3270

Same here. The script was working an ago and now gives compile errors.Will
these be counted towards total test submissions?

gcc -g -ansi -DSOLARIS  -c client.c
gcc -g -ansi -DSOLARIS  -c transport.c
gcc -g -ansi -DSOLARIS  -c network.c
gcc -g -ansi -DSOLARIS  -c mysock.c
gcc -g -ansi -DSOLARIS  -o client client.o transport.o network.o mysock.o
-lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c server.c
gcc -o server server.o transport.o network.o mysock.o -lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c finserver.c
gcc -g -ansi -DSOLARIS  -DGRADING -c network_grading.c
gcc -o finserver finserver.o transport.o network_grading.o mysock.o
-lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c finclient.c
gcc -o finclient finclient.o transport.o network.o mysock.o -lsocket -lnsl
-lm
gcc -g -ansi -DSOLARIS  -c testclient.c
gcc -g -ansi -DSOLARIS  -DFIXED_INITNUM -c transport.c -o
transport_fixed_init.o
gcc -o testclient testclient.o transport_fixed_init.o network.o mysock.o
-lsocket -lnsl -lm
chmod -f 755 client server client.window badserver finserver \
finclient testclient testserver badclient client.retrans server.no_rtt \
ftpd ftpcopy 2>/dev/null
gmake: *** [change_mod] Error 1


On Fri, 8 Feb 2002, Moh'd Saleem Saleem Alrawashdeh wrote:

> The script cannot compile my code ( which is compiling, and I tested it
> before). Can someone check for this?
>
> Thanks,
>
> Moh'd
> On Fri, 8 Feb 2002, Pablo Molinero Fernandez wrote:
>
> > Acording to our logs, you got a report at 7:42am today. Is this the one
> > that you were waiting for?
> >
> > The test script seems to have been running smoothly tonight. Sometimes
> > if there are a lot of students, it can take very long to return a report.
> >
> > Pablo
> >
> >
> > Sutin Chen wrote:
> > > i got an e-mail saying my test was started 45 min ago, but still haven't
> > > gotten any results.
> > >
> > > can anyone do anything about this? thanks!
> > >
> >
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Shyam Maddali 
Newsgroups: su.class.cs244a
Subject: Compile  Errors when Test Script is Run
Date: Fri, 08 Feb 2002 11:58:02 -0800
Lines: 44
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic16.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3271

Hi ,
   Yesterday when i ran the test most of the things looked fine,
But know the test script complains about compile errors.
i am sure that the program compiles ..

please take a look at this. 
and let me know if there is any obvious  problem..

Thanks
Lakshman Maddali

 RESULTS FOR THE DIFFERENT TESTS:

    * Test 2.A: Fails to compile
  [out of 40]

[SUMMARY] NOT OK
[RESULT]  ERROR:  could not compile (512).
-------
umask 022
gcc -g -ansi -DSOLARIS  -c client.c
gcc -g -ansi -DSOLARIS  -c transport.c
gcc -g -ansi -DSOLARIS  -c network.c
gcc -g -ansi -DSOLARIS  -c mysock.c
gcc -g -ansi -DSOLARIS  -o client client.o transport.o network.o
mysock.o -lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c server.c
gcc -o server server.o transport.o network.o mysock.o -lsocket -lnsl -lm 
gcc -g -ansi -DSOLARIS  -c finserver.c
gcc -g -ansi -DSOLARIS  -DGRADING -c network_grading.c
gcc -o finserver finserver.o transport.o network_grading.o mysock.o
-lsocket -lnsl -lm
gcc -g -ansi -DSOLARIS  -c finclient.c
gcc -o finclient finclient.o transport.o network.o mysock.o -lsocket
-lnsl -lm
gcc -g -ansi -DSOLARIS  -c testclient.c
gcc -g -ansi -DSOLARIS  -DFIXED_INITNUM -c transport.c -o
transport_fixed_init.o
gcc -o testclient testclient.o transport_fixed_init.o network.o mysock.o
-lsocket -lnsl -lm
chmod -f 755 client server client.window badserver finserver \
finclient testclient testserver badclient client.retrans server.no_rtt \
ftpd ftpcopy 2>/dev/null
gmake: *** [change_mod] Error 1
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tudor Andrei Bosman 
Newsgroups: su.class.cs244a
Subject: Re: Compile  Errors when Test Script is Run
Date: 8 Feb 2002 20:03:35 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3272

Yes, same here.  Please fix it?

Thanks,
Tudor.

Lakshman Shyam Maddali  wrote:
>     * Test 2.A: Fails to compile
>   [out of 40]

> chmod -f 755 client server client.window badserver finserver \
> finclient testclient testserver badclient client.retrans server.no_rtt \
> ftpd ftpcopy 2>/dev/null
> gmake: *** [change_mod] Error 1
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Compile  Errors when Test Script is Run
Date: Fri, 08 Feb 2002 12:14:19 -0800
Lines: 26
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.8) Gecko/20020205
X-Accept-Language: es, en-us, en, fr, de
To: Tudor Andrei Bosman 
Xref: nntp.stanford.edu su.class.cs244a:3273

Hi,

Today from 11:46am for 10 minutes the script was broken, and prevented 
people from compiling. You will not need to resubmit your code, as the 
script will rerun your test again, for those who had this error message.

Pablo

Tudor Andrei Bosman wrote:
> Yes, same here.  Please fix it?
> 
> Thanks,
> Tudor.
> 
> Lakshman Shyam Maddali  wrote:
> 
>>    * Test 2.A: Fails to compile
>>  [out of 40]
>>
> 
>>chmod -f 755 client server client.window badserver finserver \
>>finclient testclient testserver badclient client.retrans server.no_rtt \
>>ftpd ftpcopy 2>/dev/null
>>gmake: *** [change_mod] Error 1
>>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Compile errors in test script
Date: Fri, 8 Feb 2002 12:54:08 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3274

Did anyone of you who got compile errors, get a fresh evaluation report?
I haven't received anything yet.


Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Shyam Maddali 
Newsgroups: su.class.cs244a
Subject: Re: Compile errors in test script
Date: Fri, 08 Feb 2002 12:58:22 -0800
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic16.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3275


No, still waiting for the report..

~
Lakshman Shyam Maddali

Dhawal Kumar wrote:
> 
> Did anyone of you who got compile errors, get a fresh evaluation report?
> I haven't received anything yet.
> 
> Dhawal Kumar
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Pablo Molinero Fernandez 
Newsgroups: su.class.cs244a
Subject: Re: Compile errors in test script
Date: Fri, 08 Feb 2002 13:10:11 -0800
Lines: 20
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: manzanares.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.8) Gecko/20020205
X-Accept-Language: es, en-us, en, fr, de
Xref: nntp.stanford.edu su.class.cs244a:3276

Right now there are 8 people in the queue, and it takes up to 10 minutes 
to do a full test of a student. You can do the math to see how long the 
wait can be.

Pablo

Lakshman Shyam Maddali wrote:
> No, still waiting for the report..
> 
> ~
> Lakshman Shyam Maddali
> 
> Dhawal Kumar wrote:
> 
>>Did anyone of you who got compile errors, get a fresh evaluation report?
>>I haven't received anything yet.
>>
>>Dhawal Kumar
>>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tudor Andrei Bosman 
Newsgroups: su.class.cs244a
Subject: test script?
Date: 8 Feb 2002 21:41:37 GMT
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3277

Hey,

Is the test script down again?  I was one of those who submitted code when
the test script wasn't compiling correctly.  I submitted again around 12:10
- still no reply, not even the "test started" message.  Do we really have a
90 minute backlog or is something wrong with the script?

Thanks,
Tudor.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tudor Andrei Bosman 
Newsgroups: su.class.cs244a
Subject: more test script blues
Date: 8 Feb 2002 23:01:20 GMT
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3278

 RESULTS FOR THE DIFFERENT TESTS:

    * Test 2.A: Fails to compile
  [out of 40]

[SUMMARY] NOT OK
[RESULT]  ERROR:  could not compile (512).
-------
umask 022
gcc -g -ansi -DSOLARIS  -c client.c
gmake: *** No rule to make target `transport.c', needed by `transport.o'.  Stop.


Thanks,
Tudor.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: Compile errors in test script
Date: 8 Feb 2002 23:16:19 GMT
Lines: 34
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine0.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3279

Pablo Molinero Fernandez  wrote:

> Right now there are 8 people in the queue, and it takes up to 10 minutes 
> to do a full test of a student. You can do the math to see how long the 
> wait can be.

> Pablo

	I don't understand why you can't run the tests in parallel? It 
would really speed things up quite a bit and we wouldn't suffer excessive 
waits immediately before the deadline.

-Victor

> Lakshman Shyam Maddali wrote:
>> No, still waiting for the report..
>> 
>> ~
>> Lakshman Shyam Maddali
>> 
>> Dhawal Kumar wrote:
>> 
>>>Did anyone of you who got compile errors, get a fresh evaluation report?
>>>I haven't received anything yet.
>>>
>>>Dhawal Kumar
>>>


-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: more test script blues
Date: Fri, 8 Feb 2002 15:40:38 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3280

Its really frustrating to wait 1 hour for result and get things like
below. I don't know why things start to break under load and that too when
the deadline is near.

Dhawal Kumar

On 8 Feb 2002, Tudor Andrei Bosman wrote:

>  RESULTS FOR THE DIFFERENT TESTS:
>
>     * Test 2.A: Fails to compile
>   [out of 40]
>
> [SUMMARY] NOT OK
> [RESULT]  ERROR:  could not compile (512).
> -------
> umask 022
> gcc -g -ansi -DSOLARIS  -c client.c
> gmake: *** No rule to make target `transport.c', needed by `transport.o'.  Stop.
>
>
> Thanks,
> Tudor.
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Milestone C Issues
Date: Fri, 8 Feb 2002 17:06:56 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3281

I am running my transport.c with the provided ftpcopy object files and the
provided ftpd program and I have noticed the following behavior.  I was
wondering if anyone else has come accross the same issues:

Note: I grabbed the pre-built ftpd, ftpd.old, and debug_ftpd from
(/class/cs244a/homeworks/hw2/ftpd_src/src)

1.  The our.ftpcopy (running my transport layer) program does not work with
the provided ftpd at all

2.  The our.ftpcopy program (running my transport layer) works with the
debug_ftpd but only if they are running on the same machine.  It appears
that debug_ftpd assumes the IP address of the client is 127.0.0.1
(localhost).

3.  The our.ftpcopy program (running my transport layer) works with the
ftpd.old but only if they are running on the same machine.

Does anyone know what's going on or has had similar experiences?

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!fable17.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: Permission denied?
Date: Fri, 8 Feb 2002 18:25:14 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable17.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3282


Hi,

Whether I download the ftpd.o or run directly (from elaine machine)
:/afs/ir/class/cs244a/WWW/homeworks/hw2/ftpd_src/src/ftpd

I got error message : permission denied.

What is the probelm?

Thanks a lot!

Best,
-jinhui



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Leonard Sibille 
Newsgroups: su.class.cs244a
Subject: Re: Permission denied?
Date: Fri, 08 Feb 2002 18:42:58 -0800
Lines: 22
Distribution: su
Message-ID: 
References: 
Reply-To: 
NNTP-Posting-Host: saclay.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1013222899 22519 171.64.68.193 (9 Feb 2002 02:48:19 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.18 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3283

You can use the following command to copy all the files recursively in
your current directory:
wget -r www.stanford.edu/class/cs244a/homeworks/hw2/ftpd_src

Leo


Jinhui Pan wrote:
> 
> Hi,
> 
> Whether I download the ftpd.o or run directly (from elaine machine)
> :/afs/ir/class/cs244a/WWW/homeworks/hw2/ftpd_src/src/ftpd
> 
> I got error message : permission denied.
> 
> What is the probelm?
> 
> Thanks a lot!
> 
> Best,
> -jinhui
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Leonard Sibille 
Newsgroups: su.class.cs244a
Subject: Error with getsockname()
Date: Fri, 08 Feb 2002 19:05:45 -0800
Lines: 12
Distribution: su
Message-ID: 
Reply-To: 
NNTP-Posting-Host: saclay.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1013224267 22801 171.64.68.193 (9 Feb 2002 03:11:07 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.18 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3284

In accordance to the notes about milestone #3, I changed all socket
functions to their "my" counterparts. However, a call to getsockname()
just after myconnect() in the initial connection part of ftpcopy.c
returns an error. 

By the way, can we still use getsockname() in our program? Calling
mylocal_name() after myconnect() also returns an error. I am confused
about which one of these two functions we are supposed to use. 



Leo
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Test Script
Date: Fri, 8 Feb 2002 19:17:16 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3285

I tried running the Test Script for MileStone C.  I got back 3 emails (1 for
each milestone) but each one reported NOT TESTED for every category.

--Russ


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: About milestone 3
Date: 9 Feb 2002 03:39:43 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3286


>I have some questions on milestone C:

>(1) If the supplied ftpclient(object file)  runs well with STCP in
>milestone B, we actually do not need to make change to our own source
>code of ftpclient and submit it, right?

That's correct.

>(2) The delieverable files include: help.o and main.o for ftpclient,
>but I only fine libftpcopy.a, libftpd.a and libsupport.a in the
>provided dir. Are they the same thing?

They should be.  However, there seems to be a problem in the provided
ftpcopy code, in that it sends 127.0.0.1 as the local IP address rather
than the machine's actual IP address.  I'm looking at the problem, but
in the meantime you can either test with client and server running on the
same machine, or modify your ftpcopy code from assignment #1 (it takes
only a few minutes).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: About milestone 3
Date: 9 Feb 2002 03:40:54 GMT
Lines: 10
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3287


>The supplied object code does not support the -U option, is there something
>I did wrong? Also, it is not doing the right thing as it copies empty
>directory as well. So if we use the supplied object code to submit for
>milestone 3, will we get penalize for not passing the special case in PA1?

No, you would not be penalized for bugs in that code.

As for the -U option, I'll look at it when I try to find the other problem
mentioned.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Compile errors in test script
Date: 9 Feb 2002 05:19:27 GMT
Lines: 12
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3288


>	I don't understand why you can't run the tests in parallel? It 
>would really speed things up quite a bit and we wouldn't suffer excessive 
>waits immediately before the deadline.

You could avoid the problem of excessive waits before the deadline by
submitting before then, when there's less contention.

As far as running tests in parallel, this was addressed in a previous
post--there are indeed plans concerning this.  Unfortunately, such code
doesn't magically write itself.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: test server running again
Date: 9 Feb 2002 05:25:55 GMT
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3289


The server should be up again.

Please be gentle with it.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: test server running again
Date: Fri, 8 Feb 2002 22:03:56 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3290

The test script appears to require your.ftpcopy to be there or it won't run
at all.  This happens because I chose to use our.ftpcopy instead.

--Russ



RESULTS FOR THE DIFFERENT TESTS:

    * Test 3.A: Fails to compile
  [out of 10]


[SUMMARY] NOT OK
[RESULT]  ERROR:  No your.ftpcopy in
/afs/ir/class/cs244a/submissions/restricted/hw2.C/holliman/rdg12//compile.su
n.




"Matthew Jonathan Holliman"  wrote in message

>
> The server should be up again.
>
> Please be gentle with it.
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!ascwu
From: Alex Shao-chiun Wu 
Newsgroups: su.class.cs244a
Subject: Name of executables in Milestone III
Date: Fri, 8 Feb 2002 22:14:20 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3291

Dear TAs:

   The name of the executable should be "ftpcopy" or "your.ftpcopy"??

   Thanx!


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga1.Stanford.EDU!ascwu
From: Alex Shao-chiun Wu 
Newsgroups: su.class.cs244a
Subject: File not accepted when testing milestone 3
Date: Fri, 8 Feb 2002 22:33:28 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3292

On 9 Feb 2002, Matthew Jonathan Holliman wrote:

>
> >I have some questions on milestone C:
>
> >(1) If the supplied ftpclient(object file)  runs well with STCP in
> >milestone B, we actually do not need to make change to our own source
> >
> >(2) The delieverable files include: help.o and main.o for ftpclient,
> >but I only fine libftpcopy.a, libftpd.a and libsupport.a in the
> >provided dir. Are they the same thing?
>
> They should be.  However, there seems to be a problem in the provided
> ftpcopy code, in that it sends 127.0.0.1 as the local IP address rather
> than the machine's actual IP address.  I'm looking at the problem, but
> in the meantime you can either test with client and server running on the
> same machine, or modify your ftpcopy code from assignment #1 (it takes
> only a few minutes).
>
>

But when using the test script, it could not recognize those lib files and
thus those lib files are not submitted along with transport.c. So if they
are the same thing, can someone modify the test script so it can accept
those lib files as well??

Or does anyone know where the main.o or help.o really are??

Thanx!

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: test server running again
Date: 9 Feb 2002 06:56:52 GMT
Lines: 6
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3293


>The test script appears to require your.ftpcopy to be there or it won't run
>at all.  This happens because I chose to use our.ftpcopy instead.


Ok, should be fixed now to work with either of them.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Name of executables in Milestone III
Date: 9 Feb 2002 06:58:37 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3294



>   The name of the executable should be "ftpcopy" or "your.ftpcopy"??


The Makefile on the assignment page with the HW #1 object files is broken.
You should use the Makefile from the previous stages, appropriately
modified if you're building your own ftpcopy, so the executable should
be named your.ftpcopy (for yours) or our.ftpcopy (for the version built
with libftpcopy.a etc.)

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: File not accepted when testing milestone 3
Date: 9 Feb 2002 07:00:14 GMT
Lines: 10
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3295


>But when using the test script, it could not recognize those lib files and
>thus those lib files are not submitted along with transport.c. So if they
>are the same thing, can someone modify the test script so it can accept
>those lib files as well??


You shouldn't need to submit them--I could submit a version with "our.ftpcopy"
without including lib*.a.  The script seems to use its own copies of the
libraries during testing.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Error with getsockname()
Date: 9 Feb 2002 07:06:02 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3296


>In accordance to the notes about milestone #3, I changed all socket
>functions to their "my" counterparts. However, a call to getsockname()
>just after myconnect() in the initial connection part of ftpcopy.c
>returns an error. 

>By the way, can we still use getsockname() in our program? Calling
>mylocal_name() after myconnect() also returns an error. I am confused
>about which one of these two functions we are supposed to use. 


For obtaining the local IP address, there is a pointer in the
"miscellaneous notes and hints" section of the assignment.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: Compile errors in test script
Date: 9 Feb 2002 12:09:56 GMT
Lines: 31
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: elaine22.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3297

Matthew Jonathan Holliman  wrote:

>>	I don't understand why you can't run the tests in parallel? It 
>>would really speed things up quite a bit and we wouldn't suffer excessive 
>>waits immediately before the deadline.

> You could avoid the problem of excessive waits before the deadline by
> submitting before then, when there's less contention.

	I did submit fairly early for the previous assignment (hw2.A) and 
the professor decided to grant an extension for the previous assignment. I 
guess those extensions just aren't granted when I need them. :)

	I've also learned not to rely on those tests too much as if I do 
enough testing on my own, it seems that those most of those tests will 
eventually pass. (Of course I didn't get my grade for hw2.A/hw2.B yet) 

> As far as running tests in parallel, this was addressed in a previous
> post--there are indeed plans concerning this.  Unfortunately, such code
> doesn't magically write itself.

	Yeah, I understand the legacy of old code. I just thought it was 
some script that was running on epic6, and if so , why not run it on 
multiple machines? I thought it was as simple as that, but apparently it 
isn't. 

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic2.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Grading for HW2.C part.
Date: Sat, 9 Feb 2002 11:01:00 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3298

Hi,
If we use the supplied ftpclient and submit helper.o and main.o file with
the assignment then will it affect the grade with respect to if we
modified our own ftpclient and submitted the ftpclient.c file. Also what
if our code works fine with the supplied ftpclient but does not work with
our modified ftpclient code because of some bug in our ftpclient code.
Is the grading going to take into account the changes done to the
ftpclient and its correct working into account.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine29.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Compile errors in test script
Date: 9 Feb 2002 20:38:44 GMT
Lines: 18
Distribution: su
Message-ID: 
References:      
NNTP-Posting-Host: elaine29.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3299


>	I've also learned not to rely on those tests too much as if I do 
>enough testing on my own, it seems that those most of those tests will 
>eventually pass. (Of course I didn't get my grade for hw2.A/hw2.B yet) 

Yes, that's definitely the right approach to take.

>	Yeah, I understand the legacy of old code. I just thought it was 
>some script that was running on epic6, and if so , why not run it on 
>multiple machines? I thought it was as simple as that, but apparently it 
>isn't. 

Unfortunately there'd be race conditions if the same script were running on
multiple machines--these could probably be worked around, but the
infrastructure seems fragile enough (e.g. if programs run forever, consume
all disk space, etc.) that it makes more sense to fix the other problems
at the same time as well, rather than adopting an intermediate solution.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine29.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Grading for HW2.C part.
Date: 9 Feb 2002 20:45:47 GMT
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine29.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3300


>If we use the supplied ftpclient and submit helper.o and main.o file with
>the assignment then will it affect the grade with respect to if we
>modified our own ftpclient and submitted the ftpclient.c file. Also what
>if our code works fine with the supplied ftpclient but does not work with
>our modified ftpclient code because of some bug in our ftpclient code.
>Is the grading going to take into account the changes done to the
>ftpclient and its correct working into account.

As the assignment specifies you can choose either option.

If you have a bug in your ftpcopy code and you decide to use that, you should
fix it, or risk being penalized as specified in the grading guidelines for
milestone C.

If you don't fix the bug(s) and are sure that your ftp client does not behave
correctly, you should probably use the supplied libraries (subject to the
caveats posted previously).

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Jian Deng 
Newsgroups: su.class.cs244a
Subject: test server for HW2.C
Date: 9 Feb 2002 21:25:54 GMT
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3301

Hi,

I found and fixed a bug in  HW2.B, is there a way
I can run the testscript for HW2.B to verify my fix?
The HW2.C was running test A, B, and C before .
But now it was only to run C.

Thanks

Jian





.

Path: shelby.stanford.edu!nntp.stanford.edu!epic18.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Grading for HW2.C part.
Date: Sat, 9 Feb 2002 13:45:03 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic18.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Matthew Jonathan Holliman 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3302

Hi Jonathan,
But i still did not get my answer about grading difference. Will there be
any marks deducted if we use the supplied ftpcopy and don't use our
ftpcopy.
Thanks
Shankar

On 9 Feb 2002, Matthew Jonathan Holliman wrote:

>
> >If we use the supplied ftpclient and submit helper.o and main.o file with
> >the assignment then will it affect the grade with respect to if we
> >modified our own ftpclient and submitted the ftpclient.c file. Also what
> >if our code works fine with the supplied ftpclient but does not work with
> >our modified ftpclient code because of some bug in our ftpclient code.
> >Is the grading going to take into account the changes done to the
> >ftpclient and its correct working into account.
>
> As the assignment specifies you can choose either option.
>
> If you have a bug in your ftpcopy code and you decide to use that, you should
> fix it, or risk being penalized as specified in the grading guidelines for
> milestone C.
>
> If you don't fix the bug(s) and are sure that your ftp client does not behave
> correctly, you should probably use the supplied libraries (subject to the
> caveats posted previously).
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Milestone C Issues
Date: Sat, 9 Feb 2002 15:34:04 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3303

On Fri, 8 Feb 2002, Russell Greene wrote:

> I am running my transport.c with the provided ftpcopy object files and the
> provided ftpd program and I have noticed the following behavior.  I was
> wondering if anyone else has come accross the same issues:
>
> Note: I grabbed the pre-built ftpd, ftpd.old, and debug_ftpd from
> (/class/cs244a/homeworks/hw2/ftpd_src/src)
>
> 1.  The our.ftpcopy (running my transport layer) program does not work with
> the provided ftpd at all
>
> 2.  The our.ftpcopy program (running my transport layer) works with the
> debug_ftpd but only if they are running on the same machine.  It appears
> that debug_ftpd assumes the IP address of the client is 127.0.0.1
> (localhost).
>
> 3.  The our.ftpcopy program (running my transport layer) works with the
> ftpd.old but only if they are running on the same machine.
>
> Does anyone know what's going on or has had similar experiences?

Thy don't implement passive mode correctly, I suggest you use their
libraries or ftpcopy.

                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: hw2.C README
Date: Sat, 9 Feb 2002 16:30:54 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
   
 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3304


Is there anything special we need to place in the
README for part C?  All I really did was relink
to get ftpcopy working, so the design and implementation
descriptions in the past two README's are still valid.
Can we just write a quick synopsis of how we got
ftpcopy to wortk with STCP or should we restate our
design as we did in parts A and B? Thanks :-))

                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga14.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: hw2.c empty directories
Date: Sat, 9 Feb 2002 21:28:03 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3305

hi

Has the problem of empty directories being created by supplied ftpclient
solved?
I got only that error when I submitted last, using the supplied code.

thanks

Abhishek Das

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Mikhail Seregine 
Newsgroups: su.class.cs244a
Subject: Possible test script suggestion
Date: Sat, 09 Feb 2002 21:45:05 -0800
Lines: 25
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic28.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3306

Hi,

I just wanted to suggest a change to the test script: when you submit,
you are
told how many people are in the queue already, so you can have some
idea when to expect a response. This might be a simple way to reduce
stress
levels...

Additional things along the same line would be a "check_test_status"
command
that tells you where you are in the queue at some later time....

It is natural that the test scripts aren't perfect, especially since
this is
the first year they are in use; I think the above is a good way of
addressing
their flaws for now.

Thanks,

Mikhail
----------------
Mikhail Seregine

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: test scripts
Date: 10 Feb 2002 07:49:08 GMT
Organization: Stanford University, CA 94305, USA
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3307

Hi are the test scripts up? I summited a few hours ago.
Also the last time I tested it didn't test with the A and B scripts, 
can I request those to still be used.

Thank you,

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic19.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Grading for HW2.C part.
Date: Sun, 10 Feb 2002 03:38:12 -0800
Lines: 11
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic19.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3308

I have two questions:
(1) Where do I get ftpd. I get permission denied when trying to access
/afs/ir/class/cs244a/WWW/homeworks/hw2/ftpd_src/src/ftpd

and a segmentation fault when using
/afs/ir/class/cs244a/bin/ftpd

(2) How do I get helper.o and main.o from those lib files?

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Midterm review
Date: 10 Feb 2002 19:10:10 GMT
Organization: Stanford University, CA 94305, USA
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3309

Hi can we have the midterm review handout given out on friday on
the web please.


Thank you,

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine0.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: hw2.C README
Date: 10 Feb 2002 20:34:11 GMT
Lines: 15
Distribution: su
Message-ID: 
References:        
NNTP-Posting-Host: elaine0.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3310



>Is there anything special we need to place in the
>README for part C?  All I really did was relink
>to get ftpcopy working, so the design and implementation
>descriptions in the past two README's are still valid.
>Can we just write a quick synopsis of how we got
>ftpcopy to wortk with STCP or should we restate our
>design as we did in parts A and B? Thanks :-))


Not a lot is required for part C (just a sentence of two explaining whether
you used your ftpcopy or the supplied code would suffice), but please include
the design and implementation comments that you wrote for parts A and B.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine0.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: test scripts
Date: 10 Feb 2002 20:44:32 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine0.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3311

 (Mark Joseph Dolan) writes:

>Hi are the test scripts up? I summited a few hours ago.
>Also the last time I tested it didn't test with the A and B scripts, 
>can I request those to still be used.


I restarted the server a while ago--unfortunately, it tries to create
/tmp/studentname during execution, and if the student has created a directory
by that name on the machine on which it is running (elaine15 at the moment),
it dies due to a permission error.

I'll see about testing with the A and B scripts also.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine0.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: hw2.c empty directories
Date: 10 Feb 2002 20:46:49 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine0.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3312


>Has the problem of empty directories being created by supplied ftpclient
>solved?
>I got only that error when I submitted last, using the supplied code.

No, but you also need not worry about that problem since it's not in your
code.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine0.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Grading for HW2.C part.
Date: 10 Feb 2002 20:50:23 GMT
Lines: 16
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine0.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3313


>(1) Where do I get ftpd. I get permission denied when trying to access
>/afs/ir/class/cs244a/WWW/homeworks/hw2/ftpd_src/src/ftpd

As a workaround, you can use wget as posted previously.

>and a segmentation fault when using
>/afs/ir/class/cs244a/bin/ftpd

That's a regular TCP-talking ftpd (it was from assignment #1).

>(2) How do I get helper.o and main.o from those lib files?

You don't; use the Makefile included with the "support code".  You should
see targets 'your.ftpcopy' and 'our.ftpcopy'; executing 'make our.ftpcopy'
should link the libraries (libftpcopy.a) with your transport.o.
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic21.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Test for part C.
Date: Sun, 10 Feb 2002 15:41:09 -0800
Lines: 31
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3314

Hi,
I am seeing the following errors for the part C. But i have one doubt
about this. Can you please tell me the size of the files that are being
transferred for this test. It looks like the size is more than 4 GB. If
thats the case then it would create problem for STCP and our code would
definetly break. And it looks like from the test results that it is trying
for the file size greater than 4GB. Concrete sizes of the files would be
of great help.
Thanks
Shankar

 Test 3.C: Doesn't transfer all files but transfers at least one file
properly
        [out of 5]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy gz 2 
exit line said: ""
---
diff between expected tree and your tree:
1c1,3
< ./f1.gz 4294967295 0
---
> ./D6/f2.gz 4270565966 2
> ./D6/f3.gz 4103939428 2
> ./f1.gz 4219530715 2

---



.

Path: shelby.stanford.edu!nntp.stanford.edu!epic24.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: test scripts
Date: Sun, 10 Feb 2002 19:16:11 -0800
Lines: 24
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: epic24.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3320

I am also facing trouble with script response. I submitted about an hour
ago and no response. Yesterday I submitted at late night and got a
response only in the afternoon - a delay of about 10 hours.

Dhawal Kumar

On 10 Feb 2002, Matthew Jonathan Holliman wrote:

>  (Mark Joseph Dolan) writes:
>
> >Hi are the test scripts up? I summited a few hours ago.
> >Also the last time I tested it didn't test with the A and B scripts,
> >can I request those to still be used.
>
>
> I restarted the server a while ago--unfortunately, it tries to create
> /tmp/studentname during execution, and if the student has created a directory
> by that name on the machine on which it is running (elaine15 at the moment),
> it dies due to a permission error.
>
> I'll see about testing with the A and B scripts also.
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Yichen Xie" 
Newsgroups: su.class.cs244a
Subject: libftpcopy.a and PORT command
Date: Sun, 10 Feb 2002 21:33:30 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: dnab4046d7.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3321

I am using libftpcopy.a and everything seems to be working fine with the
client and server on the SAME machine. However, when I tried to test from
different machines (say, connecting from myth6 to epic6), debug_ftpd reveals
that our.ftpcopy sends PORT commands with IP set to 127.0.0.1. I highly
suspect it's caused by getsockname() (which should be replaced by
gethostname() and friends) in libftpcpoy.a. Could anyone confirm/disconfirm
this? In other words, did anyone have any success using libftpcopy.a to
connect to a server on a different machine? Thanks!


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic10.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Test scripts not working
Date: Sun, 10 Feb 2002 23:47:28 -0800
Lines: 33
Distribution: su
Message-ID: 
References:  
 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3322

I have now been waiting since 8 hours and still no response. Do I need to
resubmit?

Dhawal Kumar

On Sun, 10 Feb 2002, Dhawal Kumar wrote:

> I am also facing trouble with script response. I submitted about an hour
> ago and no response. Yesterday I submitted at late night and got a
> response only in the afternoon - a delay of about 10 hours.
>
> Dhawal Kumar
>
> On 10 Feb 2002, Matthew Jonathan Holliman wrote:
>
> >  (Mark Joseph Dolan) writes:
> >
> > >Hi are the test scripts up? I summited a few hours ago.
> > >Also the last time I tested it didn't test with the A and B scripts,
> > >can I request those to still be used.
> >
> >
> > I restarted the server a while ago--unfortunately, it tries to create
> > /tmp/studentname during execution, and if the student has created a directory
> > by that name on the machine on which it is running (elaine15 at the moment),
> > it dies due to a permission error.
> >
> > I'll see about testing with the A and B scripts also.
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine10.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: your.ftpcopy works ??
Date: Mon, 11 Feb 2002 02:12:39 -0800
Lines: 39
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3323


Hello,

wondering if anybody has got your.ftpcopy program working with
the supplied debug_ftpd. I have our.ftpcopy working, but
your.ftpcopy hangs at debug_ftpd side.

This happens after i do a LIST, and receive the list response on
the data connection of ftp. Afterwards the control channel seems
busted on the debug_ftpd side.

Appreciate your response.

Thanks, Arvind



pid 10076 is control channel
pid 10111 is for the data channel --list response

----------------------------------------------------------------
sock_io: (pid#10111) received an ACK packet, ack = 2590
send_packet: seq = 3078, len = 130, #Xmit = 0
sock_io: (pid#10111) received an ACK packet, ack = 3078
appl_io: connection close requested, state = TOSEND_FIN
sock_io: (pid#10111) received an ACK packet, ack = 3208
send_fin:
transport_sock_io: received a FIN_ACK packet
child (pid#10111):exiting...
sock_io: (pid#10076) receive a data packet, seq = 308, len = 5
send_ack: ACK = 313
----------------------------------------------------------------

The next command after LIST is acked by the ftpd transport.
But the application fails to give a response. In my case it
was "PWD" command being sent from the ftpcopy program. I have
verified that PWD works fine before LIST.


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: libftpcopy.a and PORT command
Date: 11 Feb 2002 11:01:39 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3324

Yes, to test your code versus the server you should run both on the same 
machine. This is perfectly acceptable for this part of the assignment.

Arun

Yichen Xie  wrote:
: I am using libftpcopy.a and everything seems to be working fine with the
: client and server on the SAME machine. However, when I tried to test from
: different machines (say, connecting from myth6 to epic6), debug_ftpd reveals
: that our.ftpcopy sends PORT commands with IP set to 127.0.0.1. I highly
: suspect it's caused by getsockname() (which should be replaced by
: gethostname() and friends) in libftpcpoy.a. Could anyone confirm/disconfirm
: this? In other words, did anyone have any success using libftpcopy.a to
: connect to a server on a different machine? Thanks!


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Midterm review
Date: 11 Feb 2002 11:27:33 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3325

The review handout has been posted. The answers were covered in the review 
session and will be available shortly.

Arun

Mark Joseph Dolan  wrote:
: Hi can we have the midterm review handout given out on friday on
: the web please.


: Thank you,

: -mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Leonard Sibille 
Newsgroups: su.class.cs244a
Subject: Re: Error with getsockname()
Date: Mon, 11 Feb 2002 11:12:57 -0800
Lines: 17
Distribution: su
Message-ID: 
References:  
Reply-To: 
NNTP-Posting-Host: saclay.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1013455072 20076 171.64.68.193 (11 Feb 2002 19:17:52 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.2.18 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3326

gethostname() only gives us the address of the local host. How do we
obtain the port number without using getsockname() ?

If I only use gethostname() as specified in the section, my ftp client
is unable to send any commands to the server.

Any suggestions?


Leo



Matthew Jonathan Holliman wrote:
> 
> For obtaining the local IP address, there is a pointer in the
> "miscellaneous notes and hints" section of the assignment.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Yichen Xie" 
Newsgroups: su.class.cs244a
Subject: Re: your.ftpcopy works ??
Date: Mon, 11 Feb 2002 11:38:15 -0800
Lines: 46
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dn800cb309.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3327

Same here. Took me a while to figure out that libftpcopy.a uses NLST instead
of LIST.

"Arvind Bhat"  wrote in message

>
> Hello,
>
> wondering if anybody has got your.ftpcopy program working with
> the supplied debug_ftpd. I have our.ftpcopy working, but
> your.ftpcopy hangs at debug_ftpd side.
>
> This happens after i do a LIST, and receive the list response on
> the data connection of ftp. Afterwards the control channel seems
> busted on the debug_ftpd side.
>
> Appreciate your response.
>
> Thanks, Arvind
>
>
>
> pid 10076 is control channel
> pid 10111 is for the data channel --list response
>
> ----------------------------------------------------------------
> sock_io: (pid#10111) received an ACK packet, ack = 2590
> send_packet: seq = 3078, len = 130, #Xmit = 0
> sock_io: (pid#10111) received an ACK packet, ack = 3078
> appl_io: connection close requested, state = TOSEND_FIN
> sock_io: (pid#10111) received an ACK packet, ack = 3208
> send_fin:
> transport_sock_io: received a FIN_ACK packet
> child (pid#10111):exiting...
> sock_io: (pid#10076) receive a data packet, seq = 308, len = 5
> send_ack: ACK = 313
> ----------------------------------------------------------------
>
> The next command after LIST is acked by the ftpd transport.
> But the application fails to give a response. In my case it
> was "PWD" command being sent from the ftpcopy program. I have
> verified that PWD works fine before LIST.
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Scripts up?
Date: 11 Feb 2002 20:32:41 GMT
Organization: Stanford University, CA 94305, USA
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3328

Hi are the test scripts up? I summited a test at 6am and have not heard
anything yet.

-mark

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: test server restarted
Date: 11 Feb 2002 20:46:46 GMT
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3329


The machine on which the server was running seems to be dead (or close
to it).  I've restarted the server again.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: your.ftpcopy works ??
Date: 11 Feb 2002 21:12:11 GMT
Lines: 51
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: myth7.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3330

Happens for me too, hanging after I send it "TYPE I". So are we supposed
to use NLST instead?


Yichen Xie  wrote:
: Same here. Took me a while to figure out that libftpcopy.a uses NLST instead
: of LIST.

: "Arvind Bhat"  wrote in message
: 
:>
:> Hello,
:>
:> wondering if anybody has got your.ftpcopy program working with
:> the supplied debug_ftpd. I have our.ftpcopy working, but
:> your.ftpcopy hangs at debug_ftpd side.
:>
:> This happens after i do a LIST, and receive the list response on
:> the data connection of ftp. Afterwards the control channel seems
:> busted on the debug_ftpd side.
:>
:> Appreciate your response.
:>
:> Thanks, Arvind
:>
:>
:>
:> pid 10076 is control channel
:> pid 10111 is for the data channel --list response
:>
:> ----------------------------------------------------------------
:> sock_io: (pid#10111) received an ACK packet, ack = 2590
:> send_packet: seq = 3078, len = 130, #Xmit = 0
:> sock_io: (pid#10111) received an ACK packet, ack = 3078
:> appl_io: connection close requested, state = TOSEND_FIN
:> sock_io: (pid#10111) received an ACK packet, ack = 3208
:> send_fin:
:> transport_sock_io: received a FIN_ACK packet
:> child (pid#10111):exiting...
:> sock_io: (pid#10076) receive a data packet, seq = 308, len = 5
:> send_ack: ACK = 313
:> ----------------------------------------------------------------
:>
:> The next command after LIST is acked by the ftpd transport.
:> But the application fails to give a response. In my case it
:> was "PWD" command being sent from the ftpcopy program. I have
:> verified that PWD works fine before LIST.
:>
:>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: Test script down again?
Date: Mon, 11 Feb 2002 13:20:07 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3331

I submit the test last night. But did not get any response till now.

Thanks,
-jinhui


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Failing test script
Date: 11 Feb 2002 21:52:45 GMT
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3332

I'm failing tests 3.A through 3.C, using our.ftpcopy. It works fine
when I run it myself, though, against both the public ftpd and the
one I compile with my STCP.

Can someone illuminate me on what:

  exit line said: "ERROR: Error 0 (-1) "

means in the test script?


Thanks!
Derrick
.

Path: shelby.stanford.edu!nntp.stanford.edu!myth1.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: your.ftpcopy works ??
Date: Mon, 11 Feb 2002 14:36:37 -0800
Lines: 68
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: myth1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Derrick Wen-Shiuan Tong 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3333


Was it hanging after the initial listing ?

NLST gives you the output in a different format, so you
cannot use it with the ftpcopy program from assignment 1
i think.

Bccing my TA also.

Thanks, Arvind


On 11 Feb 2002, Derrick Wen-Shiuan Tong wrote:

> Happens for me too, hanging after I send it "TYPE I". So are we supposed
> to use NLST instead?
>
>
> Yichen Xie  wrote:
> : Same here. Took me a while to figure out that libftpcopy.a uses NLST instead
> : of LIST.
>
> : "Arvind Bhat"  wrote in message
> : 
> :>
> :> Hello,
> :>
> :> wondering if anybody has got your.ftpcopy program working with
> :> the supplied debug_ftpd. I have our.ftpcopy working, but
> :> your.ftpcopy hangs at debug_ftpd side.
> :>
> :> This happens after i do a LIST, and receive the list response on
> :> the data connection of ftp. Afterwards the control channel seems
> :> busted on the debug_ftpd side.
> :>
> :> Appreciate your response.
> :>
> :> Thanks, Arvind
> :>
> :>
> :>
> :> pid 10076 is control channel
> :> pid 10111 is for the data channel --list response
> :>
> :> ----------------------------------------------------------------
> :> sock_io: (pid#10111) received an ACK packet, ack = 2590
> :> send_packet: seq = 3078, len = 130, #Xmit = 0
> :> sock_io: (pid#10111) received an ACK packet, ack = 3078
> :> appl_io: connection close requested, state = TOSEND_FIN
> :> sock_io: (pid#10111) received an ACK packet, ack = 3208
> :> send_fin:
> :> transport_sock_io: received a FIN_ACK packet
> :> child (pid#10111):exiting...
> :> sock_io: (pid#10076) receive a data packet, seq = 308, len = 5
> :> send_ack: ACK = 313
> :> ----------------------------------------------------------------
> :>
> :> The next command after LIST is acked by the ftpd transport.
> :> But the application fails to give a response. In my case it
> :> was "PWD" command being sent from the ftpcopy program. I have
> :> verified that PWD works fine before LIST.
> :>
> :>
>
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Failing test script
Date: Mon, 11 Feb 2002 14:36:47 -0800
Lines: 114
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3334

Exactly same error over here. I created the same directory structure as
mentioned in assignment1 (second example) and used our.ftpcopy to retrieve
files and that worked for me but the test is reporting
problems in 3.B and 3.C. Can the TAs point what file sizes and
directory structure are they using? FYI, earlier I had passed all the
tests.

One more request to the TAs to get the hw2A and hw2B testing script
included in the test for hw2C.

I am attaching the error for 3.B

    * Test 3.B: Doesn't transfer even a single file correctly
	[out of 10]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy gz 2 
exit line said: "ERROR: Error 0 (-1) "
---
Program output was:
elaine40.Stanford.EDU:43946
elaine40.Stanford.EDU:45437
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data from parent
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data from parent
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data from parent
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data from parent
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data from parent
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data from parent
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Timeout
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
Waiting for data sockets
Received data
Received data over network
ERROR: Error 0 (-1)



On 11 Feb 2002, Derrick Wen-Shiuan Tong wrote:

> I'm failing tests 3.A through 3.C, using our.ftpcopy. It works fine
> when I run it myself, though, against both the public ftpd and the
> one I compile with my STCP.
>
> Can someone illuminate me on what:
>
>   exit line said: "ERROR: Error 0 (-1) "
>
> means in the test script?
>
>
> Thanks!
> Derrick
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!rgustin
From: Reid Gustin 
Newsgroups: su.class.cs244a
Subject: Re: Failing test script
Date: Mon, 11 Feb 2002 15:26:55 -0800
Lines: 129
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3335

As near as I can tell from the errors I get back, the server looks to be
sending a FIN packet before waiting for ACK's for all outstanding data,
which is out of spec. I may be mistaken, but I don't think so.

The catch here is that if this is the case, there isn't a fix from the
client side, as the server is waiting for a FIN_ACK, and only a FIN_ACK,
and is certainly not expecting to have to send more data to the client.

Your mileage may vary,
 Reid

On Mon, 11 Feb 2002, Dhawal Kumar wrote:

> Exactly same error over here. I created the same directory structure as
> mentioned in assignment1 (second example) and used our.ftpcopy to retrieve
> files and that worked for me but the test is reporting
> problems in 3.B and 3.C. Can the TAs point what file sizes and
> directory structure are they using? FYI, earlier I had passed all the
> tests.
>
> One more request to the TAs to get the hw2A and hw2B testing script
> included in the test for hw2C.
>
> I am attaching the error for 3.B
>
>     * Test 3.B: Doesn't transfer even a single file correctly
> 	[out of 10]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy gz 2 
> exit line said: "ERROR: Error 0 (-1) "
> ---
> Program output was:
> elaine40.Stanford.EDU:43946
> elaine40.Stanford.EDU:45437
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data from parent
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data from parent
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data from parent
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data from parent
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data from parent
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data from parent
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Timeout
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> Waiting for data sockets
> Received data
> Received data over network
> ERROR: Error 0 (-1)
>
>
>
> On 11 Feb 2002, Derrick Wen-Shiuan Tong wrote:
>
> > I'm failing tests 3.A through 3.C, using our.ftpcopy. It works fine
> > when I run it myself, though, against both the public ftpd and the
> > one I compile with my STCP.
> >
> > Can someone illuminate me on what:
> >
> >   exit line said: "ERROR: Error 0 (-1) "
> >
> > means in the test script?
> >
> >
> > Thanks!
> > Derrick
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Review before the mid term review (by Antonious)
Date: Mon, 11 Feb 2002 15:55:20 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3336

This one still doesn't figure on the homepage.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: testing keeps failing..
Date: Mon, 11 Feb 2002 17:10:00 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3337


Hi,

Could someone please point out what is going wrong
when I try and test.. I keep getting the following error..


 RESULTS FOR THE DIFFERENT TESTS:

    * Test 2.A: Fails to compile
  [out of 40]

[SUMMARY] NOT OK
[RESULT]  ERROR:  No ftpd in /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/casado//compile.sun.


I'm not quite sure what I'm missing :-(

                  ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: testing keeps failing..
Date: 12 Feb 2002 01:23:23 GMT
Lines: 19
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3338



>Could someone please point out what is going wrong
>when I try and test.. I keep getting the following error..


> RESULTS FOR THE DIFFERENT TESTS:

>    * Test 2.A: Fails to compile
>  [out of 40]

>[SUMMARY] NOT OK
>[RESULT]  ERROR:  No ftpd in /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/casado//compile.sun.


The Makefile you submitted doesn't seem to have an 'ftpd' target.  If you
use the Makefile that was posted in the src directory, that should (I hope)
fix the problem.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: testing keeps failing..
Date: Mon, 11 Feb 2002 17:43:45 -0800
Lines: 50
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3339

Do we have to be able to build ftpd?

I am somewhat confident sure that the reason people (including me) have been
having problems with the test script (in failling 3.A-3.C) is that if you
use the Makefile from the /hw2/src directory it includes a target for ftpd.
However, if you make ftpd and try running the our.ftpcopy program you get
the
"ERROR: Error 0 (-1) " which Derrick asked about previously.  This leads me
to believe that the test script is using our Makefile to build the ftpd
which it uses to test our program.  (Can a TA verify this?)

I tried building ftpd with the Makefile that is included in the ftpd_src/src
directory but that didn't work either.  I therefore cannot build a working
version of ftpd.  My program runs perfectly with the pre-compiled version of
ftpd which is also located in ftpd_src/src.

Let me know if I am on the money with what I am saying as I am running into
all these same problems.

--Russ




"Matthew Jonathan Holliman"  wrote in message

>
>
> >Could someone please point out what is going wrong
> >when I try and test.. I keep getting the following error..
>
>
> > RESULTS FOR THE DIFFERENT TESTS:
>
> >    * Test 2.A: Fails to compile
> >  [out of 40]
>
> >[SUMMARY] NOT OK
> >[RESULT]  ERROR:  No ftpd in
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/casado//compile.su
n.
>
>
> The Makefile you submitted doesn't seem to have an 'ftpd' target.  If you
> use the Makefile that was posted in the src directory, that should (I
hope)
> fix the problem.
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: testing keeps failing..
Date: Mon, 11 Feb 2002 18:23:29 -0800
Lines: 62
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: elaine37.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3340

Thats absolutely correct - If I try to build ftpd and use it I get "ERROR:
Error 0 (-1)" and precompiled one works.

TAs please check the scripts or let us know if we are supposed to build
the ftpd as well.

Dhawal Kumar

On Mon, 11 Feb 2002, Russell Greene wrote:

> Do we have to be able to build ftpd?
>
> I am somewhat confident sure that the reason people (including me) have been
> having problems with the test script (in failling 3.A-3.C) is that if you
> use the Makefile from the /hw2/src directory it includes a target for ftpd.
> However, if you make ftpd and try running the our.ftpcopy program you get
> the
> "ERROR: Error 0 (-1) " which Derrick asked about previously.  This leads me
> to believe that the test script is using our Makefile to build the ftpd
> which it uses to test our program.  (Can a TA verify this?)
>
> I tried building ftpd with the Makefile that is included in the ftpd_src/src
> directory but that didn't work either.  I therefore cannot build a working
> version of ftpd.  My program runs perfectly with the pre-compiled version of
> ftpd which is also located in ftpd_src/src.
>
> Let me know if I am on the money with what I am saying as I am running into
> all these same problems.
>
> --Russ
>
>
>
>
> "Matthew Jonathan Holliman"  wrote in message
> 
> >
> >
> > >Could someone please point out what is going wrong
> > >when I try and test.. I keep getting the following error..
> >
> >
> > > RESULTS FOR THE DIFFERENT TESTS:
> >
> > >    * Test 2.A: Fails to compile
> > >  [out of 40]
> >
> > >[SUMMARY] NOT OK
> > >[RESULT]  ERROR:  No ftpd in
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/casado//compile.su
> n.
> >
> >
> > The Makefile you submitted doesn't seem to have an 'ftpd' target.  If you
> > use the Makefile that was posted in the src directory, that should (I
> hope)
> > fix the problem.
> >
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: testing keeps failing..
Date: 12 Feb 2002 02:30:44 GMT
Organization: Stanford University, CA 94305, USA
Lines: 4
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3341

I have the same results. What is interesting to me is that I used pass the
script. Perhaps something has changed?

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: our.ftpcopy vs your.ftpcopy for test script
Date: Mon, 11 Feb 2002 18:38:47 -0800
Lines: 65
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0100_01C1B32B.573B3C30"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3342

This is a multi-part message in MIME format.

------=_NextPart_000_0100_01C1B32B.573B3C30
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


I wonder how we can specify which of our.ftpcopy or your.ftpcopy is to =
be tested...
Is it like that our.ftpcopy is used if your.ftpcopy fails to compile,
and your.ftpcopy is used if it compiles?

And as far as I remember,
someone posted that our.ftpcopy had some errors regarding empty =
directories,
but my recent test result using our.ftpcopy says OK about it..
Is it corrected now? ( I'm just afraid that test script is somehow =
saying OK erroneously )..

Thanks
Hopil


------=_NextPart_000_0100_01C1B32B.573B3C30
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>I wonder how we can specify which of =
our.ftpcopy or=20
your.ftpcopy is to be tested...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Is it like that our.ftpcopy is used if =
your.ftpcopy=20
fails to compile,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>and your.ftpcopy is used if it=20
compiles?</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>And as far as I =
remember,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>someone posted that our.ftpcopy =
had some=20
errors regarding empty directories,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>but my recent test result using =
our.ftpcopy=20
says OK about it..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Is it corrected now? ( I'm just afraid =
that test=20
script is somehow saying OK erroneously )..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Thanks</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Hopil</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_0100_01C1B32B.573B3C30--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: your.ftpcopy works ??
Date: Mon, 11 Feb 2002 20:28:57 -0800
Lines: 84
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013488137 29277 128.12.133.48 (12 Feb 2002 04:28:57 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3343

I've gotten my ftpcopy to work with STCP. In addition to the changes
described on the class web page, I had to use NLST instead of LIST (since
the ftpd seems to die after a LIST command, as mentioned previously), and
since the result of NLST on the supplied ftpd is just the filenames and
directory names (with no other information) I had to rip out the ftpparse
part and just determine file vs. directory by trial and error.

However, it works.

Jonathan

"Arvind Bhat"  wrote in message

>
> Was it hanging after the initial listing ?
>
> NLST gives you the output in a different format, so you
> cannot use it with the ftpcopy program from assignment 1
> i think.
>
> Bccing my TA also.
>
> Thanks, Arvind
>
>
> On 11 Feb 2002, Derrick Wen-Shiuan Tong wrote:
>
> > Happens for me too, hanging after I send it "TYPE I". So are we supposed
> > to use NLST instead?
> >
> >
> > Yichen Xie  wrote:
> > : Same here. Took me a while to figure out that libftpcopy.a uses NLST
instead
> > : of LIST.
> >
> > : "Arvind Bhat"  wrote in message
> > : 
> > :>
> > :> Hello,
> > :>
> > :> wondering if anybody has got your.ftpcopy program working with
> > :> the supplied debug_ftpd. I have our.ftpcopy working, but
> > :> your.ftpcopy hangs at debug_ftpd side.
> > :>
> > :> This happens after i do a LIST, and receive the list response on
> > :> the data connection of ftp. Afterwards the control channel seems
> > :> busted on the debug_ftpd side.
> > :>
> > :> Appreciate your response.
> > :>
> > :> Thanks, Arvind
> > :>
> > :>
> > :>
> > :> pid 10076 is control channel
> > :> pid 10111 is for the data channel --list response
> > :>
> > :> ----------------------------------------------------------------
> > :> sock_io: (pid#10111) received an ACK packet, ack = 2590
> > :> send_packet: seq = 3078, len = 130, #Xmit = 0
> > :> sock_io: (pid#10111) received an ACK packet, ack = 3078
> > :> appl_io: connection close requested, state = TOSEND_FIN
> > :> sock_io: (pid#10111) received an ACK packet, ack = 3208
> > :> send_fin:
> > :> transport_sock_io: received a FIN_ACK packet
> > :> child (pid#10111):exiting...
> > :> sock_io: (pid#10076) receive a data packet, seq = 308, len = 5
> > :> send_ack: ACK = 313
> > :> ----------------------------------------------------------------
> > :>
> > :> The next command after LIST is acked by the ftpd transport.
> > :> But the application fails to give a response. In my case it
> > :> was "PWD" command being sent from the ftpcopy program. I have
> > :> verified that PWD works fine before LIST.
> > :>
> > :>
> >
> >
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga18.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: testing keeps failing..
Date: Mon, 11 Feb 2002 22:31:42 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
  
 
 
NNTP-Posting-Host: saga18.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3344

guys

Me too facing the same results. It was only yesterday when I had
everything running...not a single test used to fail. This is an unplesant
surprise....I hope the TAs give us a quick feedback on if things are
really fine with the scripts.

I checked with a couple of more guys...they are also facing the same
problem starting today.

thanks
abhishek


On 12 Feb 2002, Mark Joseph Dolan wrote:

> I have the same results. What is interesting to me is that I used pass the
> script. Perhaps something has changed?
>
> -mark
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Shyam Maddali 
Newsgroups: su.class.cs244a
Subject: peer mismatch errors!!
Date: Mon, 11 Feb 2002 23:13:10 -0800
Lines: 20
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3345

Hi,
    when i run testscript for hw2.C i get the peer mismatch errors 
but when i run it in my local environment i dont see this behaviour.
I am finding it hard to reproduce in my env.

the error message from the test is 
------------
Program output was:
myconnect: mismatch in peer names  ("0.0.0.0:33836" != "0.0.0.0:0")
ERROR: Bad file number (-1) 
TRANSPORT: local socket pair success! 
------------

it seems that network_peer_name is not filling out the peer address 
but i dont know why.

Any clue why this would occur will be really helpful.

Thanks
Lakshman Shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: test script?
Date: Tue, 12 Feb 2002 00:36:18 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3346


Hi,

I can  run well with ftpcopy  locally. But can not pass any
test script test in  milestone 3. Is there any problem with test
script?

Thanks,
-jinhui




.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: "ERROR: Error 0 (-1) "
Date: 12 Feb 2002 09:15:21 GMT
Organization: Stanford University, CA 94305, USA
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3347

Hi,

I have been having problems running our.ftpcopy with ftpd built with my
transport code, and it seems to me that others have also. 
After testing I found I was getting some crazy strings coming from the
application socket which lead me to think I have some memory
problems. After using purify on our.ftpcopy I found some memory leaks in
libftpd, no big deal, however when using purify on ftpd complied with my
transport code I am finding uninit memory reads, which is what I was
expecting(and hoping!), but purify is showing them to be in libftpd and
not transport.c.

I am pretty new to purify so I was wondering if I was making a mistake
about purify (ie the memory read could have occured in transport and not
show that in the purify output) or if anyone has noticed anything simular.

Because not everyone is having this problem I find it strange that my
problem is in libftpd but I am unsure how else to explain my results.

Thank you,

-mark

Any suggestions, responses would be greatly appreciated.
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga15.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: Yes!! "ERROR: Error 0 (-1) "
Date: Tue, 12 Feb 2002 01:28:16 -0800
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3348



  I get the same error when I run our.ftpcopy. I have tested my
transport.c in the previous milestone and found no memory leakage. I don't
know why I get this error message and what I shall do to solve the
problem. Any helps will be appreciated greatly.

  Thanks
  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Error 0 (-1) Demystified
Date: Tue, 12 Feb 2002 01:36:41 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013506602 1700 128.12.133.48 (12 Feb 2002 09:36:42 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3349

Hey all,
    I've managed to reproduce and debug this one, and here's what it is.
(This has been verified by two of my friends who were having the same
problem.)

    When the ftpd has a data connection open, there are three processes: the
parent process, and two child processes, one for the control connection and
one for the data connection. Somehow, the data child's stdout is redirected
to the control connection's local data socket. Or at least I think that's
how it happens. Either way, the effect is that any prinf's in the data child
are sent across the control connection, thus thoroughly confusing the
client.

    The fix, obviously, is to remove -DDEBUG from the Makefile and make sure
you don't have any printf's lingering in your transport.c.

    It sounds farfetched, but hold your flames until you've tried it.

Jonathan


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: Error 0 (-1) Demystified
Date: Tue, 12 Feb 2002 01:39:40 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: fable2.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3350

Just wanted to confirm Jonathan's post.

I think we owe Jonathan a big thank you on this one.

Clayton


Jonathan Keljo wrote:
> 
> Hey all,
>     I've managed to reproduce and debug this one, and here's what it is.
> (This has been verified by two of my friends who were having the same
> problem.)
> 
>     When the ftpd has a data connection open, there are three processes: the
> parent process, and two child processes, one for the control connection and
> one for the data connection. Somehow, the data child's stdout is redirected
> to the control connection's local data socket. Or at least I think that's
> how it happens. Either way, the effect is that any prinf's in the data child
> are sent across the control connection, thus thoroughly confusing the
> client.
> 
>     The fix, obviously, is to remove -DDEBUG from the Makefile and make sure
> you don't have any printf's lingering in your transport.c.
> 
>     It sounds farfetched, but hold your flames until you've tried it.
> 
> Jonathan
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: Error 0 (-1) Demystified
Date: 12 Feb 2002 09:47:45 GMT
Organization: Stanford University, CA 94305, USA
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3351

	
That is close to what I found a few hours ago, I was able to fix my
problem by removing dprintf statements in transport_init but I figured it 
was a symptom of memory problems.

-mark




.

Path: shelby.stanford.edu!nntp.stanford.edu!myth2.Stanford.EDU!rgustin
From: Reid Gustin 
Newsgroups: su.class.cs244a
Subject: Re: Error 0 (-1) Demystified
Date: Tue, 12 Feb 2002 02:09:40 -0800
Lines: 60
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3352

On going through the code, Jonathan and I found what we think is likely
the problem. The catch is that the person who wrote ftpd tried to take a
shortcut with his file descriptors (which, while cute, means he's headin'
right for a kick in the groin if I ever meet him).

Here's the deal. In ftpd.c, there are three lines of code that look like
this:

 dup2(fd, 0);
 close(fd);
 dup2(0, 1);

Which says make 0 (STDIN) point at fd, and then make 1 (STDOUT) point at
0. So essentially, he aliased stdin and stdout to both point at the
socket. This allows the programmer to play a neat trick: he can now use
printf to print to his socket. Which is nice, cause he doesn't need to
suck it up and use sprintf to create his formatted output to the socket.

It's bad news because if you should have printf's in your code to do
debugging, they get printed to the socket, which means you're just going
to see them again on the client side, which screws up your data
connections, and doesn't give you any debugging output.

So, huge props to Jonathan for finding the bug and then sitting on the
phone while we both waded through the ftpd code. In the meantime, try
commenting out your printf's, and see if you can make the code run.

If this isn't the issue, if one of the TA's could let us know, we'd
appreciate it. I will honestly be really surprised, as my code ran just
right as soon as I took the printf's out, and purify was happy both before
and after.

Reid


On Tue, 12 Feb 2002, Jonathan Keljo wrote:

> Hey all,
>     I've managed to reproduce and debug this one, and here's what it is.
> (This has been verified by two of my friends who were having the same
> problem.)
>
>     When the ftpd has a data connection open, there are three processes: the
> parent process, and two child processes, one for the control connection and
> one for the data connection. Somehow, the data child's stdout is redirected
> to the control connection's local data socket. Or at least I think that's
> how it happens. Either way, the effect is that any prinf's in the data child
> are sent across the control connection, thus thoroughly confusing the
> client.
>
>     The fix, obviously, is to remove -DDEBUG from the Makefile and make sure
> you don't have any printf's lingering in your transport.c.
>
>     It sounds farfetched, but hold your flames until you've tried it.
>
> Jonathan
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic11.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: test script
Date: Tue, 12 Feb 2002 07:43:38 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3353

Hi,
Is the test script down. Because i submitted yesterday night and also
today morning but i have not got any results back for either yet.
Thanks
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine2.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: test script
Date: 12 Feb 2002 16:36:20 GMT
Lines: 5
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine2.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3354


>Is the test script down. Because i submitted yesterday night and also
>today morning but i have not got any results back for either yet.

Yes, it's just been restarted.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Russell Greene" 
Newsgroups: su.class.cs244a
Subject: Re: Error 0 (-1) Demystified
Date: Tue, 12 Feb 2002 10:02:03 -0800
Lines: 37
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: russell2.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3355

Amazing Jonathan!  However, that appeared to only fix the ftpd in reliable
mode.  It still messes up when I run it with the -U flag.  Is that not
supported or am I missing something else?  Thanks!

--Russ


"Jonathan Keljo"  wrote in message

> Hey all,
>     I've managed to reproduce and debug this one, and here's what it is.
> (This has been verified by two of my friends who were having the same
> problem.)
>
>     When the ftpd has a data connection open, there are three processes:
the
> parent process, and two child processes, one for the control connection
and
> one for the data connection. Somehow, the data child's stdout is
redirected
> to the control connection's local data socket. Or at least I think that's
> how it happens. Either way, the effect is that any prinf's in the data
child
> are sent across the control connection, thus thoroughly confusing the
> client.
>
>     The fix, obviously, is to remove -DDEBUG from the Makefile and make
sure
> you don't have any printf's lingering in your transport.c.
>
>     It sounds farfetched, but hold your flames until you've tried it.
>
> Jonathan
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: More ERROR: Error 0 (-1)
Date: 12 Feb 2002 18:59:41 GMT
Organization: Stanford University, CA 94305, USA
Lines: 22
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3356

First I want to thank Jonathan for finding that bug, but I am also finding
that the fix does not work in unreliable mode.
After removing all the printfs from transport and submiting I get a new
error 

myconnect: mismatch in peer names

When running in unreliable mode this is also the error I get in my testing
and this is also the error when I run ftpd in reliable mode though purify.

Regardless of how I run ftpd I am still getting 4 uninit memory reads, one
of which in in myconnect, which explains to me why I get the above error.

Has anyone else seen this also? or have any explainations to why this
might be?


Thank you,

-mark


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth1.Stanford.EDU!cpjones
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: More ERROR: Error 0 (-1)
Date: Tue, 12 Feb 2002 11:20:14 -0800
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3357

I suspect the problem is coming from your print_cong_wind_and_estRTT
function.  If you are using printf... instead use dprintf...

This means that nothing will print (if you took out the -DDebug... as
mentioned by Jonathan in his previous post), but it will allow things to
work in unreliable mode.  See Reid's earlier posting about why printing to
stdout causes an error.

Clayton

On 12 Feb 2002, Mark Joseph Dolan wrote:

> First I want to thank Jonathan for finding that bug, but I am also finding
> that the fix does not work in unreliable mode.
> After removing all the printfs from transport and submiting I get a new
> error
>
> myconnect: mismatch in peer names
>
> When running in unreliable mode this is also the error I get in my testing
> and this is also the error when I run ftpd in reliable mode though purify.
>
> Regardless of how I run ftpd I am still getting 4 uninit memory reads, one
> of which in in myconnect, which explains to me why I get the above error.
>
> Has anyone else seen this also? or have any explainations to why this
> might be?
>
>
> Thank you,
>
> -mark
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: More ERROR: Error 0 (-1)
Date: 12 Feb 2002 19:44:07 GMT
Organization: Stanford University, CA 94305, USA
Lines: 5
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: myth4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3358

No, I took care of  print_cong_wind_and_estRTT.

I have no printfs left in my code. :-(

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!saga15.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: ERROR (-2) !!!
Date: Tue, 12 Feb 2002 11:57:52 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3359



   Thanks a lot to Jonathan for making clear the ERROR (-1) problem, but
this time, I get ERROR (-2) problem when I run ftpd in unreliable mode
(-U). Do we have some kind of similar problem in this mode? Any comments?

   Also, I can't use -U with our.ftpcopy. Althout it's usage includes a -U
flag, but it does not let me to use it and prints its usage again when I
use it. Any explanations?

  Yours,
  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Shyam Maddali 
Newsgroups: su.class.cs244a
Subject: Re: ERROR (-2) !!!
Date: Tue, 12 Feb 2002 12:05:36 -0800
Lines: 37
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic16.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3360

I also get the same error ERROR (-2) from the testscript printing out
assertion 
-----------
Program output was:
transport.c:1234: failed assertion `pl->minSequenceNum ==
pl->start->pkt.hdr.tcp_hdr.th_seq'
ERROR:  (-2) 
rver -p
/afs/ir/class/cs244a/submissions/restricted/hw2.C/hondroul/aag//compile.sun/S1
'  [pid: 27041]
------------------- 
Program output was:
myconnect: mismatch in peer names  ("0.0.0.0:39137" != "0.0.0.0:0")
ERROR: Bad file number (-1) 
-------------------------

After removing the printf from the code as suggested by jonathan ,
everything works
fine locally but fails the test script with the above stated errors.

Any clues on whats happening

~~~
Lakshman Maddali

Amin Firoozshahian wrote:
> 
>    Thanks a lot to Jonathan for making clear the ERROR (-1) problem, but
> this time, I get ERROR (-2) problem when I run ftpd in unreliable mode
> (-U). Do we have some kind of similar problem in this mode? Any comments?
> 
>    Also, I can't use -U with our.ftpcopy. Althout it's usage includes a -U
> flag, but it does not let me to use it and prints its usage again when I
> use it. Any explanations?
> 
>   Yours,
>   Amin
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: More ERROR: Error 0 (-1)
Date: Tue, 12 Feb 2002 12:40:01 -0800
Lines: 19
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013546402 6481 128.12.133.48 (12 Feb 2002 20:40:02 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3361

You also might want to make sure you did a clean make after
changing -DDEBUG. The makefile is a little stupid about that, and the
network.c will do dprintfs every time it drops or otherwise messes with a
packet in unreliable mode.

You should only see the one line "ftpd wrapper running on..." from the ftpd;
any other lines of output indicate a potential problem.

Jonathan

"Mark Joseph Dolan"  wrote in message

> No, I took care of  print_cong_wind_and_estRTT.
>
> I have no printfs left in my code. :-(
>
> -mark


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Anyone pass yet?
Date: Tue, 12 Feb 2002 12:44:11 -0800
Lines: 229
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:0.9.8) Gecko/20020204
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3362

I've seen some semi-conflicting info in various places, so I wanted to 
verify that the following procedure is supposed to work when using 
our.ftpcopy (from anyone who's passed the tests and is happy with the 
general direction of his/her life):

1. Get libftpcopy.a, libftpd.a, and libsupport.a from 
/class/cs244a/homeworks/hw2/hw1obj

2. Edit the project 2 Makefile s.t. 'all' now includes 'ftpd' and 
'our.ftpcopy'.

3. Remove printf statements from transport.c; in particular, don't 
forget the print_congestion_window function.

4. make clean; make all

5. Run 'ftpd -U' somewhere.  Run 'our.ftpcopy -U' somewhere.

6. See that things work.

I just got our.ftpcopy -U to work; the key is to not specify the srcdir 
  arg.  And it seems to run fine (except for some spurious directory 
creation that I believe is a flaw in their ftp client).  I submitted my 
code for testing yesterday night without the print_congestion_window 
modification.  I failed 3.b and 3.c tests, but not with ERROR(-1) or 
ERROR(-2) messages as others have reported.  I get the following:

  * Test 3.B: Doesn't transfer even a single file correctly
	[out of 10]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy gz 2 
exit line said: "ERROR:  "
---
Program output was:

ERROR:
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
Nothing be read.
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
The initial seqence num is 29
handling the syn or syn_ack
CWND: 3072; ERTT: 200
CWND: 3072; ERTT: 160
All transmissions are complete. Exiting successfully.

---


---




===============

     * Test 3.C: Doesn't transfer all files but transfers at least one 
file properly
	[out of 5]

[SUMMARY] NOT OK
[RESULT]  test used: ftpcopy gz 2 
exit line said: "ERROR:  "
---
diff between expected tree and your tree:
0a1,3

 > ./D6/f2.gz 4270565966 2
 > ./D6/f3.gz 4103939428 2
 > ./f1.gz 4219530715 2


---
Program output was:

ERROR:
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
Nothing be read.
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
The initial seqence num is 29
handling the syn or syn_ack
CWND: 3072; ERTT: 200
CWND: 3072; ERTT: 160
All transmissions are complete. Exiting successfully.

---


---

exit line said: "ERROR:  "
---
diff between expected tree and your tree:
0a1

 > ./f1.gz


---
Program output was:

ERROR:
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
Nothing be read.
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
The initial seqence num is 29
handling the syn or syn_ack
CWND: 3072; ERTT: 200
CWND: 3072; ERTT: 160
All transmissions are complete. Exiting successfully.

---


---
test used: ftpcopy gz 2 
exit line said: "ERROR:  "
---
diff between expected tree and your tree:
0a1,3

 > ./D6/f2.gz
 > ./D6/f3.gz
 > ./f1.gz


---
Program output was:

ERROR:
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
Nothing be read.
serverInfo is 33399
in ParseServerString-- serverName is elaine40.Stanford.EDU
The port is 33399
serverStr at the end of parse fn...is elaine40.Stanford.EDU
serverPort at exit of Parse is 33399
before chdir is 
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU
The command line is gz 2  
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.sun/download/elaine40.Stanford.EDU 
..
The initial seqence num is 29
handling the syn or syn_ack
CWND: 3072; ERTT: 200
CWND: 3072; ERTT: 160
All transmissions are complete. Exiting successfully.

---


---



It seems that test 3.C actually consists of multiple invocations of the 
program, and thus lots of output.  Anyway, since I haven't been able to 
get this thing to fail for me, I'm going to make the 
print_congestion_window change and re-submit for testing, hoping for the 
best.  But if anyone out there who's gotten all the tests to pass sees a 
flaw in my steps 1-6 above, please reply to this post with the proper 
corrections.  If this next test run ends up passing for me, I'll reply 
to validate the above procedure.

-Sandeep

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: ERROR (-2) !!!
Date: Tue, 12 Feb 2002 12:45:53 -0800
Lines: 58
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013546754 6608 128.12.133.48 (12 Feb 2002 20:45:54 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3363

I have seen this bad file number stuff as well. Personally I think the test
script is somehow on crack, as the error suddenly cropped up for code that
was previously working. I think this may be one of those cases where, if you
can create a tree just like theirs locally (it seems they're using the one
from the HW1 handout, with all 2-byte files), and have it work, then forget
the script.

As to the previously mentioned thing about not being able to use -U with
our.ftpcopy, make sure it's the first argument and it's capitalized. Works
fine for me.

Jonathan

"Lakshman Shyam Maddali"  wrote in message

> I also get the same error ERROR (-2) from the testscript printing out
> assertion
> -----------
> Program output was:
> transport.c:1234: failed assertion `pl->minSequenceNum ==
> pl->start->pkt.hdr.tcp_hdr.th_seq'
> ERROR:  (-2)
> rver -p
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/hondroul/aag//compile.sun/
S1
> '  [pid: 27041]
> -------------------
> Program output was:
> myconnect: mismatch in peer names  ("0.0.0.0:39137" != "0.0.0.0:0")
> ERROR: Bad file number (-1)
> -------------------------
>
> After removing the printf from the code as suggested by jonathan ,
> everything works
> fine locally but fails the test script with the above stated errors.
>
> Any clues on whats happening
>
> ~~~
> Lakshman Maddali
>
> Amin Firoozshahian wrote:
> >
> >    Thanks a lot to Jonathan for making clear the ERROR (-1) problem, but
> > this time, I get ERROR (-2) problem when I run ftpd in unreliable mode
> > (-U). Do we have some kind of similar problem in this mode? Any
comments?
> >
> >    Also, I can't use -U with our.ftpcopy. Althout it's usage includes
a -U
> > flag, but it does not let me to use it and prints its usage again when I
> > use it. Any explanations?
> >
> >   Yours,
> >   Amin


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: More ERROR: Error 0 (-1)
Date: 12 Feb 2002 20:48:13 GMT
Organization: Stanford University, CA 94305, USA
Lines: 15
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: epic4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3364


Thanks Jonathan but that was first thing I did.

"ftpd wrapper running on..."

is the only line I get from ftpd until I get my new favorite line

"myconnect: mismatch in peer names "

The problem now is trying to bebug without stdout,
ofcourse not sure what else to debug.

Thanks again,

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: ERROR: Bad file number (-1)
Date: Tue, 12 Feb 2002 12:51:29 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3365

Hey folks,

My "our.ftpclient" running against "ftpd.old" on the same machine seems
to
work fine, ( is this the only good testing combo at this point? )

But I'm failing a few tests in the part C script with the 
following errors:

[SUMMARY] NOT OK
[RESULT] test used: ftpcopy gz 2 
exit line said: "ERROR: Bad file number (-1) "
---
Program output was:
elaine40.Stanford.EDU:50716
elaine40.Stanford.EDU:43704
ERROR: Bad file number (-1)

Any ideas what this might be?

Thanks
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: Re: Anyone pass yet?
Date: 12 Feb 2002 20:57:13 GMT
Organization: Stanford University, CA 94305, USA
Lines: 4
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3366


Well I agree not to include any dir args, mine does not like that either.

-mark
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Vasco Chatalbashev" 
Newsgroups: su.class.cs244a
Subject: Re: Anyone pass yet?
Date: Tue, 12 Feb 2002 13:03:06 -0800
Lines: 279
Distribution: su
Message-ID: 
References: 
Reply-To: "Vasco Chatalbashev" 
NNTP-Posting-Host: vassil.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3367

Is your SUNetID "wjiang"? The path of the terminated process seems to be for
"wjiang."
I had the same problem for a while where the test wouldnt even test my
submission but
return various failures which displayed paths to other students' programs
(the path printed
when the process is killed.)
Anyway I have made no changes but somehow pass all the tests. I e-mailed the
TAs a couple of
times but got no response, so I am not sure if the script actually test
another student's program, but it looks
like that's what may be going on.


"Sandeep Tamhankar"  wrote in message

> I've seen some semi-conflicting info in various places, so I wanted to
> verify that the following procedure is supposed to work when using
> our.ftpcopy (from anyone who's passed the tests and is happy with the
> general direction of his/her life):
>
> 1. Get libftpcopy.a, libftpd.a, and libsupport.a from
> /class/cs244a/homeworks/hw2/hw1obj
>
> 2. Edit the project 2 Makefile s.t. 'all' now includes 'ftpd' and
> 'our.ftpcopy'.
>
> 3. Remove printf statements from transport.c; in particular, don't
> forget the print_congestion_window function.
>
> 4. make clean; make all
>
> 5. Run 'ftpd -U' somewhere.  Run 'our.ftpcopy -U' somewhere.
>
> 6. See that things work.
>
> I just got our.ftpcopy -U to work; the key is to not specify the srcdir
>   arg.  And it seems to run fine (except for some spurious directory
> creation that I believe is a flaw in their ftp client).  I submitted my
> code for testing yesterday night without the print_congestion_window
> modification.  I failed 3.b and 3.c tests, but not with ERROR(-1) or
> ERROR(-2) messages as others have reported.  I get the following:
>
>   * Test 3.B: Doesn't transfer even a single file correctly
> [out of 10]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy gz 2 
> exit line said: "ERROR:  "
> ---
> Program output was:
>
> ERROR:
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> Nothing be read.
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> The initial seqence num is 29
> handling the syn or syn_ack
> CWND: 3072; ERTT: 200
> CWND: 3072; ERTT: 160
> All transmissions are complete. Exiting successfully.
>
> ---
>
>
> ---
>
>
>
>
> ===============
>
>      * Test 3.C: Doesn't transfer all files but transfers at least one
> file properly
> [out of 5]
>
> [SUMMARY] NOT OK
> [RESULT]  test used: ftpcopy gz 2 
> exit line said: "ERROR:  "
> ---
> diff between expected tree and your tree:
> 0a1,3
>
>  > ./D6/f2.gz 4270565966 2
>  > ./D6/f3.gz 4103939428 2
>  > ./f1.gz 4219530715 2
>
>
> ---
> Program output was:
>
> ERROR:
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> Nothing be read.
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> The initial seqence num is 29
> handling the syn or syn_ack
> CWND: 3072; ERTT: 200
> CWND: 3072; ERTT: 160
> All transmissions are complete. Exiting successfully.
>
> ---
>
>
> ---
>
> exit line said: "ERROR:  "
> ---
> diff between expected tree and your tree:
> 0a1
>
>  > ./f1.gz
>
>
> ---
> Program output was:
>
> ERROR:
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> Nothing be read.
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> The initial seqence num is 29
> handling the syn or syn_ack
> CWND: 3072; ERTT: 200
> CWND: 3072; ERTT: 160
> All transmissions are complete. Exiting successfully.
>
> ---
>
>
> ---
> test used: ftpcopy gz 2 
> exit line said: "ERROR:  "
> ---
> diff between expected tree and your tree:
> 0a1,3
>
>  > ./D6/f2.gz
>  > ./D6/f3.gz
>  > ./f1.gz
>
>
> ---
> Program output was:
>
> ERROR:
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> Nothing be read.
> serverInfo is 33399
> in ParseServerString-- serverName is elaine40.Stanford.EDU
> The port is 33399
> serverStr at the end of parse fn...is elaine40.Stanford.EDU
> serverPort at exit of Parse is 33399
> before chdir is
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> The command line is gz 2 
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
n/download/elaine40.Stanford.EDU
> .
> The initial seqence num is 29
> handling the syn or syn_ack
> CWND: 3072; ERTT: 200
> CWND: 3072; ERTT: 160
> All transmissions are complete. Exiting successfully.
>
> ---
>
>
> ---
>
>
>
> It seems that test 3.C actually consists of multiple invocations of the
> program, and thus lots of output.  Anyway, since I haven't been able to
> get this thing to fail for me, I'm going to make the
> print_congestion_window change and re-submit for testing, hoping for the
> best.  But if anyone out there who's gotten all the tests to pass sees a
> flaw in my steps 1-6 above, please reply to this post with the proper
> corrections.  If this next test run ends up passing for me, I'll reply
> to validate the above procedure.
>
> -Sandeep
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: More ERROR: Error 0 (-1)
Date: Tue, 12 Feb 2002 13:10:59 -0800
Lines: 48
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013548260 6801 128.12.133.48 (12 Feb 2002 21:11:00 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3368

There's a couple of ways to debug this crap without stdout. You can enable
full printouts in the ftpcopy client and it will work fine, so if you can
debug from that end you're golden (try printing all the data in a packet
when it's received...if it is a stray printf somewhere it will show up
immediately that way). You can also hack up transport.c to UN-redirect
stdout.

You'd do that as follows:
1. Somewhere at the top of transport_init,  run dup(1) and save off the
result. ftpd calls transport_init once before redirecting stdout, so this
saves the stdout descriptor before ftpd moves it. (Make sure you only do
this the first time transport_init is called, otherwise you'll be saving the
descriptor AFTER it's moved.)

2. Somewhere in the child process in transport_init, run dup2(fid,1), where
fid is the global variable you stored the saved stdout descriptor in.

3. Comment out any dprintf's in the parent process part of transport_init
and transport_close.

This should redirect stdout for all child processes back to stdout, but
leave it redirected to the socket for the parent so that ftpd will stay
happy. (This is why you've gotta make sure the parent isn't doing any
dprintf's in transport_init or transport_close.) Once you've done this, you
should be able to turn -DDEBUG back on and go nuts with printf within the
child processes.

Jonathan

"Mark Joseph Dolan"  wrote in message

>
> Thanks Jonathan but that was first thing I did.
>
> "ftpd wrapper running on..."
>
> is the only line I get from ftpd until I get my new favorite line
>
> "myconnect: mismatch in peer names "
>
> The problem now is trying to bebug without stdout,
> ofcourse not sure what else to debug.
>
> Thanks again,
>
> -mark


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic4.Stanford.EDU!mdolan
From:  (Mark Joseph Dolan)
Newsgroups: su.class.cs244a
Subject: suggestion
Date: 12 Feb 2002 21:20:52 GMT
Organization: Stanford University, CA 94305, USA
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic4.stanford.edu
Xref: nntp.stanford.edu su.class.cs244a:3369


Hi everyone,

after reading Sandeep's post i desided to start from scratch,

ie 

make new directory, 
isssue /afs/ir/class/cs244a/WWW/homeworks/hw2/gethw2
cp my tranport.c into new directory
edit Makefile (lose the -DDEBUG)

and magic my ftpd works with our.ftpcopy with and without -U

however...now the script says

Died at /afs/ir/class/cs244a/bin/submit_test.pl line 198.

just an idea,

-mark

-thanks Jonathan for the post on debugging 
.

Path: shelby.stanford.edu!nntp.stanford.edu!myth1.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: are the test scripts working ?
Date: Tue, 12 Feb 2002 13:56:34 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3370


i submited 2 hours back and havent heard since then.

Thanks, Arvind

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga15.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: are the test scripts working ?
Date: 13 Feb 2002 00:30:02 GMT
Lines: 4
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga15.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3371


>i submited 2 hours back and havent heard since then.

Dutifully restarted.
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Bob McGrew" 
Newsgroups: su.class.cs244a
Subject: Re: are the test scripts working ?
Date: Tue, 12 Feb 2002 16:37:36 -0800
Lines: 14
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: zarathustra.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3372

    If I submitted one at 1:30 or so before the test server was restarted,
does that mean that I need to resubmit? I tried to, but it told me that my
submission was still in the queue.

Bob

"Matthew Jonathan Holliman"  wrote in message

>
> >i submited 2 hours back and havent heard since then.
>
> Dutifully restarted.


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga15.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Clarification on part C requirements
Date: 13 Feb 2002 00:40:51 GMT
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga15.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3373


There seems to be considerable confusion and/or concern as to what's
required for part C.

The objective of part C is to see your transport layer working reliably in
a real application.  So, basically, all we're looking for is that some
ftpcopy (either yours, or the provided libftpcopy.a), linked with your
transport layer, can talk to an STCP-compliant ftpd (be it a precompiled
binary, or libftpd.a linked with your transport layer).

If some combination of these work for you (e.g. libftpcopy.a linked with
your transport layer running against the precompiled ftpd is probably
simplest; note that you can *only* test this manually, as the test script
does not check this case), you can document this in your README, and submit,
happy that you've finished assignment two.

You *don't* need to make sure that all possible permutations work, because
as many of you have noticed, things seem to have changed on the system since
Saturday and some things beyond your control (e.g. libftpd.a) are broken.
(You aren't responsible for bugs in external code!)

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga15.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: are the test scripts working ?
Date: 13 Feb 2002 00:44:01 GMT
Lines: 9
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: saga15.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3374

>    If I submitted one at 1:30 or so before the test server was restarted,
>does that mean that I need to resubmit? I tried to, but it told me that my
>submission was still in the queue.


It depends as to whether it was (partially) processed or not.  In your
case, since it says a submission is still in the queue, it should process it
in due course.

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga15.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Clarification on part C requirements
Date: 13 Feb 2002 00:45:55 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga15.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3375


By the way, just to follow up---
This part of the assignment really doesn't require the test script at all.
(Which is fortunate, given the strange behaviour seen recently).

If you have a working part B, and it works in one of the two versions of
ftpcopy with ftpd, you're done.

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga15.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Error 0 (-1) Demystified
Date: 13 Feb 2002 00:48:56 GMT
Lines: 12
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga15.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3376


>If this isn't the issue, if one of the TA's could let us know, we'd
>appreciate it. I will honestly be really surprised, as my code ran just
>right as soon as I took the printf's out, and purify was happy both before
>and after.

Sounds by far the most reasonable explanation for the problems people have
seen.

As I just wrote elsewhere, if you have a working combination (following
taking out output, etc), then you can assume you're done--just focus on the
midterm :-)
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Romain Thibaux 
Newsgroups: su.class.cs244a
Subject: ftpd: Permission denied
Date: Tue, 12 Feb 2002 17:07:08 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: pup11.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1013562429 9139 171.64.66.161 (13 Feb 2002 01:07:09 GMT)
X-Complaints-To: 
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.2.1) Gecko/20010901
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3377

/afs/ir/class/cs244a/WWW/homeworks/hw2/ftpd_src/src/ftpd: Permission denied.
???
Can't we run that directly from there ?

	Romain

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: Re: Anyone pass yet?
Date: Tue, 12 Feb 2002 17:02:34 -0800
Lines: 308
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3378


The same problem for me...
My test log also says about 'wjiang'.. not me..
What's happening?

Hopil.

"Vasco Chatalbashev"  wrote in message

> Is your SUNetID "wjiang"? The path of the terminated process seems to be
for
> "wjiang."
> I had the same problem for a while where the test wouldnt even test my
> submission but
> return various failures which displayed paths to other students' programs
> (the path printed
> when the process is killed.)
> Anyway I have made no changes but somehow pass all the tests. I e-mailed
the
> TAs a couple of
> times but got no response, so I am not sure if the script actually test
> another student's program, but it looks
> like that's what may be going on.
>
>
> "Sandeep Tamhankar"  wrote in message
> 
> > I've seen some semi-conflicting info in various places, so I wanted to
> > verify that the following procedure is supposed to work when using
> > our.ftpcopy (from anyone who's passed the tests and is happy with the
> > general direction of his/her life):
> >
> > 1. Get libftpcopy.a, libftpd.a, and libsupport.a from
> > /class/cs244a/homeworks/hw2/hw1obj
> >
> > 2. Edit the project 2 Makefile s.t. 'all' now includes 'ftpd' and
> > 'our.ftpcopy'.
> >
> > 3. Remove printf statements from transport.c; in particular, don't
> > forget the print_congestion_window function.
> >
> > 4. make clean; make all
> >
> > 5. Run 'ftpd -U' somewhere.  Run 'our.ftpcopy -U' somewhere.
> >
> > 6. See that things work.
> >
> > I just got our.ftpcopy -U to work; the key is to not specify the srcdir
> >   arg.  And it seems to run fine (except for some spurious directory
> > creation that I believe is a flaw in their ftp client).  I submitted my
> > code for testing yesterday night without the print_congestion_window
> > modification.  I failed 3.b and 3.c tests, but not with ERROR(-1) or
> > ERROR(-2) messages as others have reported.  I get the following:
> >
> >   * Test 3.B: Doesn't transfer even a single file correctly
> > [out of 10]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy gz 2 
> > exit line said: "ERROR:  "
> > ---
> > Program output was:
> >
> > ERROR:
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > Nothing be read.
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > The initial seqence num is 29
> > handling the syn or syn_ack
> > CWND: 3072; ERTT: 200
> > CWND: 3072; ERTT: 160
> > All transmissions are complete. Exiting successfully.
> >
> > ---
> >
> >
> > ---
> >
> >
> >
> >
> > ===============
> >
> >      * Test 3.C: Doesn't transfer all files but transfers at least one
> > file properly
> > [out of 5]
> >
> > [SUMMARY] NOT OK
> > [RESULT]  test used: ftpcopy gz 2 
> > exit line said: "ERROR:  "
> > ---
> > diff between expected tree and your tree:
> > 0a1,3
> >
> >  > ./D6/f2.gz 4270565966 2
> >  > ./D6/f3.gz 4103939428 2
> >  > ./f1.gz 4219530715 2
> >
> >
> > ---
> > Program output was:
> >
> > ERROR:
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > Nothing be read.
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > The initial seqence num is 29
> > handling the syn or syn_ack
> > CWND: 3072; ERTT: 200
> > CWND: 3072; ERTT: 160
> > All transmissions are complete. Exiting successfully.
> >
> > ---
> >
> >
> > ---
> >
> > exit line said: "ERROR:  "
> > ---
> > diff between expected tree and your tree:
> > 0a1
> >
> >  > ./f1.gz
> >
> >
> > ---
> > Program output was:
> >
> > ERROR:
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > Nothing be read.
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > The initial seqence num is 29
> > handling the syn or syn_ack
> > CWND: 3072; ERTT: 200
> > CWND: 3072; ERTT: 160
> > All transmissions are complete. Exiting successfully.
> >
> > ---
> >
> >
> > ---
> > test used: ftpcopy gz 2 
> > exit line said: "ERROR:  "
> > ---
> > diff between expected tree and your tree:
> > 0a1,3
> >
> >  > ./D6/f2.gz
> >  > ./D6/f3.gz
> >  > ./f1.gz
> >
> >
> > ---
> > Program output was:
> >
> > ERROR:
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > Nothing be read.
> > serverInfo is 33399
> > in ParseServerString-- serverName is elaine40.Stanford.EDU
> > The port is 33399
> > serverStr at the end of parse fn...is elaine40.Stanford.EDU
> > serverPort at exit of Parse is 33399
> > before chdir is
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > The command line is gz 2 
> >
>
/afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> n/download/elaine40.Stanford.EDU
> > .
> > The initial seqence num is 29
> > handling the syn or syn_ack
> > CWND: 3072; ERTT: 200
> > CWND: 3072; ERTT: 160
> > All transmissions are complete. Exiting successfully.
> >
> > ---
> >
> >
> > ---
> >
> >
> >
> > It seems that test 3.C actually consists of multiple invocations of the
> > program, and thus lots of output.  Anyway, since I haven't been able to
> > get this thing to fail for me, I'm going to make the
> > print_congestion_window change and re-submit for testing, hoping for the
> > best.  But if anyone out there who's gotten all the tests to pass sees a
> > flaw in my steps 1-6 above, please reply to this post with the proper
> > corrections.  If this next test run ends up passing for me, I'll reply
> > to validate the above procedure.
> >
> > -Sandeep
> >
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga22.Stanford.EDU!aminf13
From: Amin Firoozshahian 
Newsgroups: su.class.cs244a
Subject: An alternative, and a problem
Date: Tue, 12 Feb 2002 17:27:10 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3379



  Hi,

  I think I've found an alternative way, I used the ftpd executable file
available instead of de-activating all dprintf commands and everything
seems to be fine.

  But this our.ftpcopy still does not let me to use -U to make it
unreliable!!! Any comments??? Please....!

  Yours,
  Amin

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Luis Robles 
Newsgroups: su.class.cs244a
Subject: Re: ERROR: Bad file number (-1)
Date: Tue, 12 Feb 2002 17:32:12 -0800
Lines: 33
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3380

As posted before, commenting out the printf in
print_cong_window_and_estRtt
got rid of the "Bad file number (-1)" business,

Cheers,

Luis Robles wrote:
> 
> Hey folks,
> 
> My "our.ftpclient" running against "ftpd.old" on the same machine seems
> to
> work fine, ( is this the only good testing combo at this point? )
> 
> But I'm failing a few tests in the part C script with the
> following errors:
> 
> [SUMMARY] NOT OK
> [RESULT] test used: ftpcopy gz 2 
> exit line said: "ERROR: Bad file number (-1) "
> ---
> Program output was:
> elaine40.Stanford.EDU:50716
> elaine40.Stanford.EDU:43704
> ERROR: Bad file number (-1)
> 
> Any ideas what this might be?
> 
> Thanks

-- 
       Luis Robles

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: Anyone pass yet?
Date: Tue, 12 Feb 2002 17:34:34 -0800
Lines: 363
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:0.9.8) Gecko/20020204
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3381

I assumed that the test script was running my program against wjiang's 
directory, but that sounds kinda weird now that I think about it.  In 
the end, I just gave up and submitted.  Need to study for the mid-term, 
and on my own, the ftpd and our.ftpcopy seems to work fine (except for 
the two or three known bugs in our.ftpcopy, which aren't the fault of 
the transport layer).

I never got the result back from my test request after I'd changed the 
printf in print_congestion... to a dprintf.  I think the test server was 
down.  But I don't really care at this point; I heard a rumor that the 
TAs are going to be fair and not trust the test script when grading part 
C since so many weird things seem to be happening.

-Sandeep

BAEHOPIL wrote:
> The same problem for me...
> My test log also says about 'wjiang'.. not me..
> What's happening?
> 
> Hopil.
> 
> "Vasco Chatalbashev"  wrote in message
> 
> 
>>Is your SUNetID "wjiang"? The path of the terminated process seems to be
>>
> for
> 
>>"wjiang."
>>I had the same problem for a while where the test wouldnt even test my
>>submission but
>>return various failures which displayed paths to other students' programs
>>(the path printed
>>when the process is killed.)
>>Anyway I have made no changes but somehow pass all the tests. I e-mailed
>>
> the
> 
>>TAs a couple of
>>times but got no response, so I am not sure if the script actually test
>>another student's program, but it looks
>>like that's what may be going on.
>>
>>
>>"Sandeep Tamhankar"  wrote in message

>>
>>>I've seen some semi-conflicting info in various places, so I wanted to
>>>verify that the following procedure is supposed to work when using
>>>our.ftpcopy (from anyone who's passed the tests and is happy with the
>>>general direction of his/her life):
>>>
>>>1. Get libftpcopy.a, libftpd.a, and libsupport.a from
>>>/class/cs244a/homeworks/hw2/hw1obj
>>>
>>>2. Edit the project 2 Makefile s.t. 'all' now includes 'ftpd' and
>>>'our.ftpcopy'.
>>>
>>>3. Remove printf statements from transport.c; in particular, don't
>>>forget the print_congestion_window function.
>>>
>>>4. make clean; make all
>>>
>>>5. Run 'ftpd -U' somewhere.  Run 'our.ftpcopy -U' somewhere.
>>>
>>>6. See that things work.
>>>
>>>I just got our.ftpcopy -U to work; the key is to not specify the srcdir
>>>  arg.  And it seems to run fine (except for some spurious directory
>>>creation that I believe is a flaw in their ftp client).  I submitted my
>>>code for testing yesterday night without the print_congestion_window
>>>modification.  I failed 3.b and 3.c tests, but not with ERROR(-1) or
>>>ERROR(-2) messages as others have reported.  I get the following:
>>>
>>>  * Test 3.B: Doesn't transfer even a single file correctly
>>>[out of 10]
>>>
>>>[SUMMARY] NOT OK
>>>[RESULT]  test used: ftpcopy gz 2 
>>>exit line said: "ERROR:  "
>>>---
>>>Program output was:
>>>
>>>ERROR:
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>Nothing be read.
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>The initial seqence num is 29
>>>handling the syn or syn_ack
>>>CWND: 3072; ERTT: 200
>>>CWND: 3072; ERTT: 160
>>>All transmissions are complete. Exiting successfully.
>>>
>>>---
>>>
>>>
>>>---
>>>
>>>
>>>
>>>
>>>===============
>>>
>>>     * Test 3.C: Doesn't transfer all files but transfers at least one
>>>file properly
>>>[out of 5]
>>>
>>>[SUMMARY] NOT OK
>>>[RESULT]  test used: ftpcopy gz 2 
>>>exit line said: "ERROR:  "
>>>---
>>>diff between expected tree and your tree:
>>>0a1,3
>>>
>>> > ./D6/f2.gz 4270565966 2
>>> > ./D6/f3.gz 4103939428 2
>>> > ./f1.gz 4219530715 2
>>>
>>>
>>>---
>>>Program output was:
>>>
>>>ERROR:
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>Nothing be read.
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>The initial seqence num is 29
>>>handling the syn or syn_ack
>>>CWND: 3072; ERTT: 200
>>>CWND: 3072; ERTT: 160
>>>All transmissions are complete. Exiting successfully.
>>>
>>>---
>>>
>>>
>>>---
>>>
>>>exit line said: "ERROR:  "
>>>---
>>>diff between expected tree and your tree:
>>>0a1
>>>
>>> > ./f1.gz
>>>
>>>
>>>---
>>>Program output was:
>>>
>>>ERROR:
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>Nothing be read.
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>The initial seqence num is 29
>>>handling the syn or syn_ack
>>>CWND: 3072; ERTT: 200
>>>CWND: 3072; ERTT: 160
>>>All transmissions are complete. Exiting successfully.
>>>
>>>---
>>>
>>>
>>>---
>>>test used: ftpcopy gz 2 
>>>exit line said: "ERROR:  "
>>>---
>>>diff between expected tree and your tree:
>>>0a1,3
>>>
>>> > ./D6/f2.gz
>>> > ./D6/f3.gz
>>> > ./f1.gz
>>>
>>>
>>>---
>>>Program output was:
>>>
>>>ERROR:
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>Nothing be read.
>>>serverInfo is 33399
>>>in ParseServerString-- serverName is elaine40.Stanford.EDU
>>>The port is 33399
>>>serverStr at the end of parse fn...is elaine40.Stanford.EDU
>>>serverPort at exit of Parse is 33399
>>>before chdir is
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>The command line is gz 2 
>>>
>>>
> /afs/ir/class/cs244a/submissions/restricted/hw2.C/xwang00/wjiang//compile.su
> 
>>n/download/elaine40.Stanford.EDU
>>
>>>.
>>>The initial seqence num is 29
>>>handling the syn or syn_ack
>>>CWND: 3072; ERTT: 200
>>>CWND: 3072; ERTT: 160
>>>All transmissions are complete. Exiting successfully.
>>>
>>>---
>>>
>>>
>>>---
>>>
>>>
>>>
>>>It seems that test 3.C actually consists of multiple invocations of the
>>>program, and thus lots of output.  Anyway, since I haven't been able to
>>>get this thing to fail for me, I'm going to make the
>>>print_congestion_window change and re-submit for testing, hoping for the
>>>best.  But if anyone out there who's gotten all the tests to pass sees a
>>>flaw in my steps 1-6 above, please reply to this post with the proper
>>>corrections.  If this next test run ends up passing for me, I'll reply
>>>to validate the above procedure.
>>>
>>>-Sandeep
>>>
>>>
>>
> 
> 


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: An alternative, and a problem
Date: Tue, 12 Feb 2002 17:37:09 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (Windows; U; Win98; en-US; rv:0.9.8) Gecko/20020204
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3382

Don't specify all the arguments; it has a bug where it's not allowing 
six args (including -U).  Do something like:

our.ftpcopy -U gz 2 

and it should work.

-Sandeep

Amin Firoozshahian wrote:
> 
>   Hi,
> 
>   I think I've found an alternative way, I used the ftpd executable file
> available instead of de-activating all dprintf commands and everything
> seems to be fine.
> 
>   But this our.ftpcopy still does not let me to use -U to make it
> unreliable!!! Any comments??? Please....!
> 
>   Yours,
>   Amin
> 
> 


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth6.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Re: More ERROR: Error 0 (-1)
Date: Tue, 12 Feb 2002 22:50:09 -0800
Lines: 17
Distribution: su
Message-ID: 
References: 
 
  
 
NNTP-Posting-Host: myth6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3383

On 12 Feb 2002, Mark Joseph Dolan wrote:

> The problem now is trying to bebug without stdout,
> ofcourse not sure what else to debug.

Unless I'm missing something, replacing all your "printf("'s with
"sprintf(stderr, " should enable you to print out as much as you want. :)

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: Re: More ERROR: Error 0 (-1)
Date: Tue, 12 Feb 2002 23:09:11 -0800
Lines: 26
Distribution: su
Message-ID: 
References:      
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1013584151 12183 128.12.133.48 (13 Feb 2002 07:09:11 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3384

That should certainly work, as ftpd doesn't touch stderr. Also for those who
hadn't found this yet, you can pipe stderr to a file using >& instead of >

Jonathan

"Nickolay Boytchev Stanev"  wrote in message

> On 12 Feb 2002, Mark Joseph Dolan wrote:
>
> > The problem now is trying to bebug without stdout,
> > ofcourse not sure what else to debug.
>
> Unless I'm missing something, replacing all your "printf("'s with
> "sprintf(stderr, " should enable you to print out as much as you want. :)
>
> NStanev
> --
> ####################################################
> # Nickolay Stanev                                  #
> # E-mail:                   #
> # Cell: (650)-269-4756                             #
> # URL: http://www.stanford.edu/~nbstanev/          #
> ####################################################
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga13.Stanford.EDU!ashmi
From: Ashmi 
Newsgroups: su.class.cs244a
Subject: test script working?
Date: Wed, 13 Feb 2002 10:37:35 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3385

hi,
I submitted a test yesterday night at around 10pm and still haven't got
any reply 12 hours later... I tried to resubmit but it says my earlier
request is still in queue. Is the test script so overloaded? Or is it not
working!
Can someone please find out what's happening?
Thanks
Ashmi

###############################################################################

				ASHMI CHOKSHI
Graduate Student				    141L Escondido Village
Dept. of Computer Science			    Stanford University
Stanford University				    Stanford Ca 94305
Stanford Ca 94305				    (650)498-1103

###############################################################################



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: test script working?
Date: 13 Feb 2002 19:06:43 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3386


>I submitted a test yesterday night at around 10pm and still haven't got
>any reply 12 hours later... I tried to resubmit but it says my earlier
>request is still in queue. Is the test script so overloaded? Or is it not
>working!

Restarted.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: testing keeps failing..
Date: 13 Feb 2002 19:10:26 GMT
Lines: 21
Distribution: su
Message-ID: 
References:         
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3387


>Me too facing the same results. It was only yesterday when I had
>everything running...not a single test used to fail. This is an unplesant
>surprise....I hope the TAs give us a quick feedback on if things are
>really fine with the scripts.

>I checked with a couple of more guys...they are also facing the same
>problem starting today.

>thanks
>abhishek


The script does indeed try to build ftpd, using libftpd.a and your transport
layer.

As posted elsewhere, and as Nick mentioned in class, a problem that's reported
by the script does *not* necessarily mean there's a problem with your program;
it just means that you must test manually (which you should be doing anyway,
before you submit anything to the test server).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Error with getsockname()
Date: 13 Feb 2002 19:13:53 GMT
Lines: 9
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3388


>gethostname() only gives us the address of the local host. How do we
>obtain the port number without using getsockname() ?

>If I only use gethostname() as specified in the section, my ftp client
>is unable to send any commands to the server.


getsockname() should still work on the socket on which you're listen()ing.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: our.ftpcopy vs your.ftpcopy for test script
Date: 13 Feb 2002 19:21:02 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3389


>I wonder how we can specify which of our.ftpcopy or your.ftpcopy is to =
>be tested...
>Is it like that our.ftpcopy is used if your.ftpcopy fails to compile,
>and your.ftpcopy is used if it compiles?

I think so.

>And as far as I remember,
>someone posted that our.ftpcopy had some errors regarding empty =
>directories,
>but my recent test result using our.ftpcopy says OK about it..

Since this isn't your problem, you don't need to worry about it.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ERROR (-2) !!!
Date: 13 Feb 2002 19:24:28 GMT
Lines: 11
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3390


>I have seen this bad file number stuff as well. Personally I think the test
>script is somehow on crack, as the error suddenly cropped up for code that
>was previously working. I think this may be one of those cases where, if you
>can create a tree just like theirs locally (it seems they're using the one
>from the HW1 handout, with all 2-byte files), and have it work, then forget
>the script.


Yep, this is exactly the right thing to do.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Anyone pass yet?
Date: 13 Feb 2002 19:29:50 GMT
Lines: 29
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3391

Sandeep Tamhankar  writes:

>I assumed that the test script was running my program against wjiang's 
>directory, but that sounds kinda weird now that I think about it.  In 
>the end, I just gave up and submitted.  Need to study for the mid-term, 
>and on my own, the ftpd and our.ftpcopy seems to work fine (except for 
>the two or three known bugs in our.ftpcopy, which aren't the fault of 
>the transport layer).

>I never got the result back from my test request after I'd changed the 
>printf in print_congestion... to a dprintf.  I think the test server was 
>down.  But I don't really care at this point; I heard a rumor that the 
>TAs are going to be fair and not trust the test script when grading part 
>C since so many weird things seem to be happening.


This is not a "rumour," it's been posted on the newsgroup (multiple times!),
and Nick mentioned in class also.  (It also doesn't just apply to this
assignment, it applies to all of them--who trusts a computer?)  Just please
specify in your README that you use our.ftpcopy (or a modified your.ftpcopy)
running against the provided ftpd, etc.

The test script is not an oracle of absolute truth, it's just a tool to
*try* and help.  In some cases it can, in some cases it can't.  Either way,
it's absolutely no substitute for manual testing.  If the test script
fails, it may or may not mean anything.

I don't know what was going on with the apparent cross-testing.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ftpd: Permission denied
Date: 13 Feb 2002 19:30:29 GMT
Lines: 6
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3392


>/afs/ir/class/cs244a/WWW/homeworks/hw2/ftpd_src/src/ftpd: Permission denied.
>???
>Can't we run that directly from there ?

The permissions are wrong; as a workaround you can do wget.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: peer mismatch errors!!
Date: 13 Feb 2002 19:31:42 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine13.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3393


>    when i run testscript for hw2.C i get the peer mismatch errors 
>but when i run it in my local environment i dont see this behaviour.

Please see the related postings concerning testing, the reliability of
the test script, known issues in our.ftpcopy (must run on the same machine
as ftpd, etc.)

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Shyam Maddali 
Newsgroups: su.class.cs244a
Subject: Questions about solution for PS1 Q6
Date: Wed, 13 Feb 2002 15:49:33 -0800
Lines: 27
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic19.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3394

Hi Professor and TAs,

   when i was looking at the solution for the PS1 question 6 , the
following questions
came up. 
  
   the question 6 says that there is a interpacket gap of I seconds on a
link.
the given solution says that a buffer of 1 packet size ( L bytes ) is
needed.

But considering a case where Ingress Link1, link2 and link3 go active
one after the other 
with out any gap. that is link1 is active between [0, L/R], link2
between [L/R, 2L/R] and
link3 between [2L/R, 3L/R] again link1 between [I, I+L/R] and so on.

if we choose c such that the router is still putting the bits of packet1
from link1 when the time  t = 3L/R, then we need a buffer of 2 packets [
2L bits ].In choosing so we can still 
have a value for I such that the router has processed all 3 packets
given in a work cycle.

Is my understanding correct? or have i misinterpreted the question.

Thanks
Lakshman Shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine25.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Questions about solution for PS1 Q6
Date: 14 Feb 2002 00:48:03 GMT
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine25.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3396


>   the question 6 says that there is a interpacket gap of I seconds on a
>link.
>the given solution says that a buffer of 1 packet size ( L bytes ) is
>needed.

>But considering a case where Ingress Link1, link2 and link3 go active
>one after the other 
>with out any gap. that is link1 is active between [0, L/R], link2
>between [L/R, 2L/R] and
>link3 between [2L/R, 3L/R] again link1 between [I, I+L/R] and so on.

>if we choose c such that the router is still putting the bits of packet1
>from link1 when the time  t = 3L/R, then we need a buffer of 2 packets [
>2L bits ].In choosing so we can still 
>have a value for I such that the router has processed all 3 packets
>given in a work cycle.


Keep in mind that the problem assumes that only one link is active.
(If more than one link were sending data, the router might well need to
buffer more than L bits, as you point out).
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Lakshman Shyam Maddali 
Newsgroups: su.class.cs244a
Subject: Re: Questions about solution for PS1 Q6
Date: Wed, 13 Feb 2002 17:13:53 -0800
Lines: 40
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: epic19.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3397

Matthew Jonathan Holliman wrote:
> 
> >   the question 6 says that there is a interpacket gap of I seconds on a
> >link.
> >the given solution says that a buffer of 1 packet size ( L bytes ) is
> >needed.
> 
> >But considering a case where Ingress Link1, link2 and link3 go active
> >one after the other
> >with out any gap. that is link1 is active between [0, L/R], link2
> >between [L/R, 2L/R] and
> >link3 between [2L/R, 3L/R] again link1 between [I, I+L/R] and so on.
> 
> >if we choose c such that the router is still putting the bits of packet1
> >from link1 when the time  t = 3L/R, then we need a buffer of 2 packets [
> >2L bits ].In choosing so we can still
> >have a value for I such that the router has processed all 3 packets
> >given in a work cycle.
> 
> Keep in mind that the problem assumes that only one link is active.
> (If more than one link were sending data, the router might well need to
> buffer more than L bits, as you point out).

Even  in the case where one link is active at a time, we may still need
buffer space of
more than L. say for the timing info

  between [0, L/R] -- L1 Active, L2, L3 Idle
  between [L/R, 2L/R] -- L1 Idle, L2 Active, L3 Idle
  between [2L/R, 3L/R] -- l1 Idle, L2 Idle, L3 Active
  between [ L/R, I+L/R] -- L1 Active, L2, L3 Idle
and this repeats. such that only one link is active.

so if c < R/2, then router would be putting bits of packet 1 from L1,
and then we need to
buffer 2 Packets.. [ if the router is not able to put L bits in 2L/R
seconds on Egress Link]

Thanks
Lakshman Shyam Maddali
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine21.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Questions about solution for PS1 Q6
Date: 14 Feb 2002 02:26:01 GMT
Lines: 24
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine21.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3398

>Even  in the case where one link is active at a time, we may still need
>buffer space of
>more than L. say for the timing info

>  between [0, L/R] -- L1 Active, L2, L3 Idle
>  between [L/R, 2L/R] -- L1 Idle, L2 Active, L3 Idle
>  between [2L/R, 3L/R] -- l1 Idle, L2 Idle, L3 Active
>  between [ L/R, I+L/R] -- L1 Active, L2, L3 Idle
>and this repeats. such that only one link is active.

>so if c < R/2, then router would be putting bits of packet 1 from L1,
>and then we need to
>buffer 2 Packets.. [ if the router is not able to put L bits in 2L/R
>seconds on Egress Link]


I understood what you meant before, but my interpretation of "one link is
active" is that only one of those links has someone at the end who wishes to
send data.  Your example looks to me like three active links (they just
aren't active simultaneously).

Other than that, I agree with what you're saying--it's just a question
of how to interpret "only one of the links is active."

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga14.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: arun's solution prob 6(b)
Date: Wed, 13 Feb 2002 23:52:22 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3399


hi

Isn't the solution assuming that the sending rate is constant? It seems
like for every ACK we are sending one more packet. But once we
get the ACK for first packet (100 bytes), we send 2 packets, i.e increase
sequence number to 300. When we get the second ACK, we send 4 packets (or 3)
depending on slow start or additive increase. So, the sequence numbers are
definately not increasing at a constant rate.

Any comments?

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: arun's solution prob 6(b)
Date: 14 Feb 2002 17:39:24 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3400



>Isn't the solution assuming that the sending rate is constant? It seems
>like for every ACK we are sending one more packet. But once we
>get the ACK for first packet (100 bytes), we send 2 packets, i.e increase
>sequence number to 300. When we get the second ACK, we send 4 packets (or 3)
>depending on slow start or additive increase. So, the sequence numbers are
>definately not increasing at a constant rate.


You're correct.  However, this effect only happens during slow start, so
it can be neglected as an approximation.  The solution assumes "steady
state"--i.e., slow start has finished, and we're back to regular AIMD.

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga14.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: arun's solution prob 6(b)
Date: Thu, 14 Feb 2002 10:17:53 -0800
Lines: 39
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: saga14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3401

Even in regular AIMD, this will be true only if the window is full, i.e we
send a packet only when we get an ACK for a single packet. In the case
when when window is not full but regular AIMD is going on, we won't be
increasing our sequence number by 100 (send 1 packet) for each ACK (after
one RTT), rather increase the window size by 100 bytes (1 packet) every
RTT, which doesn't imply that the sequence number also increase by 100,
rather by prevous (window size + 100). Have I understood things correctly?

thanks
abhishek


On 14 Feb 2002, Matthew Jonathan Holliman wrote:

>
>
> >Isn't the solution assuming that the sending rate is constant? It seems
> >like for every ACK we are sending one more packet. But once we
> >get the ACK for first packet (100 bytes), we send 2 packets, i.e increase
> >sequence number to 300. When we get the second ACK, we send 4 packets (or 3)
> >depending on slow start or additive increase. So, the sequence numbers are
> >definately not increasing at a constant rate.
>
>
> You're correct.  However, this effect only happens during slow start, so
> it can be neglected as an approximation.  The solution assumes "steady
> state"--i.e., slow start has finished, and we're back to regular AIMD.
>
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine35.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: arun's solution prob 6(b)
Date: 15 Feb 2002 02:14:39 GMT
Lines: 19
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine35.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3402


>Even in regular AIMD, this will be true only if the window is full, i.e we
>send a packet only when we get an ACK for a single packet. In the case
>when when window is not full but regular AIMD is going on, we won't be
>increasing our sequence number by 100 (send 1 packet) for each ACK (after
>one RTT), rather increase the window size by 100 bytes (1 packet) every
>RTT, which doesn't imply that the sequence number also increase by 100,
>rather by prevous (window size + 100). Have I understood things correctly?


I don't know if you care now, but--

The question "how long..." is really looking for a worst-case answer, which
would correspond to sending 4 GB of data over the link continuously.
In this case, since the problem states that no packets are dropped, in the
steady state the sender window *will* indeed be full.  (If it weren't,
TCP wouldn't be operating the link at capacity, which we know is not the
case).

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga6.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: today's anouncement
Date: Thu, 14 Feb 2002 19:20:19 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3403


can someone tell me what exactly was anounced today after the
test....regarding the early start of assignment?

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: today's anouncement
Date: 18 Feb 2002 11:49:01 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine16.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3404

This was an announcement regarding those interested in beta-testing 
assignment #3...if you are still interested you can contact Guido 
 to see if he needs any additional beta testers. 

Arun

Abhishek Das  wrote:

: can someone tell me what exactly was anounced today after the
: test....regarding the early start of assignment?

: thanks
: abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Nate Hill" 
Newsgroups: su.class.cs244a
Subject: Problem set #3
Date: Mon, 18 Feb 2002 16:27:33 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: nate.stanford.edu
X-Trace: news.Stanford.EDU 1014078455 14530 128.12.184.114 (19 Feb 2002 00:27:35 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4807.1700
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4910.0300
Xref: nntp.stanford.edu su.class.cs244a:3405

The link to ps3 on the web page is broken.  The file doesn't exist.  There
are a couple of old assignments in that directory, one in html that looks
like its from 2001 and a pdf from 2000...


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Problem set #3
Date: 19 Feb 2002 01:25:07 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3406

The assignment is not yet available. It should be available along with the 
next programming assignment shortly.

Arun

Nate Hill  wrote:
: The link to ps3 on the web page is broken.  The file doesn't exist.  There
: are a couple of old assignments in that directory, one in html that looks
: like its from 2001 and a pdf from 2000...


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga7.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: 1(c) problem set 3
Date: Thu, 21 Feb 2002 12:02:40 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3407


hi
Are the service rates equal (c) in all the routers?

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine16.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: 1(c) problem set 3
Date: 21 Feb 2002 20:21:45 GMT
Lines: 5
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine16.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3408



>Are the service rates equal (c) in all the routers?

Please see question 1c.
.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: 1(c) problem set 3
Date: Thu, 21 Feb 2002 19:48:54 -0800
Lines: 12
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3409


Is it correct to assume that we are to provide an expression
for the max sum of the queueing delays?

                ~~m
>
>
> >Are the service rates equal (c) in all the routers?
>
> Please see question 1c.
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: vr server
Date: Thu, 21 Feb 2002 21:28:15 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3410


Is the vr server down? I am unable to connect. Thanks.

                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: vr server
Date: 22 Feb 2002 06:10:48 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3411

Everything seems to be functional as of now (10 pm on Thursday).

Arun

Martin Casado  wrote:


: Is the vr server down? I am unable to connect. Thanks.

:                     ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine30.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: Re: vr server
Date: Thu, 21 Feb 2002 23:42:24 -0800
Lines: 31
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine30.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3412


Its running, on the assignment webpage its wrongly mentioned as
"www-vr1.stanford.edu:8100". It is actually "www-vr-1.stanford.edu:8100".
Note the extra "-". TA's please correct it on the webpage.

Gaurav

On 22 Feb 2002, Arun Upadhyaya Kishan wrote:

> Everything seems to be functional as of now (10 pm on Thursday).
>
> Arun
>
> Martin Casado  wrote:
>
>
> : Is the vr server down? I am unable to connect. Thanks.
>
> :                     ~~m
>
>

*************************************************************************
Gaurav Garg                          * Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             *       Escondido Village, Stanford
Department of Electrical Engineering *       CA-94305
Stanford University                  * Ph:650-498-1208
*************************************************************************

-Never ruin an apology with an excuse

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: ICMP questions
Date: Fri, 22 Feb 2002 13:23:18 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3413

A few questions about ICMP.

1) Do we embed the ICMP header into the data portion of an IP packet?  I
figure that's the only way to send it, but i just wanted to make sure
because I don't see it anywhere in the docs.
2) The vr_icmp_hdr struct given to us in the code doesn't seem to have room
for the "first 64 bits of original datagram" the RFC talks about.  Are we
not supposed to worry about that?

thanks,
Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: 1(c) problem set 3
Date: 23 Feb 2002 05:24:58 GMT
Lines: 7
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3414



>Is it correct to assume that we are to provide an expression
>for the max sum of the queueing delays?

Yes, that's right.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP questions
Date: 23 Feb 2002 05:40:59 GMT
Lines: 17
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine43.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3415


>1) Do we embed the ICMP header into the data portion of an IP packet?  I
>figure that's the only way to send it, but i just wanted to make sure
>because I don't see it anywhere in the docs.

That's right.  For a description of this, the first couple of pages of
RFC 792 describe how ICMP messages are encapsulated in IP.

>2) The vr_icmp_hdr struct given to us in the code doesn't seem to have room
>for the "first 64 bits of original datagram" the RFC talks about.  Are we
>not supposed to worry about that?

You should support this.  The vr_icmp_hdr structure is probably most
easily understood by comparing against RFC 792.  It's basically the
common fields in all headers; the offending datagram whose TTL expired
would follow this header as the ICMP payload.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: PS #3: Star Token Qs
Date: Sat, 23 Feb 2002 07:34:14 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:3416

Q5: Am I right to read that in star token networks I, II, and III, the hosts
take turns sending one or zero packets; and that only in network IV may they
send multiple packets?

Also, does the c in problem 1 mean the propagation speed of a signal?

Thanks,
Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: PS #3: one more question
Date: Sat, 23 Feb 2002 07:54:47 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:3417

what happens in 5j if a machine is forced to forfeit its turn because the
TRT has been exceeded? it can't send its token before its final packet,
because there is none. it can't send an idle packet, because that would
defeat the point of not sending a packet.

also, should we consider a steady state or try to work out some particular
scenario?

-Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: arp replys from 171.24.74.17
Date: Sat, 23 Feb 2002 11:34:38 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3418

Hi,

  Is anyone else having problems getting arp replies from 171.24.74.17?  All
  the other machines (162,146,130) reply quickly, however I do not seem to be
  getting replies from 171.24.74.17.  I was getting replies from 17 a few
  days back, so I'm not really sure what is going on. Thanks.

                                          ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: arp replys from 171.24.74.17
Date: 23 Feb 2002 20:29:53 GMT
Lines: 9
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3419


>  Is anyone else having problems getting arp replies from 171.24.74.17?  All
>  the other machines (162,146,130) reply quickly, however I do not seem to be
>  getting replies from 171.24.74.17.  I was getting replies from 17 a few
>  days back, so I'm not really sure what is going on. Thanks.

I'm not sure why they were disappearing.  After restarting the server,
things seem to be okay again.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS #3: Star Token Qs
Date: 23 Feb 2002 20:56:24 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3420


>Q5: Am I right to read that in star token networks I, II, and III, the hosts
>take turns sending one or zero packets; and that only in network IV may they
>send multiple packets?

I don't believe this material has been covered in class yet, correct?  I think
the question should make more sense (and you should have your answer) after
this is discussed in lecture.

>Also, does the c in problem 1 mean the propagation speed of a signal?

No--this wouldn't make any sense, would it?  As the problem states, it's
the minimum rate at which the router serves the flow.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS #3: one more question
Date: 23 Feb 2002 21:32:40 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3421


>what happens in 5j if a machine is forced to forfeit its turn because the
>TRT has been exceeded? it can't send its token before its final packet,
>because there is none. it can't send an idle packet, because that would
>defeat the point of not sending a packet.

Keep in mind that there is an infinite backlog of packets and that all
traffic is asynchronous.  (I don't know what else to say, without giving
away the answer, so it's better that you try and think through this yourself).

>also, should we consider a steady state or try to work out some particular
>scenario?

Since each host has an infinite backlog of packets, you can just consider the
steady state in these conditions.
.

Path: shelby.stanford.edu!nntp.stanford.edu!fable16.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: ICMP
Date: Sat, 23 Feb 2002 16:54:21 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable16.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3422


For the ICMP, shall we send a ICMP packet whenever there is some
packet dropped due to error, say, checksum error,packet length error...

Or , we just send Timeout ICMP Message and  Unreachable ICMP Messages

Thanks a lot!

Best,
-jinhui



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP
Date: 24 Feb 2002 01:11:19 GMT
Lines: 11
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3423



>For the ICMP, shall we send a ICMP packet whenever there is some
>packet dropped due to error, say, checksum error,packet length error...

>Or , we just send Timeout ICMP Message and  Unreachable ICMP Messages


Nope, you only need to generate new ICMP messages for those latter two cases.
(But as the FAQ says, if you receive other ICMP messages from other hosts,
you should forward those as normal).
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic8.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Sanity Checks
Date: Sat, 23 Feb 2002 17:35:20 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3424

The comments above the function VRClient_CheckHeader() say that:

"Do some sanity checks e.g. is protocol IP, is it Ip version 4
is the packet too long/too short etc."

- In an IP packet the protocol will be TCP, UDP etc - where does one check
  for "is protocol IP"?

- The length of payload passed to this function is calculated using
  payload_len = p->header.ip_len - (p->header.ip_hl << 2);

  Therefore, packet can never be too long. And it can never be too short
  because we read at least header len before calling this function. Or
  is it the case that too long and too short refer to something else which
  I am missing.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic8.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: ARP entry timeout
Date: Sat, 23 Feb 2002 18:00:32 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic8.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3425

Two questions:

(1) The assignment says "All entries in the ARP cache should timeout when
not used for more than 5 minutes". Is it "when not used for more than 5
minutes" or "after 5 minutes irrespective of whether used again or not".

(2) Is it permitted to modify the ARPEntry structure to add a field for
calculating timeout of the entry?

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: ICMP source address
Date: Sat, 23 Feb 2002 23:43:35 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3426

Hi,

The RFC specs for the ICMP states that the source address of the IP header
should be the address of the gateway or host that composes the ICMP
message.

To check that I got it correctly, it should be the ip address of
the machine on which we are running our VR_Client. Like if I run it on
elaine7, it should be the DNS lookup of elaine7.stanford.edu.

Am I correct?

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic14.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: ICMP source address
Date: Sun, 24 Feb 2002 00:04:54 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic14.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3427

Although your client is running on elaine, its doing the job for VR_server
(which noone else except us knows) and therefore elaine shouldn't figure
anywhere in packets sent by the client.

Dhawal Kumar

On Sat, 23 Feb 2002, Moh'd Saleem Saleem Alrawashdeh wrote:

> Hi,
>
> The RFC specs for the ICMP states that the source address of the IP header
> should be the address of the gateway or host that composes the ICMP
> message.
>
> To check that I got it correctly, it should be the ip address of
> the machine on which we are running our VR_Client. Like if I run it on
> elaine7, it should be the DNS lookup of elaine7.stanford.edu.
>
> Am I correct?
>
> Thanks,
>
> Moh'd
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga18.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: ICMP questions
Date: Sun, 24 Feb 2002 01:04:02 -0800
Lines: 20
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga18.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3428


hi

> >2) The vr_icmp_hdr struct given to us in the code doesn't seem to have room
> >for the "first 64 bits of original datagram" the RFC talks about.  Are we
> >not supposed to worry about that?
>
> You should support this.  The vr_icmp_hdr structure is probably most
> easily understood by comparing against RFC 792.  It's basically the
> common fields in all headers; the offending datagram whose TTL expired
> would follow this header as the ICMP payload.
>

you mean the icmp header consists all fileds until (and including) the
Internet header? So we append the rest of the payload (64 bits) to this
header to form the IP payload?

thanks
abhishek

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Sanity Checks
Date: 24 Feb 2002 10:12:17 GMT
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3429


>The comments above the function VRClient_CheckHeader() say that:

>"Do some sanity checks e.g. is protocol IP, is it Ip version 4
>is the packet too long/too short etc."

You should do some basic sanity checks, whatever seems reasonable.  Nothing
really to worry about.

>- In an IP packet the protocol will be TCP, UDP etc - where does one check
>  for "is protocol IP"?

You're given an IP header here to verify--thus your function assumes that it
comes from an IP packet!  So there's no field for "IP protocol" as that
wouldn't make sense in this layer.  (This field would turn up in the MAC layer,
but we're not changing this part of the code).  You can just verify that fields
are consistent and as they should be.

>  Therefore, packet can never be too long. And it can never be too short
>  because we read at least header len before calling this function. Or
>  is it the case that too long and too short refer to something else which
>  I am missing.

No, you still have access to the IP header length and the datagram length
in p->header, right?  You also know that one or two conditions must hold for
those fields to be valid.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ARP entry timeout
Date: 24 Feb 2002 10:27:48 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3430



>(1) The assignment says "All entries in the ARP cache should timeout when
>not used for more than 5 minutes". Is it "when not used for more than 5
>minutes" or "after 5 minutes irrespective of whether used again or not".

Since your program is unlikely to run for more than five minutes, it
probably doesn't matter too much!  But you should follow the assignment specs.

>(2) Is it permitted to modify the ARPEntry structure to add a field for
>calculating timeout of the entry?

Yes, that's fine.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP questions
Date: 24 Feb 2002 10:29:45 GMT
Lines: 16
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3431



>> You should support this.  The vr_icmp_hdr structure is probably most
>> easily understood by comparing against RFC 792.  It's basically the
>> common fields in all headers; the offending datagram whose TTL expired
>> would follow this header as the ICMP payload.
>>

>you mean the icmp header consists all fileds until (and including) the
>Internet header? So we append the rest of the payload (64 bits) to this
>header to form the IP payload?

I'm not sure what you mean by "Internet header."  Basically, if you look at
RFC 792, think of the structure definition that would describe those various
ICMP packet formats, and you'll see where vr_icmp_hdr comes from.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: ARP packets
Date: Sun, 24 Feb 2002 03:00:10 -0800
Lines: 28
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3432

Hi,

I want to check my understanding of the ARP packets.

Our MacTable can have a maximum of 4 entries, one for each next hop
interface to VR_Client ( if it sends arp messages for all the surronding
interfaces ). That means, we will have a Mac entry for 172.24.74.130 ,
....146, ...162, ...17

If that is the case, we can assume a fixed size data structure of 4
entries as our MacTable, which will be filled later by getting an arp
responses. Each entry in the table will be for one of the above mentioned
interfaces. (I hope that this simplification is accepted by the TA's)

As a result, when I want to forward a packet for a faraway host, All
what I will need is that Mac table of my default gateway. In other words,
If I didn't find the host ip address in my looktable, I will never need to
send an arp to get the Mac for this host. I only need the Mac for the
gateway, which will then be responsible on forwarding the packet to final
host. Since I only need the mac of my default gateway, I probably would
have it from a previous ARP packet.

Am I correct in my above description?

Thanks,

Moh'd

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ARP packets
Date: 24 Feb 2002 11:17:44 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3433



>As a result, when I want to forward a packet for a faraway host, All
>what I will need is that Mac table of my default gateway. In other words,
>If I didn't find the host ip address in my looktable, I will never need to
>send an arp to get the Mac for this host. I only need the Mac for the
>gateway, which will then be responsible on forwarding the packet to final
>host. Since I only need the mac of my default gateway, I probably would
>have it from a previous ARP packet.

That's right.  (Remember the purpose of ARP--if you're looking up the IP
address of a machine on a different network, it doesn't make much sense to
try and look up its MAC address, because that doesn't do you any good anyway!)

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: Re: ICMP source address
Date: Sun, 24 Feb 2002 11:50:24 -0800
Lines: 37
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3434

The FAQ says to use the global variable 'serverAddr' for this purpose.  The
src of your ICMP packets is supposed to be the vr-server.

Pete

"Dhawal Kumar"  wrote in message

> Although your client is running on elaine, its doing the job for VR_server
> (which noone else except us knows) and therefore elaine shouldn't figure
> anywhere in packets sent by the client.
>
> Dhawal Kumar
>
> On Sat, 23 Feb 2002, Moh'd Saleem Saleem Alrawashdeh wrote:
>
> > Hi,
> >
> > The RFC specs for the ICMP states that the source address of the IP
header
> > should be the address of the gateway or host that composes the ICMP
> > message.
> >
> > To check that I got it correctly, it should be the ip address of
> > the machine on which we are running our VR_Client. Like if I run it on
> > elaine7, it should be the DNS lookup of elaine7.stanford.edu.
> >
> > Am I correct?
> >
> > Thanks,
> >
> > Moh'd
> >
> >
> >
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: Memory Leaks
Date: Sun, 24 Feb 2002 15:56:24 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3435

It looks like vr_client.c, as given to us, leaks memory.  On line, 232, for
example, it allocates a VR_IPPacket structure which is neither freed nor
passed to our code.

So do we just need to worry about freeing the memory we create as well as
the memory passed to our functions in VR_Router?

-Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: more Q's on ICMP packets
Date: Sun, 24 Feb 2002 17:53:35 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3436

Hi,

 I'm a little confused by the ICMP replies sent by other hosts on the network.

 As far as I can tell, according to rfc an icmp time exceeded
 packet should consist of 36 bytes broken down as follows:

 1  byte  (type)
 1  byte  (code)
 2  bytes (checksum)
 4  bytes (unused)
 20 bytes (ip header)
 8  bytes (64 bits of previous packets data
 = 36 bytes

 However, ICMP time exceeded replies from other hosts on the network have ICMP
 packets of size 46 (total IP size is 66). Is there something I am missing
 here, or are they sending an extra 10 bytes? Any thoughts would be
 appreciated, thanks :-)


                        ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding the ARP replies from .17
Date: Sun, 24 Feb 2002 17:57:01 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine33.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3437

Hi,
I am not sure why but the reply from the .17 gateway seems to take a very
long time and also the packets send on that network does not seem to reach
the browser. Is there some problem on the .16 network.
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Regarding the ARP replies from .17
Date: 25 Feb 2002 02:21:03 GMT
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3438


>I am not sure why but the reply from the .17 gateway seems to take a very
>long time and also the packets send on that network does not seem to reach
>the browser. Is there some problem on the .16 network.

Things look okay in the logs (ARP requests receive replies immediately,
even for .17), and everything seems to work fine for me at the moment.
(I can connect to the web server, etc).
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Memory Leaks
Date: 25 Feb 2002 02:30:01 GMT
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine18.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3439


>It looks like vr_client.c, as given to us, leaks memory.  On line, 232, for
>example, it allocates a VR_IPPacket structure which is neither freed nor
>passed to our code.

>So do we just need to worry about freeing the memory we create as well as
>the memory passed to our functions in VR_Router?


If you're passed a pointer to something that will never be used again
(e.g. an IP packet with TTL of 0, which you replace with an ICMP "time
exceeded" message), you should free it.

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic9.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: ICMP
Date: Sun, 24 Feb 2002 19:12:40 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3440

I want to know if its possible that we will be sending "Unreachable
ICMP Message" because we have no routing entry (this is what assignment
says). Because of the default entry in the routing table, I think we'll be
sending such an ICMP message only if we keep waiting for 10 seconds and
there's no ARP response and not when there's "no routing entry". Are there
other situations (for sending Unreachable ICMP) as well?

Dhawal Kumar

On 24 Feb 2002, Matthew Jonathan Holliman wrote:

>
>
> >For the ICMP, shall we send a ICMP packet whenever there is some
> >packet dropped due to error, say, checksum error,packet length error...
>
> >Or , we just send Timeout ICMP Message and  Unreachable ICMP Messages
>
>
> Nope, you only need to generate new ICMP messages for those latter two cases.
> (But as the FAQ says, if you receive other ICMP messages from other hosts,
> you should forward those as normal).
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic9.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: ICMP time exceeded message
Date: Sun, 24 Feb 2002 20:24:42 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic9.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3441

Do we send ICMP time exceeded message if
(1) TTL = 0 in the received datagram or
(2) TTL = 0 after decrementing?

In case we are following (2) do we send ICMP with
original IP header (with TTL = 1) or the original IP Header with modified
TTL (i.e. TTL set to 0) in the ICMP payload?

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: ICMP time exceeded message
Date: Sun, 24 Feb 2002 22:23:55 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3442


Dhawal,

 I'm not sure if this is standard but
 www-vr-1.stanford.edu's icmp response contains
 the ip header with a ttl of 1 which leads me
 to believe it is using
> (2) TTL = 0 after decrementing

                ~~m


> Do we send ICMP time exceeded message if
> (1) TTL = 0 in the received datagram or
> (2) TTL = 0 after decrementing?
>
> In case we are following (2) do we send ICMP with
> original IP header (with TTL = 1) or the original IP Header with modified
> TTL (i.e. TTL set to 0) in the ICMP payload?
>
> Dhawal Kumar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: ICMP time exceeded message
Date: Sun, 24 Feb 2002 22:34:56 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Dhawal Kumar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3443

As section 4.3 of RFC 1812 says, we should follow (2) and send ICMP
with the original IP header (TTL = 1). Xin


On Sun, 24 Feb 2002, Dhawal Kumar wrote:

> Do we send ICMP time exceeded message if
> (1) TTL = 0 in the received datagram or
> (2) TTL = 0 after decrementing?
> 
> In case we are following (2) do we send ICMP with
> original IP header (with TTL = 1) or the original IP Header with modified
> TTL (i.e. TTL set to 0) in the ICMP payload?
> 
> Dhawal Kumar
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Memory Leaks
Date: Sun, 24 Feb 2002 22:56:28 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3444

Pete,

you are right and this is a known bug. I'll fix it this week
and put an updated package online.

You obvioulsy don't have to worry about memory bugs in our code, however
you should free memory that is passed to you.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Sun, 24 Feb 2002, Peter Belknap wrote:

> It looks like vr_client.c, as given to us, leaks memory.  On line, 232, for
> example, it allocates a VR_IPPacket structure which is neither freed nor
> passed to our code.
> 
> So do we just need to worry about freeing the memory we create as well as
> the memory passed to our functions in VR_Router?
> 
> -Pete
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP
Date: 25 Feb 2002 18:46:08 GMT
Lines: 10
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3445


>I want to know if its possible that we will be sending "Unreachable
>ICMP Message" because we have no routing entry (this is what assignment
>says). Because of the default entry in the routing table, I think we'll be
>sending such an ICMP message only if we keep waiting for 10 seconds and
>there's no ARP response and not when there's "no routing entry". Are there
>other situations (for sending Unreachable ICMP) as well?

Thanks, you're correct.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: more Q's on ICMP packets
Date: 25 Feb 2002 19:09:18 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3446


> 1  byte  (type)
> 1  byte  (code)
> 2  bytes (checksum)
> 4  bytes (unused)
> 20 bytes (ip header)
> 8  bytes (64 bits of previous packets data
> = 36 bytes

> However, ICMP time exceeded replies from other hosts on the network have ICMP
> packets of size 46 (total IP size is 66). Is there something I am missing
> here, or are they sending an extra 10 bytes? Any thoughts would be
> appreciated, thanks :-)


It's difficult for me to answer immediately as I cannot easily reproduce
this, but it is certainly legal (in fact, it's recommended by RFC 1812--
the long one that you didn't have to read for the assignment) for a host
to send more than 8 bytes of the offending datagram.

What does tcpdump show for the ICMP time exceeded messages that you
receive?

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Sanity Checks
Date: Mon, 25 Feb 2002 11:43:30 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3447

Dhawal,

> "Do some sanity checks e.g. is protocol IP, is it Ip version 4
> is the packet too long/too short etc."
> 
> - In an IP packet the protocol will be TCP, UDP etc - where does one check
>   for "is protocol IP"?

It is not about checking IP vs. not Ip but instead checking the version 
number of IP (the first 4 bits). If it sais IPv6 (IP, version 6) the 
structure of the IP header would be totally different. Don't expect this 
to happen (the machine can't route IPv6) but check for it anyway. 

> - The length of payload passed to this function is calculated using
>   payload_len = p->header.ip_len - (p->header.ip_hl << 2);
> 
>   Therefore, packet can never be too long. And it can never be too short
>   because we read at least header len before calling this function. Or
>   is it the case that too long and too short refer to something else which
>   I am missing.

As Matthew pointed out, if the length turns out to be -17, you should 
probably not continue to process the packer :-)

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: ARP packets
Date: Mon, 25 Feb 2002 11:53:00 -0800
Lines: 44
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3448

Moh'd,

> Our MacTable can have a maximum of 4 entries, one for each next hop
> interface to VR_Client ( if it sends arp messages for all the surronding
> interfaces ). That means, we will have a Mac entry for 172.24.74.130 ,
> ...146, ...162, ...17

This is right, altough let me add a small clarification to Matthew's 
response. You shouls not have the hosts directly in your routing table. 
Instead have the subnets of the interfaces.

This makes a difference if you get a packet for e.g. 172.24.74.133. If you 
only have 172.24.74.130 in your routing table you would send it to the 
default route. If you have the 172.24.74.128/28 subnet in the routing table 
you would send out an arp request for *.131 on the interface and after it 
times out generate a host unrechable icmp message.

Is that clear?

> If that is the case, we can assume a fixed size data structure of 4
> entries as our MacTable, which will be filled later by getting an arp
> responses. Each entry in the table will be for one of the above mentioned
> interfaces. (I hope that this simplification is accepted by the TA's)

Four sonds a little small to me but having it fixed length (e.g. 10000) is 
acceptable. Pre-initializing it with all possible ip addresses is not ok 
either, assume that you do not know what the ip addresses of the hosts are, 
you only know the subnets!

On the other hand, you do *NOT* have to implement any sophisticated search 
strategy (hash table, RB-Tree) in the arp cache.Linear search is perfectly 
fine for this exercise.

> As a result, when I want to forward a packet for a faraway host, All
> what I will need is that Mac table of my default gateway. In other words,
> If I didn't find the host ip address in my looktable, I will never need to
> send an arp to get the Mac for this host. I only need the Mac for the
> gateway, which will then be responsible on forwarding the packet to final
> host. Since I only need the mac of my default gateway, I probably would
> have it from a previous ARP packet.

Yes, as Matthew pointed out that's correct.

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Memory Leaks
Date: Mon, 25 Feb 2002 12:10:44 -0800
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3449

Peter,

the bug is fixed, a new skelleton tarball is in the hw3 directory.

  Guido

> It looks like vr_client.c, as given to us, leaks memory.  On line, 232,
> for example, it allocates a VR_IPPacket structure which is neither freed
> nor passed to our code.
> 
> So do we just need to worry about freeing the memory we create as well as
> the memory passed to our functions in VR_Router?
> 
> -Pete

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga21.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: more Q's on ICMP packets
Date: Mon, 25 Feb 2002 12:24:31 -0800
Lines: 42
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga21.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3450

It may also be the case that the IP header size is not 20 bytes...you know
the options field. however that will make it maximum 24 bytes IP
header...still doesn't explain.



On Sun, 24 Feb 2002, Martin Casado wrote:

> Hi,
>
>  I'm a little confused by the ICMP replies sent by other hosts on the network.
>
>  As far as I can tell, according to rfc an icmp time exceeded
>  packet should consist of 36 bytes broken down as follows:
>
>  1  byte  (type)
>  1  byte  (code)
>  2  bytes (checksum)
>  4  bytes (unused)
>  20 bytes (ip header)
>  8  bytes (64 bits of previous packets data
>  = 36 bytes
>
>  However, ICMP time exceeded replies from other hosts on the network have ICMP
>  packets of size 46 (total IP size is 66). Is there something I am missing
>  here, or are they sending an extra 10 bytes? Any thoughts would be
>  appreciated, thanks :-)
>
>
>                         ~~m
>
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga3.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: IP packet for server
Date: Mon, 25 Feb 2002 22:05:13 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3451


hi

What are we supposed to do to handle IP packets meant for the vr-server?
A scenario that I can conceive of is: suppose we send an ICMP packet with
TTL 64....its TTL goes becomes 1 somewhere in the network and hence that
router sends back an ICMP packet to vr-server. we receive it and then....

thanks
abhishek



.

Path: shelby.stanford.edu!nntp.stanford.edu!epic10.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Sanity Checks
Date: Mon, 25 Feb 2002 23:47:29 -0800
Lines: 34
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3452

Length can never be negative since we are interpreting the length field as
u_short.

Dhawal Kumar

On Mon, 25 Feb 2002, Guido Appenzeller wrote:

> Dhawal,
>
> > "Do some sanity checks e.g. is protocol IP, is it Ip version 4
> > is the packet too long/too short etc."
> >
> > - In an IP packet the protocol will be TCP, UDP etc - where does one check
> >   for "is protocol IP"?
>
> It is not about checking IP vs. not Ip but instead checking the version
> number of IP (the first 4 bits). If it sais IPv6 (IP, version 6) the
> structure of the IP header would be totally different. Don't expect this
> to happen (the machine can't route IPv6) but check for it anyway.
>
> > - The length of payload passed to this function is calculated using
> >   payload_len = p->header.ip_len - (p->header.ip_hl << 2);
> >
> >   Therefore, packet can never be too long. And it can never be too short
> >   because we read at least header len before calling this function. Or
> >   is it the case that too long and too short refer to something else which
> >   I am missing.
>
> As Matthew pointed out, if the length turns out to be -17, you should
> probably not continue to process the packer :-)
>
>   Guido
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic10.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: IP packet for server
Date: Mon, 25 Feb 2002 23:56:12 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3453

And what if someone maliciously/inadvertently sends a packet with
vr_server as the destination.

The original code drops ICMP packets destined to vr_server but what about
the case I mentioned above.

Dhawal Kumar

On Mon, 25 Feb 2002, Abhishek Das wrote:

>
> hi
>
> What are we supposed to do to handle IP packets meant for the vr-server?
> A scenario that I can conceive of is: suppose we send an ICMP packet with
> TTL 64....its TTL goes becomes 1 somewhere in the network and hence that
> router sends back an ICMP packet to vr-server. we receive it and then....
>
> thanks
> abhishek
>
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic10.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: IP Identification field for ICMP packets
Date: Tue, 26 Feb 2002 00:04:39 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3454

How are we supposed to set the identification field in the IP header for
ICMP packets originating from vr_server? Since half of the IP
functionality is handled by vr_server (we handle only a subset of ports)
our identification field may clash with one used by vr_server in packets
it is handling itself.

And then there are issues like wrap around.

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Romain Thibaux 
Newsgroups: su.class.cs244a
Subject: What to download ?
Date: Tue, 26 Feb 2002 00:38:28 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: thibaux.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1014712720 10905 128.12.191.77 (26 Feb 2002 08:38:40 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.76 [en] (X11; U; Linux 2.4.2-2 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3455

What are we supposed to download from the web site ? Save the page ??

    Romain

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic10.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Deliverables (cksum of downloaded files and ping)
Date: Tue, 26 Feb 2002 01:46:24 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic10.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3456

- How do I calculate cksum of downloaded files if their size is not even
 (cksum takes an array of 16 bit numbers)

- Do we have to ping a port - In fact is there such a way? Or do we have
  to attach the output of just "ping www-vr-3.stanford.edu"

Dhawal Kumar


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: ICMP Header
Date: Tue, 26 Feb 2002 13:06:47 -0800
Lines: 26
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3457

Hi all,

I am facing a weird situation here. When I try to assemble the ICMP
packet, I use the sizeof(IPHeader) as the value of the ip_hl entry in the
IPHeader structure. That means it is set to 20 ( which is correct ).

After sending this packet and view it using using ethreal, I found
that my ICMPPacket doesn't have the required min length of IPPacekt.
Ethreal tells me that the ip_hl entry which I set above is 16 and not 20.
I went and looked in the packet itself to find just two hexa numbers for
both the packet version and the packet length. For the correct case, they
should be set to 45 (where 4 is for IP_ver and 5 is the length of the
IP_Header). I have no idea how 5 representing the length of the packet
since it should be 20! In my case, when I set the ip_hl to be 20, I am
getting in the IPPacket 44 rather than 45. This is why it tells me that
my IPPacket length is 16 rather than 20.

While playing to solve this, I just changed the ip_hl to be 21 rather than
20, and I got the required 45. After that everything works great, but this
is a wrong hack, and I want to solve the above correctly :)

Does anyone have an idea why it is doing this?

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga19.Stanford.EDU!abishek
From: Abhishek Das 
Newsgroups: su.class.cs244a
Subject: Re: ICMP Header
Date: Tue, 26 Feb 2002 16:34:41 -0800
Lines: 43
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga19.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3458


ip_hl says the no. of words, i.e multiply ip_hl with 4 to get correct
header length.

On Tue, 26 Feb 2002, Moh'd Saleem Saleem Alrawashdeh wrote:

> Hi all,
>
> I am facing a weird situation here. When I try to assemble the ICMP
> packet, I use the sizeof(IPHeader) as the value of the ip_hl entry in the
> IPHeader structure. That means it is set to 20 ( which is correct ).
>
> After sending this packet and view it using using ethreal, I found
> that my ICMPPacket doesn't have the required min length of IPPacekt.
> Ethreal tells me that the ip_hl entry which I set above is 16 and not 20.
> I went and looked in the packet itself to find just two hexa numbers for
> both the packet version and the packet length. For the correct case, they
> should be set to 45 (where 4 is for IP_ver and 5 is the length of the
> IP_Header). I have no idea how 5 representing the length of the packet
> since it should be 20! In my case, when I set the ip_hl to be 20, I am
> getting in the IPPacket 44 rather than 45. This is why it tells me that
> my IPPacket length is 16 rather than 20.
>
> While playing to solve this, I just changed the ip_hl to be 21 rather than
> 20, and I got the required 45. After that everything works great, but this
> is a wrong hack, and I want to solve the above correctly :)
>
> Does anyone have an idea why it is doing this?
>
> Moh'd
>
>
>

Abhishek Das
Graduate Research Assistant
Computer Systems Lab
Stanford University

Address:-
Escondido Village 33B
Stanford CA 94305

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Darren Lewis" 
Newsgroups: su.class.cs244a
Subject: traceroute
Date: Tue, 26 Feb 2002 17:35:42 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: darren.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3459

When we traceroute to one of the www-vr-{1,2,3}, should we see www-vr-server
in the trace, or just the firewall and the destination machine?

-Darren


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Deliverables (cksum of downloaded files and ping)
Date: Tue, 26 Feb 2002 20:38:16 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Dhawal Kumar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3460

Hi,

> - How do I calculate cksum of downloaded files if their size is not even
>  (cksum takes an array of 16 bit numbers)

It should be padded with 0s to a 16-bit boundary.
 
> - Do we have to ping a port - In fact is there such a way? Or do we have
>   to attach the output of just "ping www-vr-3.stanford.edu"

It is an error in the handout. You don't need to submit the output of
"ping".

Xin

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: What to download ?
Date: Tue, 26 Feb 2002 20:58:12 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Romain Thibaux 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3461

The VR client should be able to download files from the website. Please
refer to Q8 in FAQ for examples. As for the deliverables, please refer to
the PA#3 handout. Xin




Please see "Deliverables" session in PA#3 handout. Xin


On Tue, 26 Feb 2002, Romain Thibaux wrote:

> What are we supposed to download from the web site ? Save the page ??
> 
>     Romain
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: IP Identification field for ICMP packets
Date: Tue, 26 Feb 2002 23:29:09 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Dhawal Kumar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3462

The identification field in the IP header for ICMP packets will not be
really in use, so it could be set at any arbitrary value. Xin


On Tue, 26 Feb 2002, Dhawal Kumar wrote:

> How are we supposed to set the identification field in the IP header for
> ICMP packets originating from vr_server? Since half of the IP
> functionality is handled by vr_server (we handle only a subset of ports)
> our identification field may clash with one used by vr_server in packets
> it is handling itself.
> 
> And then there are issues like wrap around.
> 
> Dhawal Kumar
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: 1(c) problem set 3
Date: 27 Feb 2002 09:30:45 GMT
Lines: 9
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga8.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3463


If the service rate of all the routers is c, isn't it true that
the maximum flow between any two routers is c? And if this is the
case, then it seems that every router after the first can
service all incoming data without delay.

What am I missing?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: 1(c) problem set 3
Date: 27 Feb 2002 19:12:40 GMT
Lines: 16
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: saga3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3464

Yes, every router can provide a service rate of c. If you think about what 
the first router does to the incoming flow (regardless of its rate), then 
you will have found the answer.

Arun

Derrick Wen-Shiuan Tong  wrote:

: If the service rate of all the routers is c, isn't it true that
: the maximum flow between any two routers is c? And if this is the
: case, then it seems that every router after the first can
: service all incoming data without delay.

: What am I missing?


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "BAEHOPIL" 
Newsgroups: su.class.cs244a
Subject: About TRT check in FDDI
Date: Wed, 27 Feb 2002 12:03:27 -0800
Lines: 23
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3465


I have some question about TRT checking in FDDI..

With actual TRT bounded by 2*TTRT+TRANSP_max,
if TRANSP_max is maximum transmit time per packet,
then it seems that TRT check is done packet-by-packet..
That is, if a host has 3 async packets to send,
but TTRT is exceeded after sending 2 packets,
then it doesn't transmit the third?
( It seems to me that allowing the host to send all the packets
defeats the purpose of TRT checking.. and TRANSP_max should
mean something else.. )

And if TRT checking is really done packet-by-packet,
 it seems that the order of sync/async is important...
That is, if sync packets are sent before async,
then async may not be sent even if token reception time was declared Early..

Thank you in advance for any comments....

Hopil Bae


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine43.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: What to download ?
Date: Wed, 27 Feb 2002 16:34:40 -0800
Lines: 46
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine43.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Xin Wang 
cc: Romain Thibaux 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3466

Hi,

Yet I don't know what is required to be downloaded from the webserver. The
PA handout says in the deliverable section:

(2)  The name and length of the three files that you downloaded from the
web servers.

However, there is only Apache test pages on the webservers, so what files
to download?

Thanks,

Moh'd

On Tue, 26 Feb 2002, Xin Wang wrote:

> The VR client should be able to download files from the website. Please
> refer to Q8 in FAQ for examples. As for the deliverables, please refer to
> the PA#3 handout. Xin
>
>
>
>
> Please see "Deliverables" session in PA#3 handout. Xin
>
>
> On Tue, 26 Feb 2002, Romain Thibaux wrote:
>
> > What are we supposed to download from the web site ? Save the page ??
> >
> >     Romain
> >
> >
>
> --
> ---------------------------------
> Xin Wang
>
> Department of Computer Science
> Stanford University
>
> 
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: Submit Script?
Date: Wed, 27 Feb 2002 18:01:34 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3467

TAs,

In the interests of having someone beat Martin Casado on one of these
competitions, I'd like to submit my VR router.  The normal submit script
doesn't seem to be set up for hw3.  Am I not seeing the one we're supposed
to use, or is it not up yet?

thanks,
Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: IP packet for server
Date: Wed, 27 Feb 2002 19:16:37 -0800
Lines: 32
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3468

Dhawal,

you can drop any packets that have the destination address of the vr-server 
itself. However you should never get them in the first place.

  Guido

> And what if someone maliciously/inadvertently sends a packet with
> vr_server as the destination.
> 
> The original code drops ICMP packets destined to vr_server but what about
> the case I mentioned above.
> 
> Dhawal Kumar
> 
> On Mon, 25 Feb 2002, Abhishek Das wrote:
> 
>>
>> hi
>>
>> What are we supposed to do to handle IP packets meant for the vr-server?
>> A scenario that I can conceive of is: suppose we send an ICMP packet with
>> TTL 64....its TTL goes becomes 1 somewhere in the network and hence that
>> router sends back an ICMP packet to vr-server. we receive it and then....
>>
>> thanks
>> abhishek
>>
>>
>>
>>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Two announcements for PA#3
Date: Wed, 27 Feb 2002 19:27:19 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3469

Two announcements for PA #3:

** Routing table - There were questions about how to implement the routing
   table. First, having one entry for each *HOST* ist not sufficient. You
   should have one entry for each *SUBNET*. E.g. your router has to know
   how to route packets for hosts that are on the subnets of one of its
   interfaces, even if this host doesn't exist.
   As an example, the correct behavior for a packet for 172.24.74.137 is
   not routing it to the default gateway. Instead your should generate an
   ARP request on the correct interface and after it times out send a
   host unreachable ICMP to the source.

** Downloading files - For the web server simply test if you are seeing
   the default Apache web page. For ftp, download the "dowhnload_me.*" file
   and either send us the checksum or correctly state what is in the
   file. For host www-vr-1 the correct answer would be a rendering of
   the multicast backbone. For the other two it should be easier to
   identify the contents. (Hint: Files that end in .au are Sun audio
   files :-)
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: What to download ?
Date: Wed, 27 Feb 2002 19:27:55 -0800
Lines: 52
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3470

Moh'd,

I just posted a clarification. There is currently nothing on the web pages, 
but there are different files on the ftp servers.

  Guido

> Hi,
> 
> Yet I don't know what is required to be downloaded from the webserver. The
> PA handout says in the deliverable section:
> 
> (2)  The name and length of the three files that you downloaded from the
> web servers.
> 
> However, there is only Apache test pages on the webservers, so what files
> to download?
> 
> Thanks,
> 
> Moh'd
> 
> On Tue, 26 Feb 2002, Xin Wang wrote:
> 
>> The VR client should be able to download files from the website. Please
>> refer to Q8 in FAQ for examples. As for the deliverables, please refer to
>> the PA#3 handout. Xin
>>
>>
>>
>>
>> Please see "Deliverables" session in PA#3 handout. Xin
>>
>>
>> On Tue, 26 Feb 2002, Romain Thibaux wrote:
>>
>> > What are we supposed to download from the web site ? Save the page ??
>> >
>> >     Romain
>> >
>> >
>>
>> --
>> ---------------------------------
>> Xin Wang
>>
>> Department of Computer Science
>> Stanford University
>>
>> 
>>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: traceroute
Date: Wed, 27 Feb 2002 19:29:09 -0800
Lines: 13
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3471

Darren,

you definitely should see www-server as a hop! Test it with demo_vr and 
www-vr-1 to see how it should look like.

  Guido

> When we traceroute to one of the www-vr-{1,2,3}, should we see
> www-vr-server in the trace, or just the firewall and the destination
> machine?
> 
> -Darren

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Submit Script?
Date: Wed, 27 Feb 2002 19:30:58 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3472

It will be up tomorrow, until then if anyone wants to time stamp their 
solution, tar it and mail it to your TA. Just make sure your final solution 
is *exactly* the same as the one you mailed.

  Guido

> TAs,
> 
> In the interests of having someone beat Martin Casado on one of these
> competitions, I'd like to submit my VR router.  The normal submit script
> doesn't seem to be set up for hw3.  Am I not seeing the one we're supposed
> to use, or is it not up yet?
> 
> thanks,
> Pete

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: 1(c) problem set 3
Date: 28 Feb 2002 04:21:01 GMT
Lines: 22
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: saga3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3473

For additional clarification/hints on this question, please refer to the 
email Professor McKeown sent to the class.

Arun

Arun Upadhyaya Kishan  wrote:
: Yes, every router can provide a service rate of c. If you think about what 
: the first router does to the incoming flow (regardless of its rate), then 
: you will have found the answer.

: Arun

: Derrick Wen-Shiuan Tong  wrote:

: : If the service rate of all the routers is c, isn't it true that
: : the maximum flow between any two routers is c? And if this is the
: : case, then it seems that every router after the first can
: : service all incoming data without delay.

: : What am I missing?


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Initial condition problem #4:wq
Date: Wed, 27 Feb 2002 20:37:34 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3474

Hello,

In problem four, are we to assume that the TRTs of the
hosts were all initialized to 8ms on startup and the
first row of the table represents the first go-around
of the token? Thanks.

                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: PS #5e)
Date: Wed, 27 Feb 2002 20:45:43 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3475

On 5e, the problem set says 'f is the size of the set F'.  However, that
can't be right if f * PROP is what the problem says it is.  That would mean
it takes longer to send the token around in this situation, when it should
take less time.

Is it instead the case that the size of the set F is Nf ?  That would seem
to make more sense...

-Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: PS #5e)
Date: 28 Feb 2002 07:58:02 GMT
Lines: 23
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga19.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3476

I'm also confused about this question. Is PROP to be interpreted
as time it takes for the token to travel from C_1 to C_2 ... 
to C_N ? Or is it simply the time it takes for data to reach 
all the other nodes (which is shorter now because of the star 
topology) ?  Question 5e seems to make more sense if PROP has
the latter interpretation, unless f is defined incorrectly,
as Pete is suggesting.

- Derrick


Peter Belknap  wrote:
: On 5e, the problem set says 'f is the size of the set F'.  However, that
: can't be right if f * PROP is what the problem says it is.  That would mean
: it takes longer to send the token around in this situation, when it should
: take less time.

: Is it instead the case that the size of the set F is Nf ?  That would seem
: to make more sense...

: -Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: Re: PS #3: one more question
Date: 28 Feb 2002 10:12:05 GMT
Lines: 35
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga19.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3477

I think what Steven was asking is "How does a machine know which packet
will be the 'last' packet it is allowed to send?" In FDDI, a machine
sends a packet if the TTRT has not yet expired--it has no idea whether
TTRT will expire during the packet send, or if it will be able to
send more packets afterward--it has no prior notion of the "last" packet.

Prof McKeown explained that this was the case because the hosts do not
know how long a packet will be, and thus how long it will take to 
send it. In the case of Star Token IV networks, however, the machines 
know exactly how long it will take to send a packet. Should we assume, 
then, that each machine does a little precalculation to determine 
whether the next packet it sends will be the last packet it is allowed 
to send (and thus, know to send the token first)?


- Derrick



Matthew Jonathan Holliman  wrote:

:>what happens in 5j if a machine is forced to forfeit its turn because the
:>TRT has been exceeded? it can't send its token before its final packet,
:>because there is none. it can't send an idle packet, because that would
:>defeat the point of not sending a packet.

: Keep in mind that there is an infinite backlog of packets and that all
: traffic is asynchronous.  (I don't know what else to say, without giving
: away the answer, so it's better that you try and think through this yourself).

:>also, should we consider a steady state or try to work out some particular
:>scenario?

: Since each host has an infinite backlog of packets, you can just consider the
: steady state in these conditions.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: Initial condition problem #4:wq
Date: Thu, 28 Feb 2002 02:35:22 -0800
Lines: 42
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3478

Hi,

I have more questions about this problem:

1. As martin asked, what are the TRT timer at the starting time (t=0) ? If
we assume that all of them were 8 msec, the the TRT of Host1 should also
be 8msec? I think that the question should provide us with the value of
the TRT timer for all the hosts at time t=0.

2. For the propogation delay, should we consider it "equally divisible"
between the hosts? So there is a delay of .25 msec when going from a host
to its neighbor?

3. I am not able to understand the first filled column. It says that TRT
was 1 msec when this host obtained token. Then the host will do the
following:

1. Send Sync data ( which takes 2msec)
2. Calculate THT ( which is now TRT - Time send Synch = 1msec - 2msec = -1
msec)
3. Since THT is lower than 0 msec, the host shouldn't send any ASYN
data. However, the filled column states that the host sent 1 msec ASYNCH
data. So, what I am missing here?

Thanks,

Moh'd

 On Wed, 27 Feb 2002, Martin Casado wrote:

> Hello,
>
> In problem four, are we to assume that the TRTs of the
> hosts were all initialized to 8ms on startup and the
> first row of the table represents the first go-around
> of the token? Thanks.
>
>                     ~~m
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth1.Stanford.EDU!rgustin
From: Reid Gustin 
Newsgroups: su.class.cs244a
Subject: Re: Initial condition problem #4:wq
Date: Thu, 28 Feb 2002 11:55:42 -0800
Lines: 25
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: myth1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3479

On Thu, 28 Feb 2002, Moh'd Saleem Saleem Alrawashdeh wrote:
> 1. As martin asked, what are the TRT timer at the starting time (t=0) ? If
> we assume that all of them were 8 msec, the the TRT of Host1 should also
> be 8msec? I think that the question should provide us with the value of
> the TRT timer for all the hosts at time t=0.

I think it's reasonable to assume they all start at TRT = 1. I don't see
another reasonable value right now.

> 3. I am not able to understand the first filled column. It says that TRT
> was 1 msec when this host obtained token. Then the host will do the
> following:

> 1. Send Sync data ( which takes 2msec)
> 2. Calculate THT ( which is now TRT - Time send Synch = 1msec - 2msec = -1
> msec)
> 3. Since THT is lower than 0 msec, the host shouldn't send any ASYN
> data. However, the filled column states that the host sent 1 msec ASYNCH
> data. So, what I am missing here?

You're missing the TTRT. Your calculation should be something along the
lines of TTRT - TRT - Time send Synch = 8 - 1 - 2 = 5.

Reid

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Derrick Wen-Shiuan Tong 
Newsgroups: su.class.cs244a
Subject: PS #5j)
Date: 28 Feb 2002 20:12:54 GMT
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga19.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3480

If the TTRT has already expired by the time C_2 receives the token,
i.e. if C_1 uses it all up, then does C_2 simply pass the token along?
It would seem that this defeats the Star Token III design of avoiding
the overhead of an idle line. Does it pass the token along and send
exactly one packet after it?

Thanks,

Derrick


.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Submissions open, Debugging
Date: Thu, 28 Feb 2002 12:38:22 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3481

Hi everyone,

the submit script is now up and running, if you are done go
for it! Note that there is no test script for this assignment.

Also if any of you has a client that consistently manages to
crash the vr server, please send it to me as an attachment with
a short line on how it works. This would be a big help for us
debugging the server!

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: Submissions open, Debugging
Date: Thu, 28 Feb 2002 13:37:31 -0800
Lines: 24
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3482

The submission script is giving me a permission error. I am not able to
submit it.

Moh'd
On Thu, 28 Feb 2002, Guido Appenzeller wrote:

> Hi everyone,
>
> the submit script is now up and running, if you are done go
> for it! Note that there is no test script for this assignment.
>
> Also if any of you has a client that consistently manages to
> crash the vr server, please send it to me as an attachment with
> a short line on how it works. This would be a big help for us
> debugging the server!
>
>   Guido
>
> ---------------------------------------------------------------
> Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
>
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Xiaoyan Cheng 
Newsgroups: su.class.cs244a
Subject: Re: Initial condition problem #4:wq
Date: Thu, 28 Feb 2002 14:26:25 -0800
Lines: 54
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: fable2.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3483

Moh'd Saleem Saleem Alrawashdeh wrote:
> 
> Hi,
> 
> I have more questions about this problem:
> 
> 1. As martin asked, what are the TRT timer at the starting time (t=0) ? If
> we assume that all of them were 8 msec, the the TRT of Host1 should also
> be 8msec? I think that the question should provide us with the value of
> the TRT timer for all the hosts at time t=0.

==>This means in the last round, no data is transmited, so the measured
TRT =propagation delay =1ms 

> 
> 2. For the propogation delay, should we consider it "equally divisible"
> between the hosts? So there is a delay of .25 msec when going from a host
> to its neighbor?

==>I think a smarter way is to substract 1ms from TTRT, so you can use
TTRT=7ms for the rest of the problem .

> 
> 3. I am not able to understand the first filled column. It says that TRT
> was 1 msec when this host obtained token. Then the host will do the
> following:
> 
> 1. Send Sync data ( which takes 2msec)
> 2. Calculate THT ( which is now TRT - Time send Synch = 1msec - 2msec = -1
> msec)

==>THT=TTRT-TRT=TTRT-1ms, so host1 is still able to transmit SYNC(2ms)
and ASYNC(1ms)

> 3. Since THT is lower than 0 msec, the host shouldn't send any ASYN
> data. However, the filled column states that the host sent 1 msec ASYNCH
> data. So, what I am missing here?
> 
> Thanks,
> 
> Moh'd
> 
>  On Wed, 27 Feb 2002, Martin Casado wrote:
> 
> > Hello,
> >
> > In problem four, are we to assume that the TRTs of the
> > hosts were all initialized to 8ms on startup and the
> > first row of the table represents the first go-around
> > of the token? Thanks.
> >
> >                     ~~m
> >
> >
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Xiaoyan Cheng 
Newsgroups: su.class.cs244a
Subject: PS3, #4
Date: Thu, 28 Feb 2002 14:34:12 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: fable2.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3484

"Each host has a maximum allowed allocation of SYN traffic. ", for
example, Host4 has Max SYNC=1ms---should we take this number as the
upper limit of the SUM of all the transmitted SYNC data for all the rows
in each column or just for each token holding time?
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: About TRT check in FDDI
Date: 1 Mar 2002 02:39:11 GMT
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3485


>I have some question about TRT checking in FDDI..

>With actual TRT bounded by 2*TTRT+TRANSP_max,
>if TRANSP_max is maximum transmit time per packet,
>then it seems that TRT check is done packet-by-packet..
>That is, if a host has 3 async packets to send,
>but TTRT is exceeded after sending 2 packets,
>then it doesn't transmit the third?
>( It seems to me that allowing the host to send all the packets
>defeats the purpose of TRT checking.. and TRANSP_max should
>mean something else.. )

>And if TRT checking is really done packet-by-packet,
> it seems that the order of sync/async is important...
>That is, if sync packets are sent before async,
>then async may not be sent even if token reception time was declared Early..


That's right--in fact, if you caught lecture today, Prof. McKeown went over
FDDI and almost this exact question was raised.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS3, #4
Date: 1 Mar 2002 02:42:59 GMT
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3486

Xiaoyan Cheng  writes:

>"Each host has a maximum allowed allocation of SYN traffic. ", for
>example, Host4 has Max SYNC=1ms---should we take this number as the
>upper limit of the SUM of all the transmitted SYNC data for all the rows
>in each column or just for each token holding time?


Keep in mind the guarantees that FDDI makes.  For TTRT to make much sense,
the hosts must have agreed upon the maximum amount of synchronous data
they'll send during a token rotation.  (Otherwise, how could they really have
a target token rotation time?)  So this is what Max SYNC corresponds to.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS #3: one more question
Date: 1 Mar 2002 02:48:44 GMT
Lines: 17
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3487


>I think what Steven was asking is "How does a machine know which packet
>will be the 'last' packet it is allowed to send?" In FDDI, a machine
>sends a packet if the TTRT has not yet expired--it has no idea whether
>TTRT will expire during the packet send, or if it will be able to
>send more packets afterward--it has no prior notion of the "last" packet.

>Prof McKeown explained that this was the case because the hosts do not
>know how long a packet will be, and thus how long it will take to 
>send it. In the case of Star Token IV networks, however, the machines 
>know exactly how long it will take to send a packet. Should we assume, 
>then, that each machine does a little precalculation to determine 
>whether the next packet it sends will be the last packet it is allowed 
>to send (and thus, know to send the token first)?


You can assume that hosts just follow the regular FDDI scheme.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS #5j)
Date: 1 Mar 2002 03:02:16 GMT
Lines: 6
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine39.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3488


>If the TTRT has already expired by the time C_2 receives the token,
>i.e. if C_1 uses it all up, then does C_2 simply pass the token along?

Yes, it behaves like FDDI.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Tim Chao 
Newsgroups: su.class.cs244a
Subject: Re: PS3, #4
Date: Thu, 28 Feb 2002 20:59:29 -0800
Lines: 22
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: slimtc.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.7 [en] (Win98; I)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3489

Can a host transmit only part of a packet while it has the token? For example, if
the token arrives at HOST A 3ms early, it gets 3ms to send asynchronous data. If
it has 1ms and 4ms packets to send, will it send the 1ms and also the first 2ms
of the second packet? or will it only send the 1ms packet?

Thanks,
Tim

Matthew Jonathan Holliman wrote:

> Xiaoyan Cheng  writes:
>
> >"Each host has a maximum allowed allocation of SYN traffic. ", for
> >example, Host4 has Max SYNC=1ms---should we take this number as the
> >upper limit of the SUM of all the transmitted SYNC data for all the rows
> >in each column or just for each token holding time?
>
> Keep in mind the guarantees that FDDI makes.  For TTRT to make much sense,
> the hosts must have agreed upon the maximum amount of synchronous data
> they'll send during a token rotation.  (Otherwise, how could they really have
> a target token rotation time?)  So this is what Max SYNC corresponds to.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: ARP packets
Date: Thu, 28 Feb 2002 21:35:10 -0800
Lines: 52
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: fable3.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3490

Would it be okay to assume a fixed size for the arp queue structure as
well?

Thanks,
Clayton

Guido Appenzeller wrote:
> 
> Moh'd,
> 
> > Our MacTable can have a maximum of 4 entries, one for each next hop
> > interface to VR_Client ( if it sends arp messages for all the surronding
> > interfaces ). That means, we will have a Mac entry for 172.24.74.130 ,
> > ...146, ...162, ...17
> 
> This is right, altough let me add a small clarification to Matthew's
> response. You shouls not have the hosts directly in your routing table.
> Instead have the subnets of the interfaces.
> 
> This makes a difference if you get a packet for e.g. 172.24.74.133. If you
> only have 172.24.74.130 in your routing table you would send it to the
> default route. If you have the 172.24.74.128/28 subnet in the routing table
> you would send out an arp request for *.131 on the interface and after it
> times out generate a host unrechable icmp message.
> 
> Is that clear?
> 
> > If that is the case, we can assume a fixed size data structure of 4
> > entries as our MacTable, which will be filled later by getting an arp
> > responses. Each entry in the table will be for one of the above mentioned
> > interfaces. (I hope that this simplification is accepted by the TA's)
> 
> Four sonds a little small to me but having it fixed length (e.g. 10000) is
> acceptable. Pre-initializing it with all possible ip addresses is not ok
> either, assume that you do not know what the ip addresses of the hosts are,
> you only know the subnets!
> 
> On the other hand, you do *NOT* have to implement any sophisticated search
> strategy (hash table, RB-Tree) in the arp cache.Linear search is perfectly
> fine for this exercise.
> 
> > As a result, when I want to forward a packet for a faraway host, All
> > what I will need is that Mac table of my default gateway. In other words,
> > If I didn't find the host ip address in my looktable, I will never need to
> > send an arp to get the Mac for this host. I only need the Mac for the
> > gateway, which will then be responsible on forwarding the packet to final
> > host. Since I only need the mac of my default gateway, I probably would
> > have it from a previous ARP packet.
> 
> Yes, as Matthew pointed out that's correct.
> 
>   Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Initial condition problem #4:wq
Date: 1 Mar 2002 09:05:27 GMT
Lines: 29
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine27.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3491

: I have more questions about this problem:

: 1. As martin asked, what are the TRT timer at the starting time (t=0) ? If
: we assume that all of them were 8 msec, the the TRT of Host1 should also
: be 8msec? I think that the question should provide us with the value of
: the TRT timer for all the hosts at time t=0.

The TRT is at 1ms meaning that in the previous round no one sent data 
(this is the PROP time). So the first host can send 7ms data in this case.

: 2. For the propogation delay, should we consider it "equally divisible"
: between the hosts? So there is a delay of .25 msec when going from a host
: to its neighbor?

This is OK and will give you the correct answer; remember that all this 
does is eliminate one second from actual available token holding time. 
Effectively, the TTRT is reduced.

: 3. I am not able to understand the first filled column. It says that TRT
: was 1 msec when this host obtained token. Then the host will do the
: following:

So in this case, the THT is 7 ms. So the sync data is sent (2 ms) and the 
1 ms async packet is also sent...

Now you should know how delayed the token is to the second host, and fill 
in the table.

Arun
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: PS3, #4
Date: 1 Mar 2002 09:08:09 GMT
Lines: 30
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine27.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3492

It cannot send fractions of packets (this was discussed in class)...so if 
1 ms is available, it will start and finish sending the next packet (even 
if this runs over time). (This is what can cause the observed TRT to be up 
to TTRT + max TRANSP).

Arun

Tim Chao  wrote:
: Can a host transmit only part of a packet while it has the token? For example, if
: the token arrives at HOST A 3ms early, it gets 3ms to send asynchronous data. If
: it has 1ms and 4ms packets to send, will it send the 1ms and also the first 2ms
: of the second packet? or will it only send the 1ms packet?

: Thanks,
: Tim

: Matthew Jonathan Holliman wrote:

:> Xiaoyan Cheng  writes:
:>
:> >"Each host has a maximum allowed allocation of SYN traffic. ", for
:> >example, Host4 has Max SYNC=1ms---should we take this number as the
:> >upper limit of the SUM of all the transmitted SYNC data for all the rows
:> >in each column or just for each token holding time?
:>
:> Keep in mind the guarantees that FDDI makes.  For TTRT to make much sense,
:> the hosts must have agreed upon the maximum amount of synchronous data
:> they'll send during a token rotation.  (Otherwise, how could they really have
:> a target token rotation time?)  So this is what Max SYNC corresponds to.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: PS #5e)
Date: 1 Mar 2002 10:30:16 GMT
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine27.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3493

Yes, in this case the size of the set F = f * N is what we intended.

Arun

Peter Belknap  wrote:
: On 5e, the problem set says 'f is the size of the set F'.  However, that
: can't be right if f * PROP is what the problem says it is.  That would mean
: it takes longer to send the token around in this situation, when it should
: take less time.

: Is it instead the case that the size of the set F is Nf ?  That would seem
: to make more sense...

: -Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: PS #5e)
Date: 1 Mar 2002 10:33:23 GMT
Lines: 30
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine27.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3494

In this case PROP has been redefined to include only those hosts in the 
circuit (fN of them) which have data to send. This is to help simplify the 
final result.

Arun

Derrick Wen-Shiuan Tong  wrote:
: I'm also confused about this question. Is PROP to be interpreted
: as time it takes for the token to travel from C_1 to C_2 ... 
: to C_N ? Or is it simply the time it takes for data to reach 
: all the other nodes (which is shorter now because of the star 
: topology) ?  Question 5e seems to make more sense if PROP has
: the latter interpretation, unless f is defined incorrectly,
: as Pete is suggesting.

: - Derrick


: Peter Belknap  wrote:
: : On 5e, the problem set says 'f is the size of the set F'.  However, that
: : can't be right if f * PROP is what the problem says it is.  That would mean
: : it takes longer to send the token around in this situation, when it should
: : take less time.

: : Is it instead the case that the size of the set F is Nf ?  That would seem
: : to make more sense...

: : -Pete


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: PS3 #2b)
Date: 1 Mar 2002 15:56:04 GMT
Lines: 16
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine31.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3495

I'm still having trouble understanding this problem. How is it that by 
getting a minimum value for a propogation time, we can now lower the value 
for TRANSP? I mean intuitively, the calculation of TRANSP should always be 
a max value (worst case), since we want to make sure that we can react in 
enough time to catch a packet transmitted on the bus while we are 
transmitting. But I just don't see how a minimum prop time value can help 
since, the time it takes to transmit could be greater and thus risk not 
seeing a collided packet.

Any help would be greatly appreciated.

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine42.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS3 #2b)
Date: 1 Mar 2002 19:02:16 GMT
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine42.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3496


>I'm still having trouble understanding this problem. How is it that by 
>getting a minimum value for a propogation time, we can now lower the value 
>for TRANSP? I mean intuitively, the calculation of TRANSP should always be 
>a max value (worst case), since we want to make sure that we can react in 
>enough time to catch a packet transmitted on the bus while we are 
>transmitting. But I just don't see how a minimum prop time value can help 
>since, the time it takes to transmit could be greater and thus risk not 
>seeing a collided packet.


If I answered your question, I'd have done your homework for you :-)

The difference in this problem from regular CSMA/CD is that in addition
to knowing PROP, you also know the smallest PROP between two adjacent
stations--this is more information than you have in the baseline case!
So you can do better--the question is asking how, and by how much?

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine42.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ARP packets
Date: 1 Mar 2002 19:11:52 GMT
Lines: 10
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine42.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3497


>Would it be okay to assume a fixed size for the arp queue structure as
>well?

Since a router implementing this in hardware has a fixed upper bound on
queue lengths due to buffer constraints, it seems reasonable that your
software implementation could as well.  (On the other hand, it's really not
much extra work to support a variable-length queue--should be just a couple
of lines of code, depending on how you implemented things).

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Submission
Date: Fri, 1 Mar 2002 12:06:08 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3498

Hi,

Till now I am not able to submit, I tried many times but I always got this
message:

/usr/class/cs244a/bin/submit assignment: hw3; TA: xwang00
mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00
mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00: Permission
denied at /usr/class/cs244a/bin/submit line 133


Thanks,

Moh'd



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Submission
Date: 1 Mar 2002 22:13:08 GMT
Lines: 7
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3499


>/usr/class/cs244a/bin/submit assignment: hw3; TA: xwang00
>mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00
>mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00: Permission
>denied at /usr/class/cs244a/bin/submit line 133

Can you try again?  Thanks.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine6.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: checksum of the download file
Date: Sat, 2 Mar 2002 01:07:25 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3500

As the "Two announcements for PA#3"
"For ftp, download the "dowhnload_me.*" file
   and either send us the checksum or correctly state what is in the
   file."
But I want to know what "send us the checksum" means. In our program the
checksum is for every packet, so for the "Deliverables" it should mean for
the whole file, right? But how can we calculate the checksum of the whole
file? Does it mean, using our program to download the file and then
calculate it by hand?
Or I made some simple mistakes?

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine8.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: checksum of the download file
Date: 2 Mar 2002 18:25:06 GMT
Lines: 14
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine8.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3501


>As the "Two announcements for PA#3"
>"For ftp, download the "dowhnload_me.*" file
>   and either send us the checksum or correctly state what is in the
>   file."
>But I want to know what "send us the checksum" means. In our program the
>checksum is for every packet, so for the "Deliverables" it should mean for
>the whole file, right? But how can we calculate the checksum of the whole
>file? Does it mean, using our program to download the file and then
>calculate it by hand?


You can use the utility 'cksum'.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: ICMP no route to host
Date: Sat, 2 Mar 2002 21:17:58 -0800
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3502

Hi,

  I'm fishing here for comments/thoughts on testing out ICMP no route to
  host.  I have not been able to get the network to route packets from
  elaine to the vr_server for a nonexistant host which should be on the
  network, if I try, ftp or lynx will just hang until the connections time-out
  without the packets bieng forwarded.  A traceroute shows that they never
  get forwarded to the firewall from Gates-gateway.

  I then modified my source so arp requests are only sent to the gateway.
  When queued packets timeout, an ICMP reply of type ICMP_UNREACH is sent
  back to the sending host.  However, as far as I can tell this has no
  affect on the sending program.  If anyone has had luck testing this out
  I would be very interested to hear about it.

  On another note, tcpdump doesn't seem to be working..

  elaine15:~/projects/router> /usr/class/cs244a/bin/tcpdump
  tcpdump: /dev/hme: Permission denied

  tcpdump would sure be helpful :-)

                                ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: IP Options
Date: Sat, 02 Mar 2002 22:25:51 -0800
Lines: 53
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
CC: Guido Appenzeller 
Xref: nntp.stanford.edu su.class.cs244a:3503

I understand that we won't be supporting IP packets with options in our 
router.  I'd thought that that meant we won't create or receive packets 
with options, but that we'd still be responsible for forwarding such 
packets.  However, I noticed that the vr_ip_hdr struct doesn't have an 
element of type ipoption for storing options.  And the 
VRClient_ProcessPackets function will munge the IP payload if there are 
options.  Here's a trimmed-down version of the relevant code:

         if(SocketFunc(READ, (char *)(&(p->header)), sizeof(IPHeader),
                       "Read IP header") < 0){
             ...
         }

         // payload len is total length - header length
         payload_len = p->header.ip_len - (p->header.ip_hl << 2);

         p->payload = (char *)Malloc(payload_len);

         if(SocketFunc(READ, p->payload, payload_len,
                       "Read IP payload") < 0){
            ...
         }

The first call to SocketFunc reads 20 bytes.  If we had options, the 
header would really be 24 bytes.  But this is ok since the IPPacket 
struct is simply header followed by payload -- so if the options end up 
in the payload, it would all pan out during forwarding.  The problem is 
that payload_len is the real payload length, not including the options 
that you might want to put in the beginning of the "payload".  This is 
because you're subtracting p->header.ip_hl instead of sizeof(IPHeader). 
  So you end up reading four bytes less of the payload than you really 
mean to.

If this project really is restricted to IP packets without options 
(meaning it won't properly handle forwarding arbitrary optionified-IP 
packets), then this won't be a problem.  But since this code is 
referring to ip_hl, it seems like it was intended to work for the 
general case.

So here's the question that pertains to me: should I ignore p->hdr.ip_hl 
and use sizeof(IPHeader) when I'm copying the IP header from my original 
packet? Sounds like I have to because of this bug.  Or am I completely 
misinterpreting all this? :)

-Sandeep

-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga4.Stanford.EDU!ascwu
From: Alex Shao-chiun Wu 
Newsgroups: su.class.cs244a
Subject: Are the www-vr-1, 2, 3 server alive now?
Date: Sat, 2 Mar 2002 23:32:50 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3504

It seems that no ARPs came back from those servers.



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Peter Belknap" 
Newsgroups: su.class.cs244a
Subject: Re: Submission
Date: Sun, 3 Mar 2002 00:58:37 -0800
Lines: 77
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: programminpete.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-Mimeole: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3505

Now I'm getting new errors:

myth5:~/cs244a/cs244aPP3> /usr/class/cs244a/bin/submit hw3 akishan
SubmitBase: /afs/ir/class/cs244a/submissions
Debug Level = 10
/usr/class/cs244a/bin/submit assignment: hw3; TA: akishan
mkdir /afs/ir/class/cs244a/submissions/grading/hw3/akishan/pbelknap

CMD: fs sa /afs/ir/class/cs244a/submissions/grading/hw3/akishan/pbelknap
system:anyuser none 2>/dev/null
PWD: /afs/ir.stanford.edu/users/p/b/pbelknap/cs244a/cs244aPP3


CMD: fs sa /afs/ir/class/cs244a/submissions/grading/hw3/akishan/pbelknap
pbelknap rlid 2>/dev/null
PWD: /afs/ir.stanford.edu/users/p/b/pbelknap/cs244a/cs244aPP3

######################################################################
DEADLINE: 3/8/2002, 17:00
NOW: 2002, 3, 3, 0, 57,15
######################################################################
Requirement: Makefile
Requirement: Makefile
find  . -name . -o -type d -prune -o -name "Makefile" -print
resp = 0 = ./Makefile
Requirement: readme:README:readme.txt:README.TXT
Requirement: readme
find  . -name . -o -type d -prune -o -name "readme" -print
resp = -1 =
Requirement: README
find  . -name . -o -type d -prune -o -name "README" -print
resp = -1 =
Requirement: readme.txt
find  . -name . -o -type d -prune -o -name "readme.txt" -print
resp = -1 =
Requirement: README.TXT
find  . -name . -o -type d -prune -o -name "README.TXT" -print
resp = -1 =
could not find any file matching readme:README:readme.txt:README.TXT in
/afs/ir.stanford.edu/users/p/b/pbelknap/cs244a/cs244aPP3!!
======================================================================
Your code could not be sumitted because of some error.
If you think there is a mistake, please send an email to your TA
(akishan;  with the relevant information.
======================================================================


mailx -s 'Cs244a: pbelknap tried to submit code UNSUCCESSFULLY '

[3-3-2002 @0:57:15] student pbelknap;TA: akishan; submission  /  => tried to
submit code UNSUCCESSFULLY

======================================================================
Your code could not be sumitted because of some error.
If you think there is a mistake, please send an email to your TA
(akishan;  with the relevant information.
======================================================================



Died at /usr/class/cs244a/bin/submit line 192.
myth5:~/cs244a/cs244aPP3>

thanks!
Pete

"Matthew Jonathan Holliman"  wrote in message

>
> >/usr/class/cs244a/bin/submit assignment: hw3; TA: xwang00
> >mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00
> >mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00: Permission
> >denied at /usr/class/cs244a/bin/submit line 133
>
> Can you try again?  Thanks.


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Sandeep Tamhankar 
Newsgroups: su.class.cs244a
Subject: Re: ICMP no route to host
Date: Sun, 03 Mar 2002 09:06:21 -0800
Lines: 50
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine4.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii; format=flowed
Content-Transfer-Encoding: 7bit
User-Agent: Mozilla/5.0 (X11; U; Linux i686; en-US; rv:0.9.7) Gecko/20011221
X-Accept-Language: en-us
Xref: nntp.stanford.edu su.class.cs244a:3506

I haven't done this part yet, but tcpdump does work -- the problem is 
that you're not root on the elaine machines.  Thus, you can only use 
tcpdump with the -r option to look at a dump file.

I'm not sure how you can use that to your advantage.  I assume that you 
can tell that the vr-server is sending out the ICMP_UNREACH.  You might 
want to verify that vr-firewall does indeed get that message.  I don't 
know how to do this since you won't have root there either.

Sorry I couldn't be more helpful.

-Sandeep

Martin Casado wrote:

> Hi,
> 
>   I'm fishing here for comments/thoughts on testing out ICMP no route to
>   host.  I have not been able to get the network to route packets from
>   elaine to the vr_server for a nonexistant host which should be on the
>   network, if I try, ftp or lynx will just hang until the connections time-out
>   without the packets bieng forwarded.  A traceroute shows that they never
>   get forwarded to the firewall from Gates-gateway.
> 
>   I then modified my source so arp requests are only sent to the gateway.
>   When queued packets timeout, an ICMP reply of type ICMP_UNREACH is sent
>   back to the sending host.  However, as far as I can tell this has no
>   affect on the sending program.  If anyone has had luck testing this out
>   I would be very interested to hear about it.
> 
>   On another note, tcpdump doesn't seem to be working..
> 
>   elaine15:~/projects/router> /usr/class/cs244a/bin/tcpdump
>   tcpdump: /dev/hme: Permission denied
> 
>   tcpdump would sure be helpful :-)
> 
>                                 ~~m
> 
> 



-- 
---------------------------------------------
Sandeep V. Tamhankar			
M.S. Student
Computer Science
Email: 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Are the www-vr-1, 2, 3 server alive now?
Date: 3 Mar 2002 21:28:24 GMT
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga3.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3507

As far as I can tell these severs respond correctly to ARP requests as of 
Sunday afternoon.

Arun

Alex Shao-chiun Wu  wrote:
: It seems that no ARPs came back from those servers.



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine3.Stanford.EDU!jinhui
From: Jinhui Pan 
Newsgroups: su.class.cs244a
Subject: traceroute
Date: Sun, 3 Mar 2002 16:00:04 -0800
Lines: 28
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine3.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3508

Hi,

I have trouble to make traceroute work cirrectly.

My router can forward ICMP packet from other routers
correctly. But it fails in construting a time exceeded
ICMP packet.

The wierd thing when I look into packets.dump, I find
that my router actually construct a ICMP packet and
send it to the source host. It shows the packet
has a total length 56, TTL is 64,protocol is ICMP
 checksum for IP header and
ICMP are both correct, ICMP packet has length of 38.

But it seems the source host can not recognize the
ICMP packe. Do I miss something in constructing the
ICMP packet?


Thanks a lot for any comments!

Best,
-jinhui




.

Path: shelby.stanford.edu!nntp.stanford.edu!saga5.Stanford.EDU!ascwu
From: Alex Shao-chiun Wu 
Newsgroups: su.class.cs244a
Subject: One quick question about ARP cache
Date: Sun, 3 Mar 2002 16:09:57 -0800
Lines: 24
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3509

Hi all:

   I don't understand about the description in the assignment page, which
says " all entries in the ARP cache should timeout when **not used** for
more than 5 minutes."

   What does it mean by "not used"??  Should it be "not updated"?

   In my own interpretation:
   "not used" means that this entry is dropped if in the past 5 minutes no
   packets are sent using that entry.

   "not updated" means that this entry should be renew every 5 minutes.


   Another question will be: we can process all incoming ARP response and
update the table, and should we update the timestamp too?  Or we just wait
for 5 minutes and renew the entry, regardless of the fact that we've done
renewal during the 5 minutes by processing someone else's ARP response.


--Alex


.

Path: shelby.stanford.edu!nntp.stanford.edu!saga5.Stanford.EDU!ascwu
From: Alex Shao-chiun Wu 
Newsgroups: su.class.cs244a
Subject: Re: traceroute
Date: Sun, 3 Mar 2002 16:15:08 -0800
Lines: 48
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3510

On Sun, 3 Mar 2002, Jinhui Pan wrote:

It seems that you must have done something wrong with your ICMP
message part if you can see that in packets.dump, so the host issuing
traceroute can't recongnize this message.

According to the previous TA's post, you have to restore everything in the
old packet's IP header (which included in the ICMP message part) . For
example, if you already decremented that TTL, you should restore it to the
previous value, and the same for checksum.

Hope it will help.

--Alex




> Hi,
>
> I have trouble to make traceroute work cirrectly.
>
> My router can forward ICMP packet from other routers
> correctly. But it fails in construting a time exceeded
> ICMP packet.
>
> The wierd thing when I look into packets.dump, I find
> that my router actually construct a ICMP packet and
> send it to the source host. It shows the packet
> has a total length 56, TTL is 64,protocol is ICMP
>  checksum for IP header and
> ICMP are both correct, ICMP packet has length of 38.
>
> But it seems the source host can not recognize the
> ICMP packe. Do I miss something in constructing the
> ICMP packet?
>
>
> Thanks a lot for any comments!
>
> Best,
> -jinhui
>
>
>
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga5.Stanford.EDU!ascwu
From: Alex Shao-chiun Wu 
Newsgroups: su.class.cs244a
Subject: should we get rid of all Debug msg upon submission???
Date: Sun, 3 Mar 2002 16:44:09 -0800
Lines: 2
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3511

As title. Thanx.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: One quick question about ARP cache
Date: 4 Mar 2002 03:12:57 GMT
Lines: 32
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine23.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3512

This is meant to read 5 minutes absolute time. So once you acquire an ARP 
response, you should timestamp it. Now, an ARP entry should be considered 
absent (and thus necessitates a new ARP request) if no ARP entry can be 
found or 5 minutes have elapsed since the ARP entry was acquired.

Arun

Alex Shao-chiun Wu  wrote:
: Hi all:

:    I don't understand about the description in the assignment page, which
: says " all entries in the ARP cache should timeout when **not used** for
: more than 5 minutes."

:    What does it mean by "not used"??  Should it be "not updated"?

:    In my own interpretation:
:    "not used" means that this entry is dropped if in the past 5 minutes no
:    packets are sent using that entry.

:    "not updated" means that this entry should be renew every 5 minutes.


:    Another question will be: we can process all incoming ARP response and
: update the table, and should we update the timestamp too?  Or we just wait
: for 5 minutes and renew the entry, regardless of the fact that we've done
: renewal during the 5 minutes by processing someone else's ARP response.


: --Alex


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: ICMP questions
Date: 4 Mar 2002 05:24:59 GMT
Lines: 32
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine33.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3513

Matthew Jonathan Holliman  wrote:

>>2) The vr_icmp_hdr struct given to us in the code doesn't seem to have room
>>for the "first 64 bits of original datagram" the RFC talks about.  Are we
>>not supposed to worry about that?

> You should support this.  The vr_icmp_hdr structure is probably most
> easily understood by comparing against RFC 792.  It's basically the
> common fields in all headers; the offending datagram whose TTL expired
> would follow this header as the ICMP payload.

I , too , am confused by the vr_icmp_hdr structure. To go through an 
example, let's say we are trying to send an ICMP-Time Exceeded message.
The type, code, and icmp_sum fields occupy 4 bytes of data. The 'hun' 
union which is unused for this message occupies another 4 bytes of data. 
But the union for the apparent next 4 bytes of data doesn't seem 
appropriately defined. It seems there is a struct for a ts, or one for a 
mask, but there is no declaration for data, even though there seems to be 
a #define for 'icmp_data' which should refer to the union dun.data? If we 
used icmp_data, wouldn't this generate a compile error?

I'd like some clarification on this since I'll be adding more data to this
header structure when I add the offending packets IP Header + the first 64
bits of data, which obviously wouldn't fit in this structure.

Thanks in advance for any info.

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine39.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: Testing ICMP Host Unreachable
Date: Mon, 4 Mar 2002 03:07:47 -0800
Lines: 58
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine39.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3514


Hi,

1)   I was wondering what would be a good way to test "ICMP Host
Unreachable". I tried the following, I blocked any arp requests to / any
arp replies from "eth3". So after >= 10 s a packet meant for www-vr-3 will
timeout and send a "ICMP Host Unreachable" packet back to src. What should
be the behavior of src on receiving "ICMP Host Unreachable" packet?
   What I observed was that src continues retrying even after I send a
"ICMP Host Unreachable" packet to it. Is it normal? Well, I dont think so
because then what is the point of having "ICMP Host Unreachable"!!
   The chances that my "ICMP Host Unreachable" packet might be wrong are
pretty slim (as "ICMP TTL Exceeded" works, which is pretty similar plus I
have reviewed it carefully).
   So, why could this be happening??

2)  On looking at the response to traceroute on running "demo_vr" I see
the following:

 1  leland-gateway.Stanford.EDU (171.64.15.97)  1.09 ms  1.23 ms  0.837 ms
 2  Core6-gateway.Stanford.EDU (171.64.1.193)  0.535 ms  0.849 ms  1.00 ms
 3  Gates-gateway.Stanford.EDU (171.64.3.89)  1.78 ms  1.92 ms  1.20 ms
 4  vr-firewall.Stanford.EDU (172.24.74.11)  0.888 ms  0.784 ms  0.837 ms
 5  vr-server-1.Stanford.EDU (172.24.74.20)  3.02 ms (ttl=44!)  2.51 ms
(ttl=44!)  2.43 ms (ttl=44!)
 6  www-vr-1.Stanford.EDU (172.24.74.130)  4.66 ms (ttl=251!)  3.97 ms
(ttl=251!)  *

The return from hop5 shows (ttl=44!) and hop6 shows (ttl=251!).
This error shows up when ttl of an "ICMP time exceeded" packet returned to
the src is not equal to its expected value. The src calculates the
expected value as INITIAL_VALUE_TTL - number of hops.
The router on leland gateway (hop1) uses an INITIAL_VALUE_TTL of 255 (can
be guessed/checked by using -l option with traceroute).

So, such an error popping up for hop5 is not unreasonable because
"demo_vr"(the router at hop5) uses a different INITIAL_VALUE_TTL(=48).

But, this should not be happening for hop6 as it uses
INITIAL_VALUE_TTL(=255). The src sees a ttl of 251 from "ICMP time
exceeded" packet from hop6 whereas it expects a ttl of 250. This can
only happen if some router in the path misbehaves which in this case can
only be "demo_vr", which I beleive is what is happening in this case, it
clearly is not decrementing the ttl of any IP packet(or atleast ICMP
Packet) passing through it. Is my understanding correct?

Gaurav

-------------------------------------------------------------------------
Gaurav Garg                          | Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             |       Escondido Village, Stanford
Department of Electrical Engineering |       CA-94305
Stanford University                  | Ph: 650-498-1208
-------------------------------------------------------------------------

Visit my webpage at http://www.stanford.edu/~ggaurav


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Submission
Date: Mon, 04 Mar 2002 11:41:54 -0800
Lines: 83
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3515

Peter,

just to double check: You do have a README file that is readable?

  Guido

> Now I'm getting new errors:
> 
> myth5:~/cs244a/cs244aPP3> /usr/class/cs244a/bin/submit hw3 akishan
> SubmitBase: /afs/ir/class/cs244a/submissions
> Debug Level = 10
> /usr/class/cs244a/bin/submit assignment: hw3; TA: akishan
> mkdir /afs/ir/class/cs244a/submissions/grading/hw3/akishan/pbelknap
> 
> CMD: fs sa /afs/ir/class/cs244a/submissions/grading/hw3/akishan/pbelknap
> system:anyuser none 2>/dev/null
> PWD: /afs/ir.stanford.edu/users/p/b/pbelknap/cs244a/cs244aPP3
> 
> 
> CMD: fs sa /afs/ir/class/cs244a/submissions/grading/hw3/akishan/pbelknap
> pbelknap rlid 2>/dev/null
> PWD: /afs/ir.stanford.edu/users/p/b/pbelknap/cs244a/cs244aPP3
> 
> ######################################################################
> DEADLINE: 3/8/2002, 17:00
> NOW: 2002, 3, 3, 0, 57,15
> ######################################################################
> Requirement: Makefile
> Requirement: Makefile
> find  . -name . -o -type d -prune -o -name "Makefile" -print
> resp = 0 = ./Makefile
> Requirement: readme:README:readme.txt:README.TXT
> Requirement: readme
> find  . -name . -o -type d -prune -o -name "readme" -print
> resp = -1 =
> Requirement: README
> find  . -name . -o -type d -prune -o -name "README" -print
> resp = -1 =
> Requirement: readme.txt
> find  . -name . -o -type d -prune -o -name "readme.txt" -print
> resp = -1 =
> Requirement: README.TXT
> find  . -name . -o -type d -prune -o -name "README.TXT" -print
> resp = -1 =
> could not find any file matching readme:README:readme.txt:README.TXT in
> /afs/ir.stanford.edu/users/p/b/pbelknap/cs244a/cs244aPP3!!
> ======================================================================
> Your code could not be sumitted because of some error.
> If you think there is a mistake, please send an email to your TA
> (akishan;  with the relevant information.
> ======================================================================
> 
> 
> mailx -s 'Cs244a: pbelknap tried to submit code UNSUCCESSFULLY '
> 
> [3-3-2002 @0:57:15] student pbelknap;TA: akishan; submission  /  => tried
> [to
> submit code UNSUCCESSFULLY
> 
> ======================================================================
> Your code could not be sumitted because of some error.
> If you think there is a mistake, please send an email to your TA
> (akishan;  with the relevant information.
> ======================================================================
> 
> 
> 
> Died at /usr/class/cs244a/bin/submit line 192.
> myth5:~/cs244a/cs244aPP3>
> 
> thanks!
> Pete
> 
> "Matthew Jonathan Holliman"  wrote in message
> 
>>
>> >/usr/class/cs244a/bin/submit assignment: hw3; TA: xwang00
>> >mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00
>> >mkdir /afs/ir/class/cs244a/submissions/grading/hw3/xwang00: Permission
>> >denied at /usr/class/cs244a/bin/submit line 133
>>
>> Can you try again?  Thanks.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: IP Options
Date: Mon, 04 Mar 2002 11:44:33 -0800
Lines: 55
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3516

Sandeep,

this project is restricted to packets without options. The dispatcher 
should only send you packets that have a header length of 5 (which is 20 
bytes).

  Guido

> I understand that we won't be supporting IP packets with options in our
> router.  I'd thought that that meant we won't create or receive packets
> with options, but that we'd still be responsible for forwarding such
> packets.  However, I noticed that the vr_ip_hdr struct doesn't have an
> element of type ipoption for storing options.  And the
> VRClient_ProcessPackets function will munge the IP payload if there are
> options.  Here's a trimmed-down version of the relevant code:
> 
>          if(SocketFunc(READ, (char *)(&(p->header)), sizeof(IPHeader),
>                        "Read IP header") < 0){
>              ...
>          }
> 
>          // payload len is total length - header length
>          payload_len = p->header.ip_len - (p->header.ip_hl << 2);
> 
>          p->payload = (char *)Malloc(payload_len);
> 
>          if(SocketFunc(READ, p->payload, payload_len,
>                        "Read IP payload") < 0){
>             ...
>          }
> 
> The first call to SocketFunc reads 20 bytes.  If we had options, the
> header would really be 24 bytes.  But this is ok since the IPPacket
> struct is simply header followed by payload -- so if the options end up
> in the payload, it would all pan out during forwarding.  The problem is
> that payload_len is the real payload length, not including the options
> that you might want to put in the beginning of the "payload".  This is
> because you're subtracting p->header.ip_hl instead of sizeof(IPHeader).
>   So you end up reading four bytes less of the payload than you really
> mean to.
> 
> If this project really is restricted to IP packets without options
> (meaning it won't properly handle forwarding arbitrary optionified-IP
> packets), then this won't be a problem.  But since this code is
> referring to ip_hl, it seems like it was intended to work for the
> general case.
> 
> So here's the question that pertains to me: should I ignore p->hdr.ip_hl
> and use sizeof(IPHeader) when I'm copying the IP header from my original
> packet? Sounds like I have to because of this bug.  Or am I completely
> misinterpreting all this? :)
> 
> -Sandeep
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: traceroute
Date: Mon, 04 Mar 2002 11:45:31 -0800
Lines: 32
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3517

Jinhui,

did you try running tcmdump wiht "-vv" to check whether the checksums are 
correct?

  Guido

> Hi,
> 
> I have trouble to make traceroute work cirrectly.
> 
> My router can forward ICMP packet from other routers
> correctly. But it fails in construting a time exceeded
> ICMP packet.
> 
> The wierd thing when I look into packets.dump, I find
> that my router actually construct a ICMP packet and
> send it to the source host. It shows the packet
> has a total length 56, TTL is 64,protocol is ICMP
>  checksum for IP header and
> ICMP are both correct, ICMP packet has length of 38.
> 
> But it seems the source host can not recognize the
> ICMP packe. Do I miss something in constructing the
> ICMP packet?
> 
> 
> Thanks a lot for any comments!
> 
> Best,
> -jinhui

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: One quick question about ARP cache
Date: Mon, 04 Mar 2002 11:48:23 -0800
Lines: 34
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3518

Alex,

sorry, this is a bug in the assignment. You should time out ofer 5 minutes 
independently of usage (if you have already submitted we'll accept 5 
minutes after last reference to it).

For updates you can do it either way but the preferred way is to update the 
time stamp every time a packet comes in.

  Guido

> Hi all:
> 
>    I don't understand about the description in the assignment page, which
> says " all entries in the ARP cache should timeout when **not used** for
> more than 5 minutes."
> 
>    What does it mean by "not used"??  Should it be "not updated"?
> 
>    In my own interpretation:
>    "not used" means that this entry is dropped if in the past 5 minutes no
>    packets are sent using that entry.
> 
>    "not updated" means that this entry should be renew every 5 minutes.
> 
> 
>    Another question will be: we can process all incoming ARP response and
> update the table, and should we update the timestamp too?  Or we just wait
> for 5 minutes and renew the entry, regardless of the fact that we've done
> renewal during the 5 minutes by processing someone else's ARP response.
> 
> 
> --Alex

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: should we get rid of all Debug msg upon submission???
Date: Mon, 04 Mar 2002 11:58:18 -0800
Lines: 4
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3519

As long as the program is running fine with the debug output you can leave 
it in.

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP questions
Date: 4 Mar 2002 20:00:57 GMT
Lines: 23
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3520


>I , too , am confused by the vr_icmp_hdr structure. To go through an 
>example, let's say we are trying to send an ICMP-Time Exceeded message.
>The type, code, and icmp_sum fields occupy 4 bytes of data. The 'hun' 
>union which is unused for this message occupies another 4 bytes of data. 
>But the union for the apparent next 4 bytes of data doesn't seem 
>appropriately defined. It seems there is a struct for a ts, or one for a 
>mask, but there is no declaration for data, even though there seems to be 
>a #define for 'icmp_data' which should refer to the union dun.data? If we 
>used icmp_data, wouldn't this generate a compile error?

There's a #define in the header for the ICMP header length for time
exceeded messages; I'd suggest malloc()ing this amount of storage,
plus whatever you need for the payload.  The only fields you care about
in the ICMP header are outside of the unions.

>I'd like some clarification on this since I'll be adding more data to this
>header structure when I add the offending packets IP Header + the first 64
>bits of data, which obviously wouldn't fit in this structure.

You can use your own type, use the #defined constants for header lengths,
etc.--whatever works for you.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP no route to host
Date: 4 Mar 2002 22:15:47 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine37.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3521



>  I'm fishing here for comments/thoughts on testing out ICMP no route to
>  host.  I have not been able to get the network to route packets from
>  elaine to the vr_server for a nonexistant host which should be on the
>  network, if I try, ftp or lynx will just hang until the connections time-out
>  without the packets bieng forwarded.  A traceroute shows that they never
>  get forwarded to the firewall from Gates-gateway.

>  I then modified my source so arp requests are only sent to the gateway.
>  When queued packets timeout, an ICMP reply of type ICMP_UNREACH is sent
>  back to the sending host.  However, as far as I can tell this has no
>  affect on the sending program.  If anyone has had luck testing this out
>  I would be very interested to hear about it.


Lynx behaves as you describe for me, also.  However, if you've (temporarily!)
modified your VR client to report a host as unreachable, you should see "!H"
popping up in the output of a traceroute to that host.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: 4 Mar 2002 22:34:44 GMT
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine37.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3522


>1)   I was wondering what would be a good way to test "ICMP Host
>Unreachable". I tried the following, I blocked any arp requests to / any
>arp replies from "eth3". So after >= 10 s a packet meant for www-vr-3 will
>timeout and send a "ICMP Host Unreachable" packet back to src. What should
>be the behavior of src on receiving "ICMP Host Unreachable" packet?
>   What I observed was that src continues retrying even after I send a
>"ICMP Host Unreachable" packet to it. Is it normal? Well, I dont think so
>because then what is the point of having "ICMP Host Unreachable"!!
>   The chances that my "ICMP Host Unreachable" packet might be wrong are
>pretty slim (as "ICMP TTL Exceeded" works, which is pretty similar plus I
>have reviewed it carefully).
>   So, why could this be happening??


Your strategy seems sound.  As I wrote in response to Martin's similar
question, traceroute might be very helpful in seeing whether your "host
unreachable" implementation is working or not.

Both ftp and lynx behave for me as you describe.  I don't know whether this
is a problem in their implementation, if it's a problem in connect(), etc.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: Mon, 04 Mar 2002 14:41:58 -0800
Lines: 30
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3523

Matthew,Everyone,

I had a look at this and it seems host unreachable icmps disappear 
somewhere. I'll investigate.

  Guido

> 
>>1)   I was wondering what would be a good way to test "ICMP Host
>>Unreachable". I tried the following, I blocked any arp requests to / any
>>arp replies from "eth3". So after >= 10 s a packet meant for www-vr-3 will
>>timeout and send a "ICMP Host Unreachable" packet back to src. What should
>>be the behavior of src on receiving "ICMP Host Unreachable" packet?
>>   What I observed was that src continues retrying even after I send a
>>"ICMP Host Unreachable" packet to it. Is it normal? Well, I dont think so
>>because then what is the point of having "ICMP Host Unreachable"!!
>>   The chances that my "ICMP Host Unreachable" packet might be wrong are
>>pretty slim (as "ICMP TTL Exceeded" works, which is pretty similar plus I
>>have reviewed it carefully).
>>   So, why could this be happening??
> 
> 
> Your strategy seems sound.  As I wrote in response to Martin's similar
> question, traceroute might be very helpful in seeing whether your "host
> unreachable" implementation is working or not.
> 
> Both ftp and lynx behave for me as you describe.  I don't know whether
> this is a problem in their implementation, if it's a problem in connect(),
> etc.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!vasco
From: Vassil Chatalbashev 
Newsgroups: su.class.cs244a
Subject: connecting to server fails
Date: Mon, 4 Mar 2002 20:34:31 -0800
Lines: 18
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3524

I am trying to start the vr client. I get:

>Client vasco connecting to Server 172.24.74.20:12345

And then it hangs...
Is the server down? (the big-brother page showed CPU problems for
www-vr-server-a, but now doesn't, and I still can't connect...)

Please post when everything is back up.

Thanks.




***********************


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: PP3 web page
Date: Mon, 4 Mar 2002 22:39:26 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine33.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3525


Can someone fix the links off the assignment page, so that we can reach
the FAQ in some civilized manner? :) Thanks,

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Phil" 
Newsgroups: su.class.cs244a
Subject: Why strdup?
Date: Mon, 4 Mar 2002 23:22:37 -0800
Lines: 67
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_01E8_01C1C3D3.78B4DFC0"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3526

This is a multi-part message in MIME format.

------=_NextPart_000_01E8_01C1C3D3.78B4DFC0
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

Well, this is not an important issue for the assignment...
but out of curiosity..

The original code of vr_router.c uses  strdup(inet_ntoa( ...))
Is there some reason why we should use strdup here?
It seems to me it's just leaking memory..

And vr_client.c seems to have some leaks..
pac->itfName allocated in line 286
p.itfName allocated in line 439
  are not freed and vr_router.c has no access to them.

Thanks
Hopil Bae


------=_NextPart_000_01E8_01C1C3D3.78B4DFC0
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2>Well, this is not an important issue =
for the=20
assignment...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>but out of curiosity..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>The original code of vr_router.c =
uses =20
strdup(inet_ntoa( ...))</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Is there some reason why we should =
use strdup=20
here?</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>It seems to me it's just leaking=20
memory..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>And vr_client.c seems to have some =

leaks..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>pac->itfName allocated in line =
286</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>p.itfName allocated in line =
439</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>  are not freed and vr_router.c =
has no access=20
to them.</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Thanks</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Hopil Bae</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_01E8_01C1C3D3.78B4DFC0--

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: PP3 web page
Date: Mon, 4 Mar 2002 23:39:34 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Nickolay Boytchev Stanev 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3527

Nickolay, it is fixed. Xin

On Mon, 4 Mar 2002, Nickolay Boytchev Stanev wrote:

> 
> Can someone fix the links off the assignment page, so that we can reach
> the FAQ in some civilized manner? :) Thanks,
> 
> 							NStanev
> -- 
> ####################################################
> # Nickolay Stanev                                  #
> # E-mail:                   #
> # Cell: (650)-269-4756                             #
> # URL: http://www.stanford.edu/~nbstanev/          #
> ####################################################
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Hopil Bae" 
Newsgroups: su.class.cs244a
Subject: About 10 sec timeout for ARP response..
Date: Tue, 5 Mar 2002 01:54:34 -0800
Lines: 75
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_024A_01C1C3E8.B2A596B0"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3528

This is a multi-part message in MIME format.

------=_NextPart_000_024A_01C1C3E8.B2A596B0
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

=20
I see that ARP cache's 5 min timeout may be implemented without using =
timer or signal....
But thinking about 10 sec timeout for ARP response,
   I think I have to use some kind of timer, because we'd better notify =
the source of the error ASAP...
Well, but select() is not feasible.. ( Main loop resides in =
vr_client.c.. I'll have to modify it.. )
   and signal() and sigset() system calls cause read() or write() to =
fail..

I wonder if it is OK just to use the simplistic method as in =
implementing the 5 min timeout..
or there is some other friendly signal mechanism which doesn't break =
read() or write() system calls..
Specifically, I wonder if someone tried sigaction()...

Thanks
Hopil Bae


------=_NextPart_000_024A_01C1C3E8.B2A596B0
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2> </FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I see that ARP cache's 5 min =
timeout may be=20
implemented without using timer or signal....</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>But thinking about 10 sec timeout for =
ARP=20
response,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>   I think I have to use some =
kind of=20
timer, because we'd better notify the source of the error =
ASAP...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Well, but select() is not feasible.. ( =
Main loop=20
resides in vr_client.c.. I'll have to modify it.. )</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>   and signal() and =
sigset()=20
system calls cause read() or write() to fail..</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>I wonder if it is OK just to use the =
simplistic=20
method as in implementing the 5 min timeout..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>or there is some other friendly signal =
mechanism=20
which doesn't break read() or write() system calls..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Specifically, I wonder if someone tried =

sigaction()...</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Thanks</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Hopil Bae</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_024A_01C1C3E8.B2A596B0--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Mudassir I Sheikha" 
Newsgroups: su.class.cs244a
Subject: Deliverables - files from the web server?
Date: Tue, 5 Mar 2002 07:41:24 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: pakistan.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3529

the handout says we have to download files from the webserver....what are
the paths to these files?

mudassir



.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: About 10 sec timeout for ARP response..
Date: Tue, 5 Mar 2002 11:56:51 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Hopil Bae 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3530

Hopil,

You can use the simple solution: only check if there are expired packets
when another packet is added to the queue.

Best,

Xin


On Tue, 5 Mar 2002, Hopil Bae wrote:

>  I see that ARP cache's 5 min timeout may be implemented without using
> timer or signal.... But thinking about 10 sec timeout for ARP
> response,
>    I think I have to use some kind of timer, because we'd better
> notify the source of the error ASAP... Well, but select() is not
> feasible.. ( Main loop resides in vr_client.c.. I'll have to modify
> it.. )
>    and signal() and sigset() system calls cause read() or write() to
> fail..
> 
> I wonder if it is OK just to use the simplistic method as in
> implementing the 5 min timeout.. or there is some other friendly
> signal mechanism which doesn't break read() or write() system calls..
> Specifically, I wonder if someone tried sigaction()...
> 
> Thanks
> Hopil Bae
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: connecting to server fails
Date: Tue, 05 Mar 2002 13:31:25 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3531

Vassil,

actually it was not down but a student had a client that sent several 
thousand ARP packets *per second* and this basically locked up the 
dispatcher for everyone else. Once the client disconnected things
went back to normal. I guess our VR is not quite robust against DOS
attacks yet :-)

  Guido



> I am trying to start the vr client. I get:
> 
>>Client vasco connecting to Server 172.24.74.20:12345
> 
> And then it hangs...
> Is the server down? (the big-brother page showed CPU problems for
> www-vr-server-a, but now doesn't, and I still can't connect...)
> 
> Please post when everything is back up.
> 
> Thanks.
> 
> 
> 
> 
> ***********************

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Deliverables - files from the web server?
Date: Tue, 5 Mar 2002 13:34:55 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Mudassir I Sheikha 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3532

Mudassir,

As Guido's post "two announcements for PA#3" said:

"For the web server simply test if you are seeing the default Apache web
page. For ftp, download the "dowhnload_me.*" file and either send us the
checksum or correctly state what is in the file. For host www-vr-1 the
correct answer would be a rendering of the multicast backbone. For the
other two it should be easier to identify the contents. (Hint: Files that
end in .au are Sun audio files :-)"

Xin



On Tue, 5 Mar 2002, Mudassir I Sheikha wrote:

> the handout says we have to download files from the webserver....what are
> the paths to these files?
> 
> mudassir
> 
> 
> 
> 


-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic20.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Deliverables - files from the web server?
Date: Tue, 5 Mar 2002 21:31:13 -0800
Lines: 44
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic20.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Xin Wang 
cc: Mudassir I Sheikha 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3533

Hi,
Can you please clarify what do u mean by rendering of the muticast
backbone for www-vr-1. I am kind of confused with this word.
Shankar

On Tue, 5 Mar 2002, Xin Wang wrote:

> Mudassir,
>
> As Guido's post "two announcements for PA#3" said:
>
> "For the web server simply test if you are seeing the default Apache web
> page. For ftp, download the "dowhnload_me.*" file and either send us the
> checksum or correctly state what is in the file. For host www-vr-1 the
> correct answer would be a rendering of the multicast backbone. For the
> other two it should be easier to identify the contents. (Hint: Files that
> end in .au are Sun audio files :-)"
>
> Xin
>
>
>
> On Tue, 5 Mar 2002, Mudassir I Sheikha wrote:
>
> > the handout says we have to download files from the webserver....what are
> > the paths to these files?
> >
> > mudassir
> >
> >
> >
> >
>
>
> --
> ---------------------------------
> Xin Wang
>
> Department of Computer Science
> Stanford University
>
> 
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Deliverables - files from the web server?
Date: Tue, 5 Mar 2002 22:32:28 -0800
Lines: 63
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
cc: Xin Wang  
    Mudassir I Sheikha 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3534

Hi,

It is just some hint about the content of the file :)

Xin


On Tue, 5 Mar 2002, Shankar Agarwal wrote:

> Hi,
> Can you please clarify what do u mean by rendering of the muticast
> backbone for www-vr-1. I am kind of confused with this word.
> Shankar
> 
> On Tue, 5 Mar 2002, Xin Wang wrote:
> 
> > Mudassir,
> >
> > As Guido's post "two announcements for PA#3" said:
> >
> > "For the web server simply test if you are seeing the default Apache web
> > page. For ftp, download the "dowhnload_me.*" file and either send us the
> > checksum or correctly state what is in the file. For host www-vr-1 the
> > correct answer would be a rendering of the multicast backbone. For the
> > other two it should be easier to identify the contents. (Hint: Files that
> > end in .au are Sun audio files :-)"
> >
> > Xin
> >
> >
> >
> > On Tue, 5 Mar 2002, Mudassir I Sheikha wrote:
> >
> > > the handout says we have to download files from the webserver....what are
> > > the paths to these files?
> > >
> > > mudassir
> > >
> > >
> > >
> > >
> >
> >
> > --
> > ---------------------------------
> > Xin Wang
> >
> > Department of Computer Science
> > Stanford University
> >
> > 
> >
> 
> 

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: Wed, 6 Mar 2002 02:44:49 -0800
Lines: 50
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3535

I've tried the same approach as Gaurav with traceroute, but traceroute is
not reporting host unreachable. Is there something I did wrong? Or
traceroute is simply not reporting host unreachable?

Guido, you mention "host unreachable icmps disappear somewhere" in your
previous post. So what's the status for that now?

Any thought is greatly appreciated.

-Henry

"Guido Appenzeller"  wrote in message

> Matthew,Everyone,
>
> I had a look at this and it seems host unreachable icmps disappear
> somewhere. I'll investigate.
>
>   Guido
>
> >
> >>1)   I was wondering what would be a good way to test "ICMP Host
> >>Unreachable". I tried the following, I blocked any arp requests to / any
> >>arp replies from "eth3". So after >= 10 s a packet meant for www-vr-3
will
> >>timeout and send a "ICMP Host Unreachable" packet back to src. What
should
> >>be the behavior of src on receiving "ICMP Host Unreachable" packet?
> >>   What I observed was that src continues retrying even after I send a
> >>"ICMP Host Unreachable" packet to it. Is it normal? Well, I dont think
so
> >>because then what is the point of having "ICMP Host Unreachable"!!
> >>   The chances that my "ICMP Host Unreachable" packet might be wrong are
> >>pretty slim (as "ICMP TTL Exceeded" works, which is pretty similar plus
I
> >>have reviewed it carefully).
> >>   So, why could this be happening??
> >
> >
> > Your strategy seems sound.  As I wrote in response to Martin's similar
> > question, traceroute might be very helpful in seeing whether your "host
> > unreachable" implementation is working or not.
> >
> > Both ftp and lynx behave for me as you describe.  I don't know whether
> > this is a problem in their implementation, if it's a problem in
connect(),
> > etc.
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine22.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: 6 Mar 2002 18:48:09 GMT
Lines: 21
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine22.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3536


>I've tried the same approach as Gaurav with traceroute, but traceroute is
>not reporting host unreachable. Is there something I did wrong? Or
>traceroute is simply not reporting host unreachable?

>Guido, you mention "host unreachable icmps disappear somewhere" in your
>previous post. So what's the status for that now?


It sounds like you may have done something wrong--at least traceroute
definitely shows one fewer entry and a host unreachable indication if
my VR client returns "ICMP host unreachable" for a host.

As far as I know, the ICMPs are arriving fine--they seem to be for me--but
if you try to test your host unreachable implementation by contacting a
non-existent host on one of the VR subnets, that looks like it's filtered
out by the firewall.  I think this is what Guido was referring to.  Unless
or until this is fixed (presuming this is the problem), the simplest way
to test your implementation is to force your code to return this error
in some cases.  (Just remember to take this test code out when you submit!)

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine22.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Why strdup?
Date: 6 Mar 2002 18:55:21 GMT
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine22.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3537


>The original code of vr_router.c uses  strdup(inet_ntoa( ...))
>Is there some reason why we should use strdup here?
>It seems to me it's just leaking memory..

I presume you're referring to the Debug message in VRClient_ForwardPacket()?
Remember that inet_ntoa returns a pointer to a buffer that is overwritten
on each call.  Since it's called twice in the debug message, the first
result must be saved.  (Based on this, there's a strdup for each call,
since the order in which the calls are evaluated is undefined).

>And vr_client.c seems to have some leaks..
>pac->itfName allocated in line 286
>p.itfName allocated in line 439
>  are not freed and vr_router.c has no access to them.

Could be :-)

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: Wed, 06 Mar 2002 16:09:51 -0800
Lines: 39
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3538

Hi evertyone,

forget my earlier comment, host unreachable works and gets through the fw 
just fine. Testing it is a little trickier though. Try the following and it 
should work:

- Just using a nonexistant host such as 172.24.74.131 won't work for
  complex reasons you don't have to understand. I have now set up
  the ip address 172.24.74.133. Use it for testing and everything
  should be fine.
  [The reason why it doesn't work is that vr-firewall doesn't have proxy
  arp entries for the other ip addresses. If you really want to know why
  it needs that read the VR docs].
- Telnet won't report anything so it is not much good for testing. Instead
  use traceroute. For traceroute testing to work make sure the ARP Queue
  time out is set to something very low (maximum 1 second). Then run
  traceroute and you should get something like:

-- snip --
saga4:~> traceroute -p 8000 172.24.74.133
traceroute to 172.24.74.133 (172.24.74.133): 1-30 hops, 38 byte packets
 1  leland-gateway.Stanford.EDU (171.64.15.129)  1.37 ms  1.24 ms  1.58 ms
 2  Core6-gateway.Stanford.EDU (171.64.1.193)  0.831 ms  0.902 ms  0.513 ms
 3  Gates-gateway.Stanford.EDU (171.64.3.89)  1.40 ms  1.31 ms  1.42 ms
 4  vr-firewall.Stanford.EDU (172.24.74.11)  0.754 ms  1.20 ms  1.34 ms
 5  vr-server-a-1.Stanford.EDU (172.24.74.18)  97.6 ms  8.14 ms  2.70 ms
 6  vr-server-a-1.Stanford.EDU (172.24.74.18)  219 ms (ttl=251!) !H  990 ms 
(ttl=251!) !H  990 ms (ttl=251!) !H
--- snip ---

  Note the "!H" denoting host unreachable. For your final submission set ARP
  time out back to 5 seconds.

- The demo_vr client uses the correct ARP time out, it can't be used for
  testing.

I hope that helps,

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Hopil Bae" 
Newsgroups: su.class.cs244a
Subject: recvd close session..
Date: Wed, 6 Mar 2002 17:05:08 -0800
Lines: 62
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_0026_01C1C531.11DDC320"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3539

This is a multi-part message in MIME format.

------=_NextPart_000_0026_01C1C531.11DDC320
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


Sometimes my vr prints "recvd close session." message and then =
terminates..
It seems that the vr server is sending CLOSE_SESSION command...
But then, when does it send it?
I guessed that vr server is imposing some timeout on vr clients..
but sometimes I could see it very soon after I started my client..
So I'm a little afraid that vr server may also send it
   when my vr client is doing something wrong..
Could I receive some comments on this?

Hopil Bae


------=_NextPart_000_0026_01C1C531.11DDC320
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>Sometimes </FONT><FONT face=3DArial =
size=3D2>my vr=20
prints "recvd close session." message and then terminates..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>It seems that the vr server is sending=20
CLOSE_SESSION command...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>But then, when does it send =
it?</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I guessed that vr server is imposing =
some timeout=20
on vr clients..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>but sometimes I could see it very soon =
after I=20
started my client..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>So I'm a little afraid that vr =
server may also=20
send it</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>   when my vr client is =
doing=20
</FONT><FONT face=3DArial size=3D2>something wrong..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Could I receive some comments on =
this?</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Hopil Bae</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_0026_01C1C531.11DDC320--

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: Wed, 6 Mar 2002 18:22:24 -0800
Lines: 53
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3540

I've managed to get the (!H) traceroute output following the below
instructions, but only from time to time, not all the time. Is there
something I did wrong?

Henry

"Guido Appenzeller"  wrote in message

> Hi evertyone,
>
> forget my earlier comment, host unreachable works and gets through the fw
> just fine. Testing it is a little trickier though. Try the following and
it
> should work:
>
> - Just using a nonexistant host such as 172.24.74.131 won't work for
>   complex reasons you don't have to understand. I have now set up
>   the ip address 172.24.74.133. Use it for testing and everything
>   should be fine.
>   [The reason why it doesn't work is that vr-firewall doesn't have proxy
>   arp entries for the other ip addresses. If you really want to know why
>   it needs that read the VR docs].
> - Telnet won't report anything so it is not much good for testing. Instead
>   use traceroute. For traceroute testing to work make sure the ARP Queue
>   time out is set to something very low (maximum 1 second). Then run
>   traceroute and you should get something like:
>
> -- snip --
> saga4:~> traceroute -p 8000 172.24.74.133
> traceroute to 172.24.74.133 (172.24.74.133): 1-30 hops, 38 byte packets
>  1  leland-gateway.Stanford.EDU (171.64.15.129)  1.37 ms  1.24 ms  1.58 ms
>  2  Core6-gateway.Stanford.EDU (171.64.1.193)  0.831 ms  0.902 ms  0.513
ms
>  3  Gates-gateway.Stanford.EDU (171.64.3.89)  1.40 ms  1.31 ms  1.42 ms
>  4  vr-firewall.Stanford.EDU (172.24.74.11)  0.754 ms  1.20 ms  1.34 ms
>  5  vr-server-a-1.Stanford.EDU (172.24.74.18)  97.6 ms  8.14 ms  2.70 ms
>  6  vr-server-a-1.Stanford.EDU (172.24.74.18)  219 ms (ttl=251!) !H  990
ms
> (ttl=251!) !H  990 ms (ttl=251!) !H
> --- snip ---
>
>   Note the "!H" denoting host unreachable. For your final submission set
ARP
>   time out back to 5 seconds.
>
> - The demo_vr client uses the correct ARP time out, it can't be used for
>   testing.
>
> I hope that helps,
>
>   Guido


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: recvd close session..
Date: Wed, 06 Mar 2002 18:38:03 -0800
Lines: 21
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3541

Hopil,

this can happen if the server is restarted, which is once per hour. It 
could also happen during certain error conditions but as far as I can see 
from the log that was not the case.

How often did you experience this?

  Guido

> Sometimes my vr prints "recvd close session." message and then
> terminates.. It seems that the vr server is sending CLOSE_SESSION
> command... But then, when does it send it?
> I guessed that vr server is imposing some timeout on vr clients..
> but sometimes I could see it very soon after I started my client..
> So I'm a little afraid that vr server may also send it
>    when my vr client is doing something wrong..
> Could I receive some comments on this?
> 
> Hopil Bae

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: Wed, 06 Mar 2002 18:39:16 -0800
Lines: 62
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3542

Henry,

I guess there is something wrong, it should be deterministic. The load on 
the dispatcher was a little high this afternoon. Maybe try again and also
decrease the ARP timeout to well below a second.

  Guido

> I've managed to get the (!H) traceroute output following the below
> instructions, but only from time to time, not all the time. Is there
> something I did wrong?
> 
> Henry
> 
> "Guido Appenzeller"  wrote in message
> 
>> Hi evertyone,
>>
>> forget my earlier comment, host unreachable works and gets through the fw
>> just fine. Testing it is a little trickier though. Try the following and
> it
>> should work:
>>
>> - Just using a nonexistant host such as 172.24.74.131 won't work for
>>   complex reasons you don't have to understand. I have now set up
>>   the ip address 172.24.74.133. Use it for testing and everything
>>   should be fine.
>>   [The reason why it doesn't work is that vr-firewall doesn't have proxy
>>   arp entries for the other ip addresses. If you really want to know why
>>   it needs that read the VR docs].
>> - Telnet won't report anything so it is not much good for testing.
>> Instead
>>   use traceroute. For traceroute testing to work make sure the ARP Queue
>>   time out is set to something very low (maximum 1 second). Then run
>>   traceroute and you should get something like:
>>
>> -- snip --
>> saga4:~> traceroute -p 8000 172.24.74.133
>> traceroute to 172.24.74.133 (172.24.74.133): 1-30 hops, 38 byte packets
>>  1  leland-gateway.Stanford.EDU (171.64.15.129)  1.37 ms  1.24 ms  1.58
>>  ms
>>  2  Core6-gateway.Stanford.EDU (171.64.1.193)  0.831 ms  0.902 ms  0.513
> ms
>>  3  Gates-gateway.Stanford.EDU (171.64.3.89)  1.40 ms  1.31 ms  1.42 ms
>>  4  vr-firewall.Stanford.EDU (172.24.74.11)  0.754 ms  1.20 ms  1.34 ms
>>  5  vr-server-a-1.Stanford.EDU (172.24.74.18)  97.6 ms  8.14 ms  2.70 ms
>>  6  vr-server-a-1.Stanford.EDU (172.24.74.18)  219 ms (ttl=251!) !H  990
> ms
>> (ttl=251!) !H  990 ms (ttl=251!) !H
>> --- snip ---
>>
>>   Note the "!H" denoting host unreachable. For your final submission set
> ARP
>>   time out back to 5 seconds.
>>
>> - The demo_vr client uses the correct ARP time out, it can't be used for
>>   testing.
>>
>> I hope that helps,
>>
>>   Guido

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: Wed, 6 Mar 2002 19:38:29 -0800
Lines: 81
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3543

Looks like if a set the ARP timeout to 250ms then it works all the time.

Thanks,
Henry

"Guido Appenzeller"  wrote in message

> Henry,
>
> I guess there is something wrong, it should be deterministic. The load on
> the dispatcher was a little high this afternoon. Maybe try again and also
> decrease the ARP timeout to well below a second.
>
>   Guido
>
> > I've managed to get the (!H) traceroute output following the below
> > instructions, but only from time to time, not all the time. Is there
> > something I did wrong?
> >
> > Henry
> >
> > "Guido Appenzeller"  wrote in message
> > 
> >> Hi evertyone,
> >>
> >> forget my earlier comment, host unreachable works and gets through the
fw
> >> just fine. Testing it is a little trickier though. Try the following
and
> > it
> >> should work:
> >>
> >> - Just using a nonexistant host such as 172.24.74.131 won't work for
> >>   complex reasons you don't have to understand. I have now set up
> >>   the ip address 172.24.74.133. Use it for testing and everything
> >>   should be fine.
> >>   [The reason why it doesn't work is that vr-firewall doesn't have
proxy
> >>   arp entries for the other ip addresses. If you really want to know
why
> >>   it needs that read the VR docs].
> >> - Telnet won't report anything so it is not much good for testing.
> >> Instead
> >>   use traceroute. For traceroute testing to work make sure the ARP
Queue
> >>   time out is set to something very low (maximum 1 second). Then run
> >>   traceroute and you should get something like:
> >>
> >> -- snip --
> >> saga4:~> traceroute -p 8000 172.24.74.133
> >> traceroute to 172.24.74.133 (172.24.74.133): 1-30 hops, 38 byte packets
> >>  1  leland-gateway.Stanford.EDU (171.64.15.129)  1.37 ms  1.24 ms  1.58
> >>  ms
> >>  2  Core6-gateway.Stanford.EDU (171.64.1.193)  0.831 ms  0.902 ms
0.513
> > ms
> >>  3  Gates-gateway.Stanford.EDU (171.64.3.89)  1.40 ms  1.31 ms  1.42 ms
> >>  4  vr-firewall.Stanford.EDU (172.24.74.11)  0.754 ms  1.20 ms  1.34 ms
> >>  5  vr-server-a-1.Stanford.EDU (172.24.74.18)  97.6 ms  8.14 ms  2.70
ms
> >>  6  vr-server-a-1.Stanford.EDU (172.24.74.18)  219 ms (ttl=251!) !H
990
> > ms
> >> (ttl=251!) !H  990 ms (ttl=251!) !H
> >> --- snip ---
> >>
> >>   Note the "!H" denoting host unreachable. For your final submission
set
> > ARP
> >>   time out back to 5 seconds.
> >>
> >> - The demo_vr client uses the correct ARP time out, it can't be used
for
> >>   testing.
> >>
> >> I hope that helps,
> >>
> >>   Guido
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jing Jiang" 
Newsgroups: su.class.cs244a
Subject: ICMP
Date: Wed, 6 Mar 2002 20:54:37 -0800
Lines: 32
Distribution: su
Message-ID: 
NNTP-Posting-Host: jjiang.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2314.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
Xref: nntp.stanford.edu su.class.cs244a:3544

"As section 4.3 of RFC 1812 says, we should follow (2) and send ICMP
with the original IP header (TTL = 1). Xin"

I assume for ICMP Destination Unreachable Message, we also send with the
original TTL and checksum? Thanks!

-Jing

On Sun, 24 Feb 2002, Dhawal Kumar wrote:

> Do we send ICMP time exceeded message if
> (1) TTL = 0 in the received datagram or
> (2) TTL = 0 after decrementing?
>
> In case we are following (2) do we send ICMP with
> original IP header (with TTL = 1) or the original IP Header with modified
> TTL (i.e. TTL set to 0) in the ICMP payload?
>
> Dhawal Kumar
>
>

--
---------------------------------
Xin Wang

Department of Computer Science
Stanford University




.

Path: shelby.stanford.edu!nntp.stanford.edu!saga6.Stanford.EDU!hershd
From: Hersh Satyanarayan Dangayach 
Newsgroups: su.class.cs244a
Subject: purify problem
Date: Wed, 6 Mar 2002 21:38:14 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3545

Hi,

I am getting the following error when I try using purify. Any ideas?

Purify slave: Error: Can't access cache sub-directory "/tmp/usr/lib" --
check permissions.
System error code: Permission denied
Purify: Read from rtslave failed, rtslave shutdown.


Thanks,
Hersh


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: cant connect to server
Date: Wed, 6 Mar 2002 21:55:52 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine37.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3546

hi,

i have not been able to connect to the server for about an hour now.

It says:

=============
VRClient Starting.
Client anuragg connecting to Server 172.24.74.20:-1407694319
Connect to server error: Connection refused
=============

Any clues whats wrong with the server (or my code)?

thanks
-anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine41.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: how to test ftp?
Date: Wed, 6 Mar 2002 23:47:26 -0800
Lines: 69
Distribution: su
Message-ID: 
References: 
 
 
 
NNTP-Posting-Host: elaine41.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3547

Hi, could someone let me know how to test ftp.  Thanks!

On Tue, 5 Mar 2002, Xin Wang wrote:

> Hi,
>
> It is just some hint about the content of the file :)
>
> Xin
>
>
> On Tue, 5 Mar 2002, Shankar Agarwal wrote:
>
> > Hi,
> > Can you please clarify what do u mean by rendering of the muticast
> > backbone for www-vr-1. I am kind of confused with this word.
> > Shankar
> >
> > On Tue, 5 Mar 2002, Xin Wang wrote:
> >
> > > Mudassir,
> > >
> > > As Guido's post "two announcements for PA#3" said:
> > >
> > > "For the web server simply test if you are seeing the default Apache web
> > > page. For ftp, download the "dowhnload_me.*" file and either send us the
> > > checksum or correctly state what is in the file. For host www-vr-1 the
> > > correct answer would be a rendering of the multicast backbone. For the
> > > other two it should be easier to identify the contents. (Hint: Files that
> > > end in .au are Sun audio files :-)"
> > >
> > > Xin
> > >
> > >
> > >
> > > On Tue, 5 Mar 2002, Mudassir I Sheikha wrote:
> > >
> > > > the handout says we have to download files from the webserver....what are
> > > > the paths to these files?
> > > >
> > > > mudassir
> > > >
> > > >
> > > >
> > > >
> > >
> > >
> > > --
> > > ---------------------------------
> > > Xin Wang
> > >
> > > Department of Computer Science
> > > Stanford University
> > >
> > > 
> > >
> >
> >
>
> --
> ---------------------------------
> Xin Wang
>
> Department of Computer Science
> Stanford University
>
> 
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine41.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: traceroute
Date: Wed, 6 Mar 2002 23:57:26 -0800
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine41.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3548

Hi there,
I got the result for traceroute.  I am wondering what the * * * for hop 5
is.  any hint?
Thanks!

elaine41:~> traceroute -p 8020 www-vr-1
traceroute to www-vr-1.Stanford.EDU (172.24.74.130): 1-30 hops, 38 byte
packets
 1  leland-gateway.Stanford.EDU (171.64.15.97)  1.22 ms  1.99 ms  0.744 ms
 2  Core2-gateway.Stanford.EDU (171.64.1.233)  0.749 ms  0.791 ms  0.616
ms
 3  Gates-gateway.Stanford.EDU (171.64.3.41)  1.66 ms  2.96 ms  1.73 ms
 4  vr-firewall.Stanford.EDU (172.24.74.11)  0.838 ms  0.964 ms  0.947 ms
 5  *  *  *
 6  www-vr-1.Stanford.EDU (172.24.74.130)  11.1 ms  23.1 ms  10.3 ms



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Clayton Pierce Jones 
Newsgroups: su.class.cs244a
Subject: Re: traceroute
Date: Thu, 07 Mar 2002 00:23:26 -0800
Lines: 22
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: fable20.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3549

It means you are forwarding packets to the server okay, but you are not
returning valid icmp timeout responses.

Clayton

Huan Wanda Jiang wrote:
> 
> Hi there,
> I got the result for traceroute.  I am wondering what the * * * for hop 5
> is.  any hint?
> Thanks!
> 
> elaine41:~> traceroute -p 8020 www-vr-1
> traceroute to www-vr-1.Stanford.EDU (172.24.74.130): 1-30 hops, 38 byte
> packets
>  1  leland-gateway.Stanford.EDU (171.64.15.97)  1.22 ms  1.99 ms  0.744 ms
>  2  Core2-gateway.Stanford.EDU (171.64.1.233)  0.749 ms  0.791 ms  0.616
> ms
>  3  Gates-gateway.Stanford.EDU (171.64.3.41)  1.66 ms  2.96 ms  1.73 ms
>  4  vr-firewall.Stanford.EDU (172.24.74.11)  0.838 ms  0.964 ms  0.947 ms
>  5  *  *  *
>  6  www-vr-1.Stanford.EDU (172.24.74.130)  11.1 ms  23.1 ms  10.3 ms
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Bryan Paul Alfaro 
Newsgroups: su.class.cs244a
Subject: connecting to ftp thru www-vr-1,2,3
Date: Thu, 07 Mar 2002 00:30:48 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: raptor12.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; Linux 2.4.9-13 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3550

how exactly are we suppose to connect to the www-vr sites for ftp use?
i download web pages without a problem but hang on the ftp

here is what i am doing:



then it asks me for a password and give anonymous

am i going about this the wrong way

any help would be appreciated

thanks,
bryan
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jing Jiang" 
Newsgroups: su.class.cs244a
Subject: vr_protocol.h file
Date: Thu, 7 Mar 2002 01:53:16 -0800
Lines: 7
Distribution: su
Message-ID: 
NNTP-Posting-Host: jjiang.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.00.2314.1300
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2314.1300
Xref: nntp.stanford.edu su.class.cs244a:3551

I noticed that the vr_protocol.h file is now different from what I copied
from the class directory before. Can we still use the old version? My
definition of ICMP packet depends on the old file. Thanks!

-Jing


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Henry Fu" 
Newsgroups: su.class.cs244a
Subject: Re: connecting to ftp thru www-vr-1,2,3
Date: Thu, 7 Mar 2002 02:30:03 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: bernese.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3552

Bryan,

You need to connect to the www-vr-1, www-vr-2, www-vr-3 in elaine or saga
machine.
Then, you can try to connect by typing "ftp www-vr-1.stanford.edu 8002" in
the shell.

Hope it helps,
Henry


"Bryan Paul Alfaro"  wrote in message

> how exactly are we suppose to connect to the www-vr sites for ftp use?
> i download web pages without a problem but hang on the ftp
>
> here is what i am doing:
>
> 
>
> then it asks me for a password and give anonymous
>
> am i going about this the wrong way
>
> any help would be appreciated
>
> thanks,
> bryan


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: One quick question about ARP cache
Date: 7 Mar 2002 15:30:09 GMT
Lines: 53
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine22.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3553

Hello,

I'm still not clear about the timing out of ARP caches. Do we timeout only 
the entry that is over 5 minutes, or do we flush our entire cache if 
there's an entry over 5 minutes old?
Also, since we don't have to implement timers/multiple threads, how often 
do we check to see when we should flush entries from the ARP cache? I'm 
guessing we should check when we receive new ARP requests, but what about 
when we just receive ANY packet?

Thanks in advance for any clarification.

Guido Appenzeller  wrote:
> Alex,

> sorry, this is a bug in the assignment. You should time out ofer 5 minutes 
> independently of usage (if you have already submitted we'll accept 5 
> minutes after last reference to it).

> For updates you can do it either way but the preferred way is to update the 
> time stamp every time a packet comes in.

>   Guido

>> Hi all:
>> 
>>    I don't understand about the description in the assignment page, which
>> says " all entries in the ARP cache should timeout when **not used** for
>> more than 5 minutes."
>> 
>>    What does it mean by "not used"??  Should it be "not updated"?
>> 
>>    In my own interpretation:
>>    "not used" means that this entry is dropped if in the past 5 minutes no
>>    packets are sent using that entry.
>> 
>>    "not updated" means that this entry should be renew every 5 minutes.
>> 
>> 
>>    Another question will be: we can process all incoming ARP response and
>> update the table, and should we update the timestamp too?  Or we just wait
>> for 5 minutes and renew the entry, regardless of the fact that we've done
>> renewal during the 5 minutes by processing someone else's ARP response.
>> 
>> 
>> --Alex


-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: One quick question about ARP cache
Date: Thu, 7 Mar 2002 07:59:25 -0800
Lines: 23
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3554

On 7 Mar 2002, Victor Tung wrote:

> Hello,
>
> I'm still not clear about the timing out of ARP caches. Do we timeout only
> the entry that is over 5 minutes, or do we flush our entire cache if
> there's an entry over 5 minutes old?

You should clear entries piecewise that are 5 minutes old or older (not the
entire cache).

> Also, since we don't have to implement timers/multiple threads, how often
> do we check to see when we should flush entries from the ARP cache?

One suggestion is to check whether or not to prune your arp queue/cache
everytime you receive a packet (any packet).

You should take a peak at the FAQ #14.


                                HTH,
                                ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine38.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP
Date: 7 Mar 2002 18:00:19 GMT
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine38.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3555

"Jing Jiang"  writes:

>"As section 4.3 of RFC 1812 says, we should follow (2) and send ICMP
>with the original IP header (TTL = 1). Xin"

>I assume for ICMP Destination Unreachable Message, we also send with the
>original TTL and checksum? Thanks!


Yep.  You can look at section 4.3.2.3 in RFC 1812 for a discussion of ICMP
message handling.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine38.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: how to test ftp?
Date: 7 Mar 2002 18:03:16 GMT
Lines: 6
Distribution: su
Message-ID: 
References:        
NNTP-Posting-Host: elaine38.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3556


>Hi, could someone let me know how to test ftp.  Thanks!

ftp www-vr-[1-3] N
where N is your assigned port number.

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: vr_protocol.h file
Date: Thu, 07 Mar 2002 12:01:18 -0800
Lines: 10
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3557

Jing,

> I noticed that the vr_protocol.h file is now different from what I copied
> from the class directory before. Can we still use the old version? My
> definition of ICMP packet depends on the old file. Thanks!

Yes, that's ok, just make sure you state this clearly in your README. The 
differences should be pretty minimal anyway.

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: connecting to ftp thru www-vr-1,2,3
Date: Thu, 07 Mar 2002 12:01:59 -0800
Lines: 8
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3558

Bryan,

> how exactly are we suppose to connect to the www-vr sites for ftp use?
> i download web pages without a problem but hang on the ftp

Did you check out the FAQ (active vs. passive mode)?

  Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: cant connect to server
Date: Thu, 07 Mar 2002 12:04:32 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: dynamogen.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7Bit
User-Agent: KNode/0.6.1
Xref: nntp.stanford.edu su.class.cs244a:3559

Anurag,

this is strange, according to the log plenty of users used the system 
during this time. What machine were you connecting from? Did the
sample client work? Did anyone else experience a longer (> 1 min) outage
during this time?

  Guido

> hi,
> 
> i have not been able to connect to the server for about an hour now.
> 
> It says:
> 
> =============
> VRClient Starting.
> Client anuragg connecting to Server 172.24.74.20:-1407694319
> Connect to server error: Connection refused
> =============
> 
> Any clues whats wrong with the server (or my code)?
> 
> thanks
> -anurag

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine32.Stanford.EDU!wjiang
From: Huan Wanda Jiang 
Newsgroups: su.class.cs244a
Subject: what if ICMP packet itself is timed out in the queue?
Date: Thu, 7 Mar 2002 16:58:24 -0800
Lines: 5
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine32.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3560

Hi there,
I just drop the icmp packet if it is timed out.  Would that be ok?

Thanks!

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine13.Stanford.EDU!homa
From: Alex Khomenko 
Newsgroups: su.class.cs244a
Subject: download_me_very_slowly.au (1509006 bytes)
Date: Thu, 7 Mar 2002 18:13:53 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine13.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3561


I'm downloading download_me.au from www-vr-3, and it is very slow. It will
definitely take > 10 minutes. Anyone cares to share how long it took them?
:>

Thanks,
Alex

-- 
Knowledge is a deadly friend when no one sets the rules.
The fate of all mankind I see is in the hands of fools. (KC "Epitaph")

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: Re: download_me_very_slowly.au (1509006 bytes)
Date: Thu, 07 Mar 2002 18:52:14 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic25.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3562

It's been slow for me too - i wonder if maybe there are alot of people
testing right now, because yesterday it wasn't nearly as slow.


Alex Khomenko wrote:
> 
> I'm downloading download_me.au from www-vr-3, and it is very slow. It will
> definitely take > 10 minutes. Anyone cares to share how long it took them?
> :>
> 
> Thanks,
> Alex
> 
> --
> Knowledge is a deadly friend when no one sets the rules.
> The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: Re: download_me_very_slowly.au (1509006 bytes)
Date: Thu, 07 Mar 2002 18:54:52 -0800
Lines: 16
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic25.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3563

I'm also seeing a whole bunch of ARP responses on the network (hopefully
this isn't because of me - i'm pretty sure it's not). But that might be
another reason for it being so slow.

Alex Khomenko wrote:
> 
> I'm downloading download_me.au from www-vr-3, and it is very slow. It will
> definitely take > 10 minutes. Anyone cares to share how long it took them?
> :>
> 
> Thanks,
> Alex
> 
> --
> Knowledge is a deadly friend when no one sets the rules.
> The fate of all mankind I see is in the hands of fools. (KC "Epitaph")
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: traceroute
Date: 8 Mar 2002 02:58:17 GMT
Lines: 27
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3564

This means you are not sending the ICMP response for a timed out packet
correctly (TTL of 5 at the source will have TTL of 0 at the vr_server).  
You need to create a ICMP time exceeded packet and return it to the host
to get this hop to appear correctly in your program.

Arun


 Huan Wanda Jiang  wrote:
: Hi there,
: I got the result for traceroute.  I am wondering what the * * * for hop 5
: is.  any hint?
: Thanks!

: elaine41:~> traceroute -p 8020 www-vr-1
: traceroute to www-vr-1.Stanford.EDU (172.24.74.130): 1-30 hops, 38 byte
: packets
:  1  leland-gateway.Stanford.EDU (171.64.15.97)  1.22 ms  1.99 ms  0.744 ms
:  2  Core2-gateway.Stanford.EDU (171.64.1.233)  0.749 ms  0.791 ms  0.616
: ms
:  3  Gates-gateway.Stanford.EDU (171.64.3.41)  1.66 ms  2.96 ms  1.73 ms
:  4  vr-firewall.Stanford.EDU (172.24.74.11)  0.838 ms  0.964 ms  0.947 ms
:  5  *  *  *
:  6  www-vr-1.Stanford.EDU (172.24.74.130)  11.1 ms  23.1 ms  10.3 ms



.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: what if ICMP packet itself is timed out in the queue?
Date: Thu, 7 Mar 2002 19:28:00 -0800
Lines: 28
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: myth4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Huan Wanda Jiang 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3565


Hi Huan,

You cannot send an ICMP message about the ICMP message error,
so should be ok to drop if it is pass thro' ICMP, i think.

I wonder what to do if the ARP to default gateway times out
when the VRClient is generating and sending an ICMP packet
to an outside source. Do we have to do retransmits for this ?

Or should we drop this ICMP message silently along with the
original IP packet ? I think dropping both the packets is OK
as link layer was supposed to be reliable.

Please correct me if i am wrong :-)

Thanks, Arvind


On Thu, 7 Mar 2002, Huan Wanda Jiang wrote:

> Hi there,
> I just drop the icmp packet if it is timed out.  Would that be ok?
>
> Thanks!
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: what if ICMP packet itself is timed out in the queue?
Date: 8 Mar 2002 04:07:50 GMT
Lines: 36
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3566

Yes, it is okay to drop the ICMP packet if the ARP request times out for
the default gateway when attempting to send the ICMP packet back to the 
source.

Arun


Arvind Bhat  wrote:

: Hi Huan,

: You cannot send an ICMP message about the ICMP message error,
: so should be ok to drop if it is pass thro' ICMP, i think.

: I wonder what to do if the ARP to default gateway times out
: when the VRClient is generating and sending an ICMP packet
: to an outside source. Do we have to do retransmits for this ?

: Or should we drop this ICMP message silently along with the
: original IP packet ? I think dropping both the packets is OK
: as link layer was supposed to be reliable.

: Please correct me if i am wrong :-)

: Thanks, Arvind


: On Thu, 7 Mar 2002, Huan Wanda Jiang wrote:

:> Hi there,
:> I just drop the icmp packet if it is timed out.  Would that be ok?
:>
:> Thanks!
:>
:>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: Re: recvd close session..
Date: Thu, 07 Mar 2002 20:13:33 -0800
Lines: 30
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: epic25.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3567

Do you know what "certain error conditions" this might occur for?
Currently, I'm trying to download from vr-3, having successfully
downloaded from vr-1 and 2, and i get the recvd close session message.

Thanks!
Tim


Guido Appenzeller wrote:
> 
> Hopil,
> 
> this can happen if the server is restarted, which is once per hour. It
> could also happen during certain error conditions but as far as I can see
> from the log that was not the case.
> 
> How often did you experience this?
> 
>   Guido
> 
> > Sometimes my vr prints "recvd close session." message and then
> > terminates.. It seems that the vr server is sending CLOSE_SESSION
> > command... But then, when does it send it?
> > I guessed that vr server is imposing some timeout on vr clients..
> > but sometimes I could see it very soon after I started my client..
> > So I'm a little afraid that vr server may also send it
> >    when my vr client is doing something wrong..
> > Could I receive some comments on this?
> >
> > Hopil Bae
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Receiving other people's packets..
Date: 8 Mar 2002 06:30:39 GMT
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: saga9.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3568

Hello,

I was wondering if anyone had any similar experiences as I've been having
tonight. IT seems when I test, I can sometimes get a web page or
traceroute, but then something happens where I start getting other
people's packets. Once this happens, my connection no longer works. Has
anyone else been experiencing this tonight? I'm wondering if it has
anything to do with the load tonight (people are quitting their programs,
but still have their client programs running, and then I get allocated
their old port) or something else. But the problem it seems is once it 
get's other people's packets, my packets never seem to get in anymore.

Thanks in advance for any explanation.

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine20.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Re: Receiving other people's packets..
Date: Thu, 7 Mar 2002 23:00:42 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine20.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3569

On 8 Mar 2002, Victor Tung wrote:

> Hello,
>
> I was wondering if anyone had any similar experiences as I've been having
> tonight. IT seems when I test, I can sometimes get a web page or
> traceroute, but then something happens where I start getting other
> people's packets. Once this happens, my connection no longer works. Has
> anyone else been experiencing this tonight? I'm wondering if it has
> anything to do with the load tonight (people are quitting their programs,
> but still have their client programs running, and then I get allocated
> their old port) or something else. But the problem it seems is once it
> get's other people's packets, my packets never seem to get in anymore.
>
> Thanks in advance for any explanation.

It might be possible that sometimes people get assigned the same port
numbers... I was able to get a webpage on my port number before my
program was even half-done. Guido, did you find anything on that?

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Timothy Tay Chao 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: Thu, 07 Mar 2002 23:36:00 -0800
Lines: 74
Distribution: su
Message-ID: 
References:     
NNTP-Posting-Host: epic25.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.75 [en] (X11; U; SunOS 5.8 sun4u)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3570

If we don't check timeouts unless a packet arrives, then it wouldn't
exactly be deterministic right? If no packets are arriving at our router
then we'll never time-out the queue entries and thus never send the ICMP
host unreachable message. Am I right?

I can only get the !H traceroute response by sending my router extra
packets so that it times-out the traceroute packets.

Tim


Guido Appenzeller wrote:
> 
> Henry,
> 
> I guess there is something wrong, it should be deterministic. The load on
> the dispatcher was a little high this afternoon. Maybe try again and also
> decrease the ARP timeout to well below a second.
> 
>   Guido
> 
> > I've managed to get the (!H) traceroute output following the below
> > instructions, but only from time to time, not all the time. Is there
> > something I did wrong?
> >
> > Henry
> >
> > "Guido Appenzeller"  wrote in message
> > 
> >> Hi evertyone,
> >>
> >> forget my earlier comment, host unreachable works and gets through the fw
> >> just fine. Testing it is a little trickier though. Try the following and
> > it
> >> should work:
> >>
> >> - Just using a nonexistant host such as 172.24.74.131 won't work for
> >>   complex reasons you don't have to understand. I have now set up
> >>   the ip address 172.24.74.133. Use it for testing and everything
> >>   should be fine.
> >>   [The reason why it doesn't work is that vr-firewall doesn't have proxy
> >>   arp entries for the other ip addresses. If you really want to know why
> >>   it needs that read the VR docs].
> >> - Telnet won't report anything so it is not much good for testing.
> >> Instead
> >>   use traceroute. For traceroute testing to work make sure the ARP Queue
> >>   time out is set to something very low (maximum 1 second). Then run
> >>   traceroute and you should get something like:
> >>
> >> -- snip --
> >> saga4:~> traceroute -p 8000 172.24.74.133
> >> traceroute to 172.24.74.133 (172.24.74.133): 1-30 hops, 38 byte packets
> >>  1  leland-gateway.Stanford.EDU (171.64.15.129)  1.37 ms  1.24 ms  1.58
> >>  ms
> >>  2  Core6-gateway.Stanford.EDU (171.64.1.193)  0.831 ms  0.902 ms  0.513
> > ms
> >>  3  Gates-gateway.Stanford.EDU (171.64.3.89)  1.40 ms  1.31 ms  1.42 ms
> >>  4  vr-firewall.Stanford.EDU (172.24.74.11)  0.754 ms  1.20 ms  1.34 ms
> >>  5  vr-server-a-1.Stanford.EDU (172.24.74.18)  97.6 ms  8.14 ms  2.70 ms
> >>  6  vr-server-a-1.Stanford.EDU (172.24.74.18)  219 ms (ttl=251!) !H  990
> > ms
> >> (ttl=251!) !H  990 ms (ttl=251!) !H
> >> --- snip ---
> >>
> >>   Note the "!H" denoting host unreachable. For your final submission set
> > ARP
> >>   time out back to 5 seconds.
> >>
> >> - The demo_vr client uses the correct ARP time out, it can't be used for
> >>   testing.
> >>
> >> I hope that helps,
> >>
> >>   Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine20.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Max payload length
Date: Thu, 7 Mar 2002 23:46:49 -0800
Lines: 12
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine20.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3571


Is ETHERNET_MTU the maximum acceptable payload length? Thanks,

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Testing ICMP Host Unreachable
Date: 8 Mar 2002 08:27:18 GMT
Lines: 81
Distribution: su
Message-ID: 
References:      
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3572

This is fine...you can assume that packets arrive often enough that the 
queue is checked at the appropriate interval. We made this change to help 
simplify the code you need to write.

Arun

Timothy Tay Chao  wrote:
: If we don't check timeouts unless a packet arrives, then it wouldn't
: exactly be deterministic right? If no packets are arriving at our router
: then we'll never time-out the queue entries and thus never send the ICMP
: host unreachable message. Am I right?

: I can only get the !H traceroute response by sending my router extra
: packets so that it times-out the traceroute packets.

: Tim


: Guido Appenzeller wrote:
:> 
:> Henry,
:> 
:> I guess there is something wrong, it should be deterministic. The load on
:> the dispatcher was a little high this afternoon. Maybe try again and also
:> decrease the ARP timeout to well below a second.
:> 
:>   Guido
:> 
:> > I've managed to get the (!H) traceroute output following the below
:> > instructions, but only from time to time, not all the time. Is there
:> > something I did wrong?
:> >
:> > Henry
:> >
:> > "Guido Appenzeller"  wrote in message
:> > 
:> >> Hi evertyone,
:> >>
:> >> forget my earlier comment, host unreachable works and gets through the fw
:> >> just fine. Testing it is a little trickier though. Try the following and
:> > it
:> >> should work:
:> >>
:> >> - Just using a nonexistant host such as 172.24.74.131 won't work for
:> >>   complex reasons you don't have to understand. I have now set up
:> >>   the ip address 172.24.74.133. Use it for testing and everything
:> >>   should be fine.
:> >>   [The reason why it doesn't work is that vr-firewall doesn't have proxy
:> >>   arp entries for the other ip addresses. If you really want to know why
:> >>   it needs that read the VR docs].
:> >> - Telnet won't report anything so it is not much good for testing.
:> >> Instead
:> >>   use traceroute. For traceroute testing to work make sure the ARP Queue
:> >>   time out is set to something very low (maximum 1 second). Then run
:> >>   traceroute and you should get something like:
:> >>
:> >> -- snip --
:> >> saga4:~> traceroute -p 8000 172.24.74.133
:> >> traceroute to 172.24.74.133 (172.24.74.133): 1-30 hops, 38 byte packets
:> >>  1  leland-gateway.Stanford.EDU (171.64.15.129)  1.37 ms  1.24 ms  1.58
:> >>  ms
:> >>  2  Core6-gateway.Stanford.EDU (171.64.1.193)  0.831 ms  0.902 ms  0.513
:> > ms
:> >>  3  Gates-gateway.Stanford.EDU (171.64.3.89)  1.40 ms  1.31 ms  1.42 ms
:> >>  4  vr-firewall.Stanford.EDU (172.24.74.11)  0.754 ms  1.20 ms  1.34 ms
:> >>  5  vr-server-a-1.Stanford.EDU (172.24.74.18)  97.6 ms  8.14 ms  2.70 ms
:> >>  6  vr-server-a-1.Stanford.EDU (172.24.74.18)  219 ms (ttl=251!) !H  990
:> > ms
:> >> (ttl=251!) !H  990 ms (ttl=251!) !H
:> >> --- snip ---
:> >>
:> >>   Note the "!H" denoting host unreachable. For your final submission set
:> > ARP
:> >>   time out back to 5 seconds.
:> >>
:> >> - The demo_vr client uses the correct ARP time out, it can't be used for
:> >>   testing.
:> >>
:> >> I hope that helps,
:> >>
:> >>   Guido
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Max payload length
Date: 8 Mar 2002 08:29:16 GMT
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3573

Yes, this is the maximum total length of a packet before it needs to be 
fragmented went sent across an ethernet link. 

Arun

Nickolay Boytchev Stanev  wrote:

: Is ETHERNET_MTU the maximum acceptable payload length? Thanks,

: 							NStanev
: -- 
: ####################################################
: # Nickolay Stanev                                  #
: # E-mail:                   #
: # Cell: (650)-269-4756                             #
: # URL: http://www.stanford.edu/~nbstanev/          #
: ####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: Receiving other people's packets..
Date: 8 Mar 2002 08:40:54 GMT
Lines: 44
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine38.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3574

Hello Nickolay,

I noticed you posted from elaine20. If you ran your web browser from 
elaine20, I think I might have routed some of your packets. :)
I noticed packets coming from elaine20 at around 9:00pm.

-Victor

Nickolay Boytchev Stanev  wrote:
> On 8 Mar 2002, Victor Tung wrote:

>> Hello,
>>
>> I was wondering if anyone had any similar experiences as I've been having
>> tonight. IT seems when I test, I can sometimes get a web page or
>> traceroute, but then something happens where I start getting other
>> people's packets. Once this happens, my connection no longer works. Has
>> anyone else been experiencing this tonight? I'm wondering if it has
>> anything to do with the load tonight (people are quitting their programs,
>> but still have their client programs running, and then I get allocated
>> their old port) or something else. But the problem it seems is once it
>> get's other people's packets, my packets never seem to get in anymore.
>>
>> Thanks in advance for any explanation.

> It might be possible that sometimes people get assigned the same port
> numbers... I was able to get a webpage on my port number before my
> program was even half-done. Guido, did you find anything on that?

> 							NStanev
> -- 
> ####################################################
> # Nickolay Stanev                                  #
> # E-mail:                   #
> # Cell: (650)-269-4756                             #
> # URL: http://www.stanford.edu/~nbstanev/          #
> ####################################################


-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!epic27.Stanford.EDU!ggaurav
From: Gaurav Garg 
Newsgroups: su.class.cs244a
Subject: The server seems to have become more robust!!
Date: Fri, 8 Mar 2002 00:48:35 -0800
Lines: 21
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3575


This is for the TA's just out of curiosity!!

When I tested my code a week back the server used to crash pretty
frequently... but today it is faithfully running for very long times. I
have made no changes to my code. Why could this be so(did you guys got
rid of a bug or something)?

Although its pretty slow today because of heavy traffic!!

Gaurav

-------------------------------------------------------------------------
Gaurav Garg                          | Contact Info: 20A Comstock Circle
1st Yr, Graduate Student             |       Escondido Village, Stanford
Department of Electrical Engineering |       CA-94305
Stanford University                  | Ph: 650-498-1208
-------------------------------------------------------------------------

Visit my webpage at http://www.stanford.edu/~ggaurav

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine37.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Max payload length
Date: Fri, 8 Mar 2002 00:56:30 -0800
Lines: 27
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: elaine37.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3576

Do we need to enforce this (max payload). We shouldn't be designing things
based on a particular environment.

Dhawal Kumar

On 8 Mar 2002, Arun Upadhyaya Kishan wrote:

> Yes, this is the maximum total length of a packet before it needs to be
> fragmented went sent across an ethernet link.
>
> Arun
>
> Nickolay Boytchev Stanev  wrote:
>
> : Is ETHERNET_MTU the maximum acceptable payload length? Thanks,
>
> : 							NStanev
> : --
> : ####################################################
> : # Nickolay Stanev                                  #
> : # E-mail:                   #
> : # Cell: (650)-269-4756                             #
> : # URL: http://www.stanford.edu/~nbstanev/          #
> : ####################################################
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!abhat
From: Arvind Bhat 
Newsgroups: su.class.cs244a
Subject: Re: Max payload length
Date: Fri, 8 Mar 2002 01:22:21 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
  
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Dhawal Kumar 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3577


it is typical to set the mtu size for a particular interface
in this case being ethernet, 1500.

but the assignment does not require us to consider fragmentation
at all, so i guess it is ok to drop paks exceeding that length.

Thanks, Arvind

On Fri, 8 Mar 2002, Dhawal Kumar wrote:

> Do we need to enforce this (max payload). We shouldn't be designing things
> based on a particular environment.
>
> Dhawal Kumar
>
> On 8 Mar 2002, Arun Upadhyaya Kishan wrote:
>
> > Yes, this is the maximum total length of a packet before it needs to be
> > fragmented went sent across an ethernet link.
> >
> > Arun
> >
> > Nickolay Boytchev Stanev  wrote:
> >
> > : Is ETHERNET_MTU the maximum acceptable payload length? Thanks,
> >
> > : 							NStanev
> > : --
> > : ####################################################
> > : # Nickolay Stanev                                  #
> > : # E-mail:                   #
> > : # Cell: (650)-269-4756                             #
> > : # URL: http://www.stanford.edu/~nbstanev/          #
> > : ####################################################
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: Receiving other people's packets..
Date: 8 Mar 2002 11:52:30 GMT
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine38.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3578


Well, after looking at packets.dump with ethereal as opposed to tcpdump, I 
realized they were ICMP packets sent to VR_Client from the gateway. 
(ethereal definitely makes reading the dump files easier) I seem to 
remember a TA stating that this was unavoidable. But I felt that I did see 
an enormous amount of these packets headed to/from these other hosts 
earlier. I guess now the load must have died down. I don't seem to be 
getting as many of these "foreign" packets, but the downloads are still slow.


Victor Tung  wrote:
> Hello,

> I was wondering if anyone had any similar experiences as I've been having
> tonight. IT seems when I test, I can sometimes get a web page or
> traceroute, but then something happens where I start getting other
> people's packets. Once this happens, my connection no longer works. Has
> anyone else been experiencing this tonight? I'm wondering if it has
> anything to do with the load tonight (people are quitting their programs,
> but still have their client programs running, and then I get allocated
> their old port) or something else. But the problem it seems is once it 
> get's other people's packets, my packets never seem to get in anymore.

> Thanks in advance for any explanation.

-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Steven Siuhong Ngai" 
Newsgroups: su.class.cs244a
Subject: Suggestion
Date: Fri, 8 Mar 2002 03:58:44 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: ngai001.stanford.edu
X-Newsreader: Microsoft Outlook Express 4.72.3110.5
X-MimeOLE: Produced By Microsoft MimeOLE V4.72.3110.3
Xref: nntp.stanford.edu su.class.cs244a:3579

It was mentioned a while ago on this group that the server is periodically
restarted, and consequently at that time all connected clients are dropped.

Can the TAs go without this restart for a while so that people can get an
uninterrupted download of the last file? Average times for me are running in
40min-1hr range.

Thanks,
Steve


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine28.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: ICMP host unreachable...
Date: Fri, 8 Mar 2002 05:35:44 -0800
Lines: 39
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine28.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3580


TA's,

I'm failing to obtain reasonable output from traceroute when doing
traceroute -p <port> 172.24.74.133
Below is what I found in the packets.dump file regarding the ICMP packets
I'm sending. Is that information enough to assume that my messages are
properly formatted, or is there some subtle (or not so subtle) problem
that I could still be having? This has been a very annoying problem and I
want to make sure that I am, in fact, tracking down a bug and not just
wasting sleep... Thanks!

05:21:16.169031 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
time exc
eeded in-transit for elaine43.Stanford.EDU.36744 > 172.24.74.133.8114:
[no cksu
m] udp 10 (DF) [ttl 1] (id 58259, len 38) (ttl 63, id 0, len 56)

05:21:16.255500 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
time exc
eeded in-transit for elaine43.Stanford.EDU.36744 > 172.24.74.133.8115:
[no cksu
m] udp 10 (DF) [ttl 1] (id 58260, len 38) (ttl 63, id 0, len 56)

05:21:55.951094 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
host 172
..24.74.133 unreachable for elaine43.Stanford.EDU.36744 >
172.24.74.133.8119:  [n
o cksum] udp 10 (DF) (ttl 2, id 58264, len 38) (ttl 63, id 0, len 56)

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: recvd close session..
Date: 8 Mar 2002 13:57:52 GMT
Lines: 45
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: elaine38.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3581

Guido,

I'm getting this as well. It seems to only happen on the  really long file 
which throughout Thurs. night has been taking close to an hour to 
download. The other files have downloaded OK. I'm guessing if the server 
gets reset every hour, the files are taking close to that long to 
download.  Though I'm downloading after 2 "closes" and the file seems to 
be downloading much faster now.

Oh, another thing is that once it "closes" our VR client, I decided to let 
the ftp continue, and the whole file downloaded completely and as far as I 
can tell correctly. But I'm guessing maybe it was running someone else's 
VR client, so I'd like to test my implementation thoroughly.

-Victor


Guido Appenzeller  wrote:
> Hopil,

> this can happen if the server is restarted, which is once per hour. It 
> could also happen during certain error conditions but as far as I can see 
> from the log that was not the case.

> How often did you experience this?

>   Guido

>> Sometimes my vr prints "recvd close session." message and then
>> terminates.. It seems that the vr server is sending CLOSE_SESSION
>> command... But then, when does it send it?
>> I guessed that vr server is imposing some timeout on vr clients..
>> but sometimes I could see it very soon after I started my client..
>> So I'm a little afraid that vr server may also send it
>>    when my vr client is doing something wrong..
>> Could I receive some comments on this?
>> 
>> Hopil Bae


-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine33.Stanford.EDU!anuragg
From: Anurag Gupta 
Newsgroups: su.class.cs244a
Subject: submission errors
Date: Fri, 8 Mar 2002 07:23:12 -0800
Lines: 19
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine33.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3582

hi,

i am having problems submitting hw3.

it says

============
/usr/class/cs244a/bin/submit assignment: hw3; TA: hondroul
mkdir /afs/ir/class/cs244a/submissions/grading/hw3/hondroul
mkdir /afs/ir/class/cs244a/submissions/grading/hw3/hondroul: Permission
denied at /usr/class/cs244a/bin/submit line 133
============

any clues?

thanks
-anurag


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: ICMP host unreachable...
Date: 8 Mar 2002 19:31:28 GMT
Lines: 30
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3583

>I'm failing to obtain reasonable output from traceroute when doing
>traceroute -p <port> 172.24.74.133
>Below is what I found in the packets.dump file regarding the ICMP packets
>I'm sending. Is that information enough to assume that my messages are
>properly formatted, or is there some subtle (or not so subtle) problem
>that I could still be having? This has been a very annoying problem and I
>want to make sure that I am, in fact, tracking down a bug and not just
>wasting sleep... Thanks!

>05:21:16.169031 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
>time exc
>eeded in-transit for elaine43.Stanford.EDU.36744 > 172.24.74.133.8114:
>[no cksu
>m] udp 10 (DF) [ttl 1] (id 58259, len 38) (ttl 63, id 0, len 56)

>05:21:16.255500 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
>time exc
>eeded in-transit for elaine43.Stanford.EDU.36744 > 172.24.74.133.8115:
>[no cksu
>m] udp 10 (DF) [ttl 1] (id 58260, len 38) (ttl 63, id 0, len 56)


traceroute to 133 doesn't show anything reasonable for me either; I'm not
sure if the configuration has changed recently or not.  Your messages 
look reasonable--although as someone else mentioned, the output is easier
to read (and thus easier to find mistakes) in ethereal than in tcpdump.

If you force your implementation to return "host unreachable" even for
valid hosts, does traceroute behave as expected?  (This case certainly
seems to be working on a consistent basis).
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: submission errors
Date: 8 Mar 2002 19:32:26 GMT
Lines: 15
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine15.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3584


>i am having problems submitting hw3.

Are you trying to submit for Antonios?  (And if so, why?)


>it says

>============
>/usr/class/cs244a/bin/submit assignment: hw3; TA: hondroul
>mkdir /afs/ir/class/cs244a/submissions/grading/hw3/hondroul
>mkdir /afs/ir/class/cs244a/submissions/grading/hw3/hondroul: Permission
>denied at /usr/class/cs244a/bin/submit line 133
>============

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: ICMP host unreachable...
Date: Fri, 8 Mar 2002 11:43:34 -0800
Lines: 55
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3585

As Matthew said, the output looks ok to me. Try running it with
"-vv" (this is actually a FAQ) to see whether the checksums 
are ok. Also make sure that the timing issue (see previous 
posting) is not the problem when you test.

  Guido


---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Fri, 8 Mar 2002, Nickolay Boytchev Stanev wrote:

> 
> TA's,
> 
> I'm failing to obtain reasonable output from traceroute when doing
> traceroute -p <port> 172.24.74.133
> Below is what I found in the packets.dump file regarding the ICMP packets
> I'm sending. Is that information enough to assume that my messages are
> properly formatted, or is there some subtle (or not so subtle) problem
> that I could still be having? This has been a very annoying problem and I
> want to make sure that I am, in fact, tracking down a bug and not just
> wasting sleep... Thanks!
> 
> 05:21:16.169031 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
> time exc
> eeded in-transit for elaine43.Stanford.EDU.36744 > 172.24.74.133.8114:
> [no cksu
> m] udp 10 (DF) [ttl 1] (id 58259, len 38) (ttl 63, id 0, len 56)
> 
> 05:21:16.255500 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
> time exc
> eeded in-transit for elaine43.Stanford.EDU.36744 > 172.24.74.133.8115:
> [no cksu
> m] udp 10 (DF) [ttl 1] (id 58260, len 38) (ttl 63, id 0, len 56)
> 
> 05:21:55.951094 vr-server-1.Stanford.EDU > elaine43.Stanford.EDU: icmp:
> host 172
> .24.74.133 unreachable for elaine43.Stanford.EDU.36744 >
> 172.24.74.133.8119:  [n
> o cksum] udp 10 (DF) (ttl 2, id 58264, len 38) (ttl 63, id 0, len 56)
> 
> 							NStanev
> -- 
> ####################################################
> # Nickolay Stanev                                  #
> # E-mail:                   #
> # Cell: (650)-269-4756                             #
> # URL: http://www.stanford.edu/~nbstanev/          #
> ####################################################
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: Suggestion
Date: Fri, 8 Mar 2002 11:48:35 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3586

40 minutes sounds very long to me, are you sure that your
code is not slowing things down? Try testing speed with the
demo_vr and see how speeds compare. Even with the current
very high load, I'd expect a ownload time of minutes.

And BTW, if you manage to get a partial download it is 
probably enough identify the file contents.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Fri, 8 Mar 2002, Steven Siuhong Ngai wrote:

> It was mentioned a while ago on this group that the server is periodically
> restarted, and consequently at that time all connected clients are dropped.
> 
> Can the TAs go without this restart for a while so that people can get an
> uninterrupted download of the last file? Average times for me are running in
> 40min-1hr range.
> 
> Thanks,
> Steve
> 
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!appenz
From: Guido Appenzeller 
Newsgroups: su.class.cs244a
Subject: Re: The server seems to have become more robust!!
Date: Fri, 8 Mar 2002 11:52:12 -0800
Lines: 35
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3587

Yes, stability has increased a lot in the last week. I fixed
two major bugs on Monday and the server has been running with
hardly any crashes since.

  Guido

---------------------------------------------------------------
Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
 - office: 650 7253545  cell: 650 7042781

On Fri, 8 Mar 2002, Gaurav Garg wrote:

> 
> This is for the TA's just out of curiosity!!
> 
> When I tested my code a week back the server used to crash pretty
> frequently... but today it is faithfully running for very long times. I
> have made no changes to my code. Why could this be so(did you guys got
> rid of a bug or something)?
> 
> Although its pretty slow today because of heavy traffic!!
> 
> Gaurav
> 
> -------------------------------------------------------------------------
> Gaurav Garg                          | Contact Info: 20A Comstock Circle
> 1st Yr, Graduate Student             |       Escondido Village, Stanford
> Department of Electrical Engineering |       CA-94305
> Stanford University                  | Ph: 650-498-1208
> -------------------------------------------------------------------------
> 
> Visit my webpage at http://www.stanford.edu/~ggaurav
> 
> 

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic27.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Suggestion
Date: Fri, 8 Mar 2002 12:35:03 -0800
Lines: 40
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic27.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3588


A couple of days ago I was able to download files in 5 seconds, but now
the time has definitely increased several times. Latest download: I had to
wait 10 mins to download file from vr2.

Dhawal Kumar

On Fri, 8 Mar 2002, Guido Appenzeller wrote:

> 40 minutes sounds very long to me, are you sure that your
> code is not slowing things down? Try testing speed with the
> demo_vr and see how speeds compare. Even with the current
> very high load, I'd expect a ownload time of minutes.
>
> And BTW, if you manage to get a partial download it is
> probably enough identify the file contents.
>
>   Guido
>
> ---------------------------------------------------------------
> Guido Appenzeller, Ph.D. Candiate, Computer Sc., Stanford Univ.
>  - office: 650 7253545  cell: 650 7042781
>
> On Fri, 8 Mar 2002, Steven Siuhong Ngai wrote:
>
> > It was mentioned a while ago on this group that the server is periodically
> > restarted, and consequently at that time all connected clients are dropped.
> >
> > Can the TAs go without this restart for a while so that people can get an
> > uninterrupted download of the last file? Average times for me are running in
> > 40min-1hr range.
> >
> > Thanks,
> > Steve
> >
> >
> >
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!cardinal1.Stanford.EDU!fsun
From: Fang Sun 
Newsgroups: su.class.cs244a
Subject: about the final exam
Date: Mon, 11 Mar 2002 14:31:49 -0800
Lines: 4
Distribution: su
Message-ID: 
NNTP-Posting-Host: cardinal1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3589

Will the final exam cover all the materials or only the second half after
the midterm?
Thanks

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Eric Watkins 
Newsgroups: su.class.cs244a
Subject: Alternate final?
Date: Tue, 12 Mar 2002 01:34:25 -0800
Lines: 3
Distribution: su
Message-ID: 
NNTP-Posting-Host: e-man.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Mailer: Mozilla 4.77 [en] (Windows NT 5.0; U)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3590

Will there be one?  Where / when?  I'm taking a class concurrently.

-Eric
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Problem set #4: 8
Date: Tue, 12 Mar 2002 20:17:19 -0800
Lines: 13
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3591

Hi,

  I'm a bit confused by question number 8.  The diagram shows
  a circuit with a generator of 3 bits, but part (a) says the
  generator is 4 bits.. G(x) = 1011.  Are we supposed to
  fill in the extra register and column in the graph?
  Please let me know if I'm off base here... perhaps there is
  something fundamental I simply don't understand.  Thanks very
  much in advance :-)


                    ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic5.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding the question 6.
Date: Tue, 12 Mar 2002 21:19:09 -0800
Lines: 11
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic5.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3592

Hi,
Is the condition for the question 6 4/5 b problem correct because i am not
able to form more than 12 codes using the constrained. When we are saying
no 2 consecutive codes can have 4 0's and 1's does that mean any 2 codes or
consecutive in number that is say only 0000 and 0001 of the four bit code
represented by 5 bits cannot have 4 0's and 1's.
Thanks
Regards
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: Problem set #4: 8
Date: Tue, 12 Mar 2002 21:29:07 -0800
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3593


Nevermind on this... I'm dense. (thanks shankar)

                    ~~m

> Hi,
>
>   I'm a bit confused by question number 8.  The diagram shows
>   a circuit with a generator of 3 bits, but part (a) says the
>   generator is 4 bits.. G(x) = 1011.  Are we supposed to
>   fill in the extra register and column in the graph?
>   Please let me know if I'm off base here... perhaps there is
>   something fundamental I simply don't understand.  Thanks very
>   much in advance :-)
>
>
>                     ~~m
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!Xenon.Stanford.EDU!xwang
From: Xin Wang 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the question 6.
Date: Tue, 12 Mar 2002 21:56:56 -0800
Lines: 29
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: xenon.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Shankar Agarwal 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3594

Hi Shankar,

> Is the condition for the question 6 4/5 b problem correct because i am not
> able to form more than 12 codes using the constrained. 

The condition is correct.

> When we are saying no 2 consecutive codes can have 4 0's and 1's does
> that mean any 2 codes or consecutive in number that is say only 0000
> and 0001 of the four bit code represented by 5 bits cannot have 4 0's
> and 1's. Thanks Regards Shankar


Any pair of 5-bit codes should not have MORE THAN 4 consecutive 1s or
0s. The problem says "no more than 4". Did you consider it when you solve
the problem? 

Best,

Xin

-- 
---------------------------------
Xin Wang

Department of Computer Science
Stanford University


.

Path: shelby.stanford.edu!nntp.stanford.edu!epic1.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Regarding the question 6.
Date: Wed, 13 Mar 2002 08:38:44 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Xin Wang 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3595

Hi Xin,
Thanks a lot. By mistake i hade read it as 4 or more than than more than
4.
Thanks
Shankar

On Tue, 12 Mar 2002, Xin Wang wrote:

> Hi Shankar,
>
> > Is the condition for the question 6 4/5 b problem correct because i am not
> > able to form more than 12 codes using the constrained.
>
> The condition is correct.
>
> > When we are saying no 2 consecutive codes can have 4 0's and 1's does
> > that mean any 2 codes or consecutive in number that is say only 0000
> > and 0001 of the four bit code represented by 5 bits cannot have 4 0's
> > and 1's. Thanks Regards Shankar
>
>
> Any pair of 5-bit codes should not have MORE THAN 4 consecutive 1s or
> 0s. The problem says "no more than 4". Did you consider it when you solve
> the problem?
>
> Best,
>
> Xin
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic1.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding question 5
Date: Wed, 13 Mar 2002 08:47:37 -0800
Lines: 9
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic1.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3596

Hi.
I am getting the interpacket gap of 40 microsecond but the question says
that the interpacket gap is 1 microsecond. Using the equation given in the
class interpacketgap should be >= B/2 * Rmax i get 40 microsecond where i
am taking R to be 100 Mb/s. Can you please let me know if this is correct
and i should be getting 1 micorsecond and not 40 microsecond.
Thanks
Shankar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic6.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Re: Regarding question 5
Date: Thu, 14 Mar 2002 09:24:34 -0800
Lines: 18
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: epic6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3597

Hi,
I again made a mistake here. So please ignore this mail.
Thanks
Shankar

On Wed, 13 Mar 2002, Shankar Agarwal wrote:

> Hi.
> I am getting the interpacket gap of 40 microsecond but the question says
> that the interpacket gap is 1 microsecond. Using the equation given in the
> class interpacketgap should be >= B/2 * Rmax i get 40 microsecond where i
> am taking R to be 100 Mb/s. Can you please let me know if this is correct
> and i should be getting 1 micorsecond and not 40 microsecond.
> Thanks
> Shankar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine18.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: SITN fax number (SITN students only)
Date: Thu, 14 Mar 2002 19:16:36 -0800
Lines: 6
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine18.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3598

Anyone knows SITN fax no? I have been able to figure out only the SCPD fax
number. Thanks, in advance.


Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!myth4.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Phase-lock loop
Date: Thu, 14 Mar 2002 23:16:47 -0800
Lines: 14
Distribution: su
Message-ID: 
NNTP-Posting-Host: myth4.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3599


With the risk of appearing amusingly ignorant, I would like to ask the
following: what is a phase lock loop and is that relevant to problem 7.b.
on PS4 at all... Thanks,

							NStanev
-- 
####################################################
# Nickolay Stanev                                  #
# E-mail:                   #
# Cell: (650)-269-4756                             #
# URL: http://www.stanford.edu/~nbstanev/          #
####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Phase-lock loop
Date: 15 Mar 2002 08:18:28 GMT
Lines: 22
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: saga4.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3600

Knowledge of phase lock loops is not needed to answer this question. Look 
at the bit patterns from 00000 - 11111 and see what kind of patterns will 
allow you to make the guarantees you need to make (you can eliminate 
several of the patterns first).

Arun

Nickolay Boytchev Stanev  wrote:

: With the risk of appearing amusingly ignorant, I would like to ask the
: following: what is a phase lock loop and is that relevant to problem 7.b.
: on PS4 at all... Thanks,

: 							NStanev
: -- 
: ####################################################
: # Nickolay Stanev                                  #
: # E-mail:                   #
: # Cell: (650)-269-4756                             #
: # URL: http://www.stanford.edu/~nbstanev/          #
: ####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic22.Stanford.EDU!nbstanev
From: Nickolay Boytchev Stanev 
Newsgroups: su.class.cs244a
Subject: Re: Phase-lock loop
Date: Fri, 15 Mar 2002 05:16:12 -0800
Lines: 30
Distribution: su
Message-ID: 
References: 
 
NNTP-Posting-Host: epic22.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3601


Um, I have that. I was talking about part b: why the restrictions are
necessary. Whatever PLL is, it could affect the answer. Are we just
supposed to explain why long unchanging sequences of bits are undesirable
in general? Thanks!

On 15 Mar 2002, Arun Upadhyaya Kishan wrote:

> Knowledge of phase lock loops is not needed to answer this question. Look
> at the bit patterns from 00000 - 11111 and see what kind of patterns will
> allow you to make the guarantees you need to make (you can eliminate
> several of the patterns first).
>
> Arun
>
> Nickolay Boytchev Stanev  wrote:
>
> : With the risk of appearing amusingly ignorant, I would like to ask the
> : following: what is a phase lock loop and is that relevant to problem 7.b.
> : on PS4 at all... Thanks,
>
> : 							NStanev
> : --
> : ####################################################
> : # Nickolay Stanev                                  #
> : # E-mail:                   #
> : # Cell: (650)-269-4756                             #
> : # URL: http://www.stanford.edu/~nbstanev/          #
> : ####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine34.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Question regarding error detection theorem (ii) - handout 14
Date: Fri, 15 Mar 2002 19:54:36 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine34.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3602

I have a doubt regarding the theorem (ii) on slide 14 of handout 14 (shown
as 15 on the bottom of slide). The theorem says two isolated errors can be
detected if (x^k + 1) is not a factor of G(x) for any k. Now consider the
following:
E(x) = x^3 + 1 (1001) - Note isolated errors.
G(x) = x^2 + x + 1 (111)

Clearly (x^k + 1) is not a factor of G(x) for k = 1, 2, 3 ... but G(x)
divides E(x) i.e. error will go undetected.

If you don't want to consider base 2 arithmetic while division i.e. you
want to consider the normal polynomial arithmetic, use
E(x) = x^3 + 1
G(x) = x^2 - x + 1

Dhawal Kumar

.

Path: shelby.stanford.edu!nntp.stanford.edu!epic11.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Regarding the slide 33 handout 3
Date: Fri, 15 Mar 2002 20:08:50 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: epic11.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3603

Hi,
I am not able to figure out the average queue occupancy in the sample
given on this slide. Firt the Q(t) i given as 0.5 * (0.1 * 500) = 25 bits
then again over 1 second it is given as 5 bits. I am not getting anything
about it. Can someone tell me how did we get the first average of 25 bits.
What does 0.5 , 0.1 and 500 represent here
Thanks
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine15.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: Handout 10 slide 14
Date: Sat, 16 Mar 2002 15:40:39 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine15.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3604

Hi,
Can someone explain me how did we get the equation
A = (a * 0) + (1 - a)(1 + A)
I am slightly at a loss about this equation.
Thanks
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Phase-lock loop
Date: 17 Mar 2002 01:54:55 GMT
Lines: 41
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: elaine21.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3605

Knowledge of what a PLL is could help but is not required to solve that
problem -- a PLL circuit is used for frequency control and is often used 
for clock recovery. This use provides the basis for this question, which 
was also something Prof. McKeown alluded to in class. With long 
consecutive streams of 1s and 0s, the receiver will begin to face 
difficulty in ascertaining transitions from one signal level to another 
due to baseline wandering over time.

Arun

Nickolay Boytchev Stanev  wrote:

: Um, I have that. I was talking about part b: why the restrictions are
: necessary. Whatever PLL is, it could affect the answer. Are we just
: supposed to explain why long unchanging sequences of bits are undesirable
: in general? Thanks!

: On 15 Mar 2002, Arun Upadhyaya Kishan wrote:

:> Knowledge of phase lock loops is not needed to answer this question. Look
:> at the bit patterns from 00000 - 11111 and see what kind of patterns will
:> allow you to make the guarantees you need to make (you can eliminate
:> several of the patterns first).
:>
:> Arun
:>
:> Nickolay Boytchev Stanev  wrote:
:>
:> : With the risk of appearing amusingly ignorant, I would like to ask the
:> : following: what is a phase lock loop and is that relevant to problem 7.b.
:> : on PS4 at all... Thanks,
:>
:> : 							NStanev
:> : --
:> : ####################################################
:> : # Nickolay Stanev                                  #
:> : # E-mail:                   #
:> : # Cell: (650)-269-4756                             #
:> : # URL: http://www.stanford.edu/~nbstanev/          #
:> : ####################################################

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Regarding the slide 33 handout 3
Date: 17 Mar 2002 02:17:00 GMT
Lines: 20
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3606


>I am not able to figure out the average queue occupancy in the sample
>given on this slide. Firt the Q(t) i given as 0.5 * (0.1 * 500) = 25 bits
>then again over 1 second it is given as 5 bits. I am not getting anything
>about it. Can someone tell me how did we get the first average of 25 bits.
>What does 0.5 , 0.1 and 500 represent here

Not quite--the slide says that E(Q(t) | Q(t) > 0) = 25 bits, while
E(Q(t)) = 5 bits.

Remember that Q(t) = A(t) - S(t), and that the average queue occupancy over
some interval [0,T] is just 1/T integral(0,T) Q(t) dt.  The expression on the
slide is taking a little shortcut to evaluate that.  (For example, you could
just compute E(Q(t)) = (0.5*b*h) / 1 s, where b/h are the base/height of the
triangle shown.  This gives you 5 ms, as the slide says).

There were a few posts on this very slide a long time ago--if you look
for the posts entitled "Fluid model example" or something similar from
around that time (first page or two in the list of posts), there's quite
a bit of discussion there.
.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine9.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: Handout 10 slide 14
Date: 17 Mar 2002 02:19:41 GMT
Lines: 12
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3607


>Can someone explain me how did we get the equation
>A = (a * 0) + (1 - a)(1 + A)

A is the number of time slots we expect to wait before a packet is
transmitted successfully.  With probability a, exactly one node
transmits in a slot, so the expected number of slots is zero.  With
probability (1-a), we experience a collision, which means we must
wait for one time slot before repeating the experiment; when we repeat
the experiment we again expect to wait for A time slots before successfully
transmitting the packet.

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Questions about the solution of Problem 4 (Q1 & 2)
Date: Sat, 16 Mar 2002 20:34:23 -0800
Lines: 84
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3608

Hi,

I have the following questions about the solutions of problem 1 and 2:

I will start with Problem2 (Since it is easier):

Assuming we have PROPmin between any two hosts, and assuming that at time
t0 host X just finished placing the last bit of its packet. I argue that
host X will need "at least" 2*PROPmin to hear anything else on the line if
there was no collision. To explain why, let us follow the timing diagram
for a non-collided packet:

Assume we have host X currently transmitting its packet, and its neighbor,
host Y has a packet to send directly after host X will finish:

1. Host X places the last bit of its packet at T0.
2. Host Y will see the last bit of hostX at T0+PROPmin.
3. Host Y will start transmitting the first bit of its packet at
the same time T0+PROPmin.
4. Host X will receive the first bit of Host Y packet at T0+2*PROPmin.

Therefore, if Host X packet went through without collision, host X will
see a silent period of 2*PROPmin on the etherent cable directly after it
finishes its transmission.

That means, Host X can actually ends up its packet "2*PROPmin" before the
conventional time of 2*PROP. During the last 2*PROPmin, host X senses the
line, and if it finds anything on the line, it will know that a collision
took place and its packet didn't make it through.

If my logic is correct, then the solution should be 2*PROP-2*PROPmin, and
not 2*PROP-PROPmin. I lost most of the question points for this (2/6).


For Problem1:
I want to ask about part C. I think that the max queuing delay along N
routers is the queuing delay along *one router*. The question solution
says : "The actual queuing delay will be lower than this", and personally
I believe it is *much* lower than this. To describe why, I do this
emperically using a case of 2 routers, a source, and a destination:

Assume we have just 2 routers, each has a maximum output rate R1, R2. Also,
assume that the source has a maximum output rate R. Also I will assume
that the source started its transmission with R for the "segma" burst, and
after that, source continues with the "row" rate. Then, argue that for any
state of this system, you will not get a queuing delay larger than:
(segma)( (1/row) - (1/R)) where R is the highest R among the egress
routers rates (including sender rate). To check this, I will try to
enumerate the four basic cases:

(a) router 1 and 2 can only operate at "row" (Since they are loaded from
other sources, and can only give the allocated rate).

Then we will have a delay (segma) ((1/row) - (1/R)) for the first router.
We will have *zero* delay for the second router (since the data is coming
in "row" rate, which it promised to handle immediately). Totally, we have
a max delay of one router.

(b) If we have both of them operating at R1, and R2 (assume they have no
other load to handle). Then we will have much lower delay, and this is not
intresting for calculating the max delay.

(c) For the remaining two cases (either router 1 can operate at R1, and
router2 can operate at "row", or vice versa). If you calculate these
two last cases (which is easy to do, and I will not go throug it here),
you will reach that the queuing delay is exactely the same as max delay of
one router!

If my logic above is correct, that means for the case of 2 routers, source
and destination, the max queue delay is the same as the max
queuing delay for one router. In my solution I generalized this to N
routers and tried to solve it mathematically to show that. However, I got
0 out of 5. I don't know if my grade goes with the policy of the HW, which
states "Show your reasoning clearly. If your reasoning is correct, you may
recieve a partial credit".

I will deeply appreciate any clarifications, or If I am missing something
above.

Thanks,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine7.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Problem set 4 - Question 8
Date: Sat, 16 Mar 2002 22:26:13 -0800
Lines: 15
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3610

Hi,

I wanna ask about the solution. After all the bits of Message enetered to
the CRC circuit (which happened at time = 7), We got the remainder 100
(which I think is the answer). However, the solution goes further and does
another 3 extra time steps (assuming that the input data is 0). I am not
understanding why it didn't stop at time = 7?

Thanks for any help :)

Good luck for you all,

Moh'd


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Victor Tung 
Newsgroups: su.class.cs244a
Subject: Re: Problem set 4 - Question 8
Date: 17 Mar 2002 07:34:37 GMT
Lines: 38
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine11.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3611

Hello,

I was wondering the exact same thing until I looked at the class notes. If
you look at the long division example in the notes, when the G(x) is 4
bits, he padded the message, M, with 3 extra bits when performing the
division. I think this is basically the reason why they didn't stop at
time t=7 as the circuit is really performing the division for you. As for
why the 3 digits are added to the end, I thought about it for a while, and
my only guess is that since you are trying to make M divisible by G(x),
you are really trying to make M + (3 bits of the remainder) you are trying
to make divisible by G(x), therefore you are really performing division on 
M + (3 bits of 000 at the end of M) by G(x), and not M divided by G(x). 
Again, this is just a guess. Also, when I said 3 bits, this is assuming 
G(x) is 4 bits, if G(x) has let's say 6 bits, then you need to add on 5 
more bits of 0's at the end.

Moh'd Saleem Saleem Alrawashdeh  wrote:
> Hi,

> I wanna ask about the solution. After all the bits of Message enetered to
> the CRC circuit (which happened at time = 7), We got the remainder 100
> (which I think is the answer). However, the solution goes further and does
> another 3 extra time steps (assuming that the input data is 0). I am not
> understanding why it didn't stop at time = 7?

> Thanks for any help :)

> Good luck for you all,

> Moh'd



-- 
Victor Tung			| 
------------------------------------------------------------------------------
Metaphysics is the science of proving what we don't understand.
                                -- Josh Billings (Henry Wheeler Shaw)
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Phil" 
Newsgroups: su.class.cs244a
Subject: HO#3 slide 28.. and TCP sender Tx window
Date: Sat, 16 Mar 2002 23:51:04 -0800
Lines: 101
Distribution: su
Message-ID: 
NNTP-Posting-Host: hopils.stanford.edu
Mime-Version: 1.0
Content-Type: multipart/alternative;
	boundary="----=_NextPart_000_003B_01C1CD45.6F4BCA40"
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3612

This is a multi-part message in MIME format.

------=_NextPart_000_003B_01C1CD45.6F4BCA40
Content-Type: text/plain;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable


1)
I'm wondering about the latency equation in slide 28 of HO#3,
which says that the latency is M/R_MIN + sum(PROP_i)..
M/R_MIN doesn't increase as the number of hops increases,
so it looks counterintuitive to me...
I think latency should be something like sum(P/R)+sum(PROP_i)
where P is the packet size...=20

2)=20
And as in the midterm TCP problem,
Tx sends packet 4,5,6,7 when Tx win_size=3D4,
and 5,6,7 are received but 4 is lost...
Then Tx win size becomes 2, so only 4 and 5 are retransmitted..
4 is received OK this time,
then since Rx has received 4,5,6,7 correctly, it will send ack for 7.

But now Tx window includes only 4 and 5..
7 is outside the window..Then, should Tx ignore the ack?
although ignoring would result in a deadlock..
And if Tx accepts ack for 7, then maybe it should send 8,9,10...
Or should it send 6,7,8?

Thank you.
Hopil Bae


------=_NextPart_000_003B_01C1CD45.6F4BCA40
Content-Type: text/html;
	charset="ks_c_5601-1987"
Content-Transfer-Encoding: quoted-printable

<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN">
<HTML><HEAD>
<META content=3D"text/html; charset=3Dks_c_5601-1987" =
http-equiv=3DContent-Type>
<META content=3D"MSHTML 5.00.3315.2870" name=3DGENERATOR>
<STYLE></STYLE>
</HEAD>
<BODY bgColor=3D#ffffff>
<DIV><FONT face=3DArial size=3D2></FONT> </DIV>
<DIV><FONT face=3DArial size=3D2>1)</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I'm wondering about the latency =
equation in slide=20
28 of HO#3,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>which says that the latency is M/R_MIN =
+=20
sum(PROP_i)..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>M/R_MIN doesn't increase as the <FONT =
face=3DArial=20
size=3D2>number of hops increases,</FONT></FONT></DIV>
<DIV><FONT face=3DArial size=3D2>so it looks counterintuitive to=20
me...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>I think latency should be something =
like=20
sum(P/R)+sum(PROP_i)</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>where P is the packet size... =
</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>2) </FONT></DIV>
<DIV><FONT face=3DArial size=3D2>And as in the midterm TCP =
problem,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Tx sends packet 4,5,6,7 when Tx=20
win_size=3D4,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>and 5,6,7 are received but 4 is=20
lost...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Then Tx win size becomes 2, so only 4 =
and 5 are=20
retransmitted..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>4 is received OK this =
time,</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>then since Rx has received =
4,5,6,7 correctly,=20
it will send ack for 7.</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>But now Tx window includes only 4 =
and=20
5..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>7 is outside the window..</FONT><FONT =
face=3DArial=20
size=3D2>Then, should Tx ignore the ack?</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>although ignoring would result in a=20
deadlock..</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>And if Tx accepts ack for 7, then maybe =
it should=20
send 8,9,10...</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Or should it send 6,7,8?</FONT></DIV>
<DIV> </DIV>
<DIV><FONT face=3DArial size=3D2>Thank you.</FONT></DIV>
<DIV><FONT face=3DArial size=3D2>Hopil Bae</FONT></DIV>
<DIV> </DIV></BODY></HTML>

------=_NextPart_000_003B_01C1CD45.6F4BCA40--

.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine6.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Re: Question regarding error detection theorem (ii) - handout 14
Date: Sun, 17 Mar 2002 11:20:50 -0800
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine6.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3613


I hope this question didn't escape TA's eyes.

Dhawal Kumar

On Fri, 15 Mar 2002, Dhawal Kumar wrote:

> I have a doubt regarding the theorem (ii) on slide 14 of handout 14 (shown
> as 15 on the bottom of slide). The theorem says two isolated errors can be
> detected if (x^k + 1) is not a factor of G(x) for any k. Now consider the
> following:
> E(x) = x^3 + 1 (1001) - Note isolated errors.
> G(x) = x^2 + x + 1 (111)
>
> Clearly (x^k + 1) is not a factor of G(x) for k = 1, 2, 3 ... but G(x)
> divides E(x) i.e. error will go undetected.
>
> If you don't want to consider base 2 arithmetic while division i.e. you
> want to consider the normal polynomial arithmetic, use
> E(x) = x^3 + 1
> G(x) = x^2 - x + 1
>
> Dhawal Kumar
>
>

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Leonard Sibille" 
Newsgroups: su.class.cs244a
Subject: Previous years Final Exams
Date: Sun, 17 Mar 2002 12:46:08 -0800
Lines: 10
Distribution: su
Message-ID: 
NNTP-Posting-Host: leo22.stanford.edu
X-Trace: news.Stanford.EDU 1016397766 28497 128.12.188.148 (17 Mar 2002 20:42:46 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4133.2400
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
Xref: nntp.stanford.edu su.class.cs244a:3614

Hi,

Would it be possible to have previous years final exams and solutions
accessible on the class website, so that we can work on them?  Thank you
very much.


Leo


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine41.Stanford.EDU!shankara
From: Shankar Agarwal 
Newsgroups: su.class.cs244a
Subject: PS 1 problem 7
Date: Sun, 17 Mar 2002 14:40:19 -0800
Lines: 8
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine41.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3615

Hi,
Can you please elaborate the d(t) for this problem
I am not able to understand how did d(t) = 9(t-i) + 0.1 i <= t <=i+0.1
came.
Thanks
Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: francois-marie lefevere 
Newsgroups: su.class.cs244a
Subject: Re: Questions about the solution of Problem 4 (Q1 & 2)
Date: Sun, 17 Mar 2002 16:16:30 -0800
Lines: 37
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: nastia.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1016410591 125 128.12.187.116 (18 Mar 2002 00:16:31 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.4.9-31 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3616

I completely agree with your two answers and also got the same score as you
for these questions :

My justification for  1 c) was
-If we consider the whole set of routers as a black box, in input we have the
same thing as in question 1 a).
At the output we have a rate greater or equal rho because ALL the routers
guarantee at least a rate of c >= rho.
Therefore the comportment of the whole system is like what we had in 1 a) and
in the worst case we obtain the same answer as in 1 a) for the global delay
(and we don't have to care about  what is going on inside the black box (we
didn't care when we did question 1a !!!! so why would we care now ??)).
I can't see any flaw in that very simple proof. If somebody else see one, he
can email me a detailed counter example, I would be interested.

- For question 2, I just wrote down the equations.
We consider two hosts A and B. A is sending at time 0,  and B at time t.
We call b the propagation time between A and B.
For B to be able to send we must have :
     *    b > t  (B sends before it receives the first bits from A and there
will be collision)
or *   t  >  b + TRANSP  (B sends after it receives the last bits from A and
there is no collision)
We know that PROP > b > PROPMIN
Therefore :
in the first case we have   2 PROP > 2 b > b + t
and in the second case  we have b + t > 2 b + TRANSP > 2 PROPMIN + TRANSP
We want to be able distinguish between between these two cases, thus, we want
 2 PROPMIN + TRANSP > 2 PROP

Does anybody see where this is wrong ?

Thanks.
Francois-Marie.



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Jonathan Keljo" 
Newsgroups: su.class.cs244a
Subject: PS3 #2b Solution
Date: Sun, 17 Mar 2002 17:02:18 -0800
Lines: 33
Distribution: su
Message-ID: 
NNTP-Posting-Host: nordic.stanford.edu
X-Trace: news.Stanford.EDU 1016413339 448 128.12.133.48 (18 Mar 2002 01:02:19 GMT)
X-Complaints-To: 
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3617

I'm trying to figure out why the answer is 2PROP - PROPMIN instead of
2PROP-2PROPMIN.

The solution says that anything that arrives within PROPMIN since the last
packet left must have collided with the last packet. This is certainly true,
but I believe you can make the same assertion for anything that arrives
within 2PROPMIN.

Say the last bit of packet P1 leaves A at time t=0. It arrives at B no
earlier than time t=PROPMIN. B can then start sending, and its first bit of
data arrives at A at time t=2PROPMIN. Thus, anything that arrives before
that, regardless of source, must have collided with P1 at B.

The example of C starting to send just before the first bit of A's packet
arrives still holds. C detects the collision immediately because it is
receiving and sending at the same time. A receives the first bit of C's
packet somewhat less than 2PROPMIN after it stopped sending, so it knows
that C's packet collided with A's last one.

Whether you reduce the transmission time by PROPMIN or 2PROPMIN, it's
possible that A sent back-to-back packets, so that C's packet collided with
both of A's packets. The first would be detected by a timer, and the second
would be detected in the usual manner.

What's wrong with this reasoning?

Thanks,

Jonathan




.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Romain Thibaux" 
Newsgroups: su.class.cs244a
Subject: Re: Questions about the solution of Problem 4 (Q1 & 2)
Date: Sun, 17 Mar 2002 17:57:45 -0800
Lines: 8
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: thibaux.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4133.2400
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4133.2400
Xref: nntp.stanford.edu su.class.cs244a:3618

I completely agree. I had about the same explanations as you, with the same
conclusion, and the same grade. I really don't see where the flaw in these
reasonings is either, and I am not at all convinced by the official
solution, which in both cases is not a proof.

    Romain


.

Path: shelby.stanford.edu!nntp.stanford.edu!myth7.Stanford.EDU!casado
From: Martin Casado 
Newsgroups: su.class.cs244a
Subject: Re: PS3 #2b Solution
Date: Sun, 17 Mar 2002 20:15:48 -0800
Lines: 57
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: myth7.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3619

> I'm trying to figure out why the answer is 2PROP - PROPMIN instead of
> 2PROP-2PROPMIN.
>
> The solution says that anything that arrives within PROPMIN since the last
> packet left must have collided with the last packet. This is certainly true,
> but I believe you can make the same assertion for anything that arrives
> within 2PROPMIN.
>
> Say the last bit of packet P1 leaves A at time t=0. It arrives at B no
> earlier than time t=PROPMIN. B can then start sending, and its first bit of
> data arrives at A at time t=2PROPMIN. Thus, anything that arrives before
> that, regardless of source, must have collided with P1 at B.
>
> The example of C starting to send just before the first bit of A's packet
> arrives still holds. C detects the collision immediately because it is
> receiving and sending at the same time. A receives the first bit of C's
> packet somewhat less than 2PROPMIN after it stopped sending, so it knows
> that C's packet collided with A's last one.
>
> Whether you reduce the transmission time by PROPMIN or 2PROPMIN, it's
> possible that A sent back-to-back packets, so that C's packet collided with
> both of A's packets. The first would be detected by a timer, and the second
> would be detected in the usual manner.
>
> What's wrong with this reasoning?
>

Hi,

  I think under some conditions your solution can produce incorrect results.
  One of the problems is that 2PROP - 2PROPMIN in a network where PROP =
  PROPMIN allows for any size packet.  This becomes problematic in a hubbed
  star network.  Assume the topology below:

      B
      |
  A - H - C

  Lets say A and B and C are all PROPMIN distance apart.  Since we are allowed
  to send a single bit packet (or any proportionally insignificant sized
  packet), A can send a packet p to B. After p hits the hub, it is home free to
  B.  Now assume, that, moments before p hits the hub, C sends a very small
  packet.  It is quite possible that p goes through the hub, on to B with no
  problems.  However, C's very small packet (assuming a non-switched hub) will
  have been mangled. A, gets the mangled remains of C's packet and assumes that
  there was a collision, and sends a duplicate.

  I don't think you run into this situation if TRANSP >= 2PROP - PROPMIN, since
  the smallest packet on any network must be at least of size PROP.

  fwiw, I think needing a counter-example using a hub is a stretch and I find
  your answer to be perfectly reasonable. Secondly, I'm not even sure if it
  is valid to have a hub within PROPMIN distance of a node, however according
  to the posted solution, it is OK.

                        ~~m

.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Hopil Bae" 
Newsgroups: su.class.cs244a
Subject: Re: PS3 #2b Solution
Date: Sun, 17 Mar 2002 23:46:14 -0800
Lines: 109
Distribution: su
Message-ID: 
References:  
NNTP-Posting-Host: hopils.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 5.50.4522.1200
X-MimeOLE: Produced By Microsoft MimeOLE V5.50.4522.1200
Xref: nntp.stanford.edu su.class.cs244a:3620


Hi..

May I add my words?..
I'm also more inclined to 2(PROP-PROPMIN)..
I assumed that A sends packets 1 and 2, with a small gap between them
so that other hosts may start transmitting during the gap..
And the objective is to differentiate between "1 colliding with B" and "2
colliding with C"
Then, if A starts transmitting packet number 1 at t=0,
then A can detect collision with B no later than 2*PROP
And since C can start transmitting no earlier than TRANSP+PROPMIN,
A can detect collision with C no earlier than TRANSP+2*PROPMIN.
Then to differentiate between the two cases,
2*PROP <= TRANSP + 2*PROPMIN...

It seems to me that TRANSP>=2*PROP- PROPMIN has extra margin of PROPMIN..
...
And I think c's packet need not be mangled in Martin's example..
p and c's packet do cross each other on the link between C and H,
but they don't interfere with each other.. Just propagate independently..
So minimum length of 0 doesn't seem strange to me..
Say if A sends a small packet a at t=0,
and sees a small packet c from C at t=PROP,
then A can clearly know that a and c arrived at B at the same time(t=PROP).

There may be some holes in my reasoning...
Anyway, good luck for everyone...
Hopil Bae

"Martin Casado"  wrote in message

> > I'm trying to figure out why the answer is 2PROP - PROPMIN instead of
> > 2PROP-2PROPMIN.
> >
> > The solution says that anything that arrives within PROPMIN since the
last
> > packet left must have collided with the last packet. This is certainly
true,
> > but I believe you can make the same assertion for anything that arrives
> > within 2PROPMIN.
> >
> > Say the last bit of packet P1 leaves A at time t=0. It arrives at B no
> > earlier than time t=PROPMIN. B can then start sending, and its first bit
of
> > data arrives at A at time t=2PROPMIN. Thus, anything that arrives before
> > that, regardless of source, must have collided with P1 at B.
> >
> > The example of C starting to send just before the first bit of A's
packet
> > arrives still holds. C detects the collision immediately because it is
> > receiving and sending at the same time. A receives the first bit of C's
> > packet somewhat less than 2PROPMIN after it stopped sending, so it knows
> > that C's packet collided with A's last one.
> >
> > Whether you reduce the transmission time by PROPMIN or 2PROPMIN, it's
> > possible that A sent back-to-back packets, so that C's packet collided
with
> > both of A's packets. The first would be detected by a timer, and the
second
> > would be detected in the usual manner.
> >
> > What's wrong with this reasoning?
> >
>
> Hi,
>
>   I think under some conditions your solution can produce incorrect
results.
>   One of the problems is that 2PROP - 2PROPMIN in a network where PROP =
>   PROPMIN allows for any size packet.  This becomes problematic in a
hubbed
>   star network.  Assume the topology below:
>
>       B
>       |
>   A - H - C
>
>   Lets say A and B and C are all PROPMIN distance apart.  Since we are
allowed
>   to send a single bit packet (or any proportionally insignificant sized
>   packet), A can send a packet p to B. After p hits the hub, it is home
free to
>   B.  Now assume, that, moments before p hits the hub, C sends a very
small
>   packet.  It is quite possible that p goes through the hub, on to B with
no
>   problems.  However, C's very small packet (assuming a non-switched hub)
will
>   have been mangled. A, gets the mangled remains of C's packet and assumes
that
>   there was a collision, and sends a duplicate.
>
>   I don't think you run into this situation if TRANSP >= 2PROP - PROPMIN,
since
>   the smallest packet on any network must be at least of size PROP.
>
>   fwiw, I think needing a counter-example using a hub is a stretch and I
find
>   your answer to be perfectly reasonable. Secondly, I'm not even sure if
it
>   is valid to have a hub within PROPMIN distance of a node, however
according
>   to the posted solution, it is OK.
>
>                         ~~m
>


.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: francois-marie lefevere 
Newsgroups: su.class.cs244a
Subject: Re: PS3 #2b Solution
Date: Mon, 18 Mar 2002 00:39:02 -0800
Lines: 8
Distribution: su
Message-ID: 
References:   
NNTP-Posting-Host: nastia.stanford.edu
Mime-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
X-Trace: news.Stanford.EDU 1016440746 3937 128.12.187.116 (18 Mar 2002 08:39:06 GMT)
X-Complaints-To: 
X-Mailer: Mozilla 4.77 [en] (X11; U; Linux 2.4.9-31 i686)
X-Accept-Language: en
Xref: nntp.stanford.edu su.class.cs244a:3621

>

Same for me, I am more inclined to 2*(PROP-PROPMIN) .
See thread  uncorrectly labelled "Questions about the solution of Problem 4 (Q1
& 2)"



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: HO#3 slide 28.. and TCP sender Tx window
Date: 18 Mar 2002 11:14:44 GMT
Lines: 36
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine22.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3622


: 1)
: I'm wondering about the latency equation in slide 28 of HO#3,
: which says that the latency is M/R_MIN + sum(PROP_i)..
: M/R_MIN doesn't increase as the number of hops increases,
: so it looks counterintuitive to me...
: I think latency should be something like sum(P/R)+sum(PROP_i)
: where P is the packet size... 

Yes, the total end to end delay would be M/R + sum(P/Ri) (i=1 to 3) and 
sum(PROP_i) (i=0 to 3) (PROP_0 is prop from A->R1), from start of 
transmission at A to end of reception at B.

: 2) 
: And as in the midterm TCP problem,
: Tx sends packet 4,5,6,7 when Tx win_size=4,
: and 5,6,7 are received but 4 is lost...
: Then Tx win size becomes 2, so only 4 and 5 are retransmitted..
: 4 is received OK this time,
: then since Rx has received 4,5,6,7 correctly, it will send ack for 7.


: But now Tx window includes only 4 and 5..
: 7 is outside the window..Then, should Tx ignore the ack?
: although ignoring would result in a deadlock..
: And if Tx accepts ack for 7, then maybe it should send 8,9,10...
: Or should it send 6,7,8?

Yes this is correct. In the next window 8,9,10 could have been sent. This 
was the alternate solution to the problem and we accepted it as correct 
when grading the problem. Under this scenario things become a bit more 
complicated when you consider the loss of 9's ACK; if you are curious, you 
can think about this. However, this is not the interpretation we had 
intended.

Arun
.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: PS 1 problem 7
Date: 18 Mar 2002 11:35:39 GMT
Lines: 27
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine22.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3623

If you consider the diagram, bits only arrive during the periods i <= t <= 
i+1, i=0, 1, ....

In this period the arrival process can be written as 100i + 1000(t-i).

The departure process is always 100(t-.1).

Now we want to solve for d(t) in the intervals when bits arrive. The delay 
is departure time (td) - arrival time (t). Note at this time both arrival 
and departure processes have the same value and thus:

100i + 1000(t-i) = 100(td - .1) =>
td = i + 10(t-i) + .1 = -9i + 10t + .1

d(t) = td - t = (-9i + 10t + .1) - t = 9(t-i) + .1, in i <= t <= i+.1

Arun

Shankar Agarwal  wrote:
: Hi,
: Can you please elaborate the d(t) for this problem
: I am not able to understand how did d(t) = 9(t-i) + 0.1 i <= t <=i+0.1
: came.
: Thanks
: Shankar


.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine22.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS3 #2b Solution
Date: 18 Mar 2002 13:31:59 GMT
Lines: 10
Distribution: su
Message-ID: 
References:    
NNTP-Posting-Host: elaine22.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3624


If you use 2*(PROP-PROPMIN), what happens in the case where you have two
nodes?

A-----B (so PROP=PROPMIN).  This is even simpler than Martin's example.

You'd have no requirements on TRANSP (short of TRANSP > 0)--which means if
there's a collision, a host might have absolutely no idea which packet
collided!

.

Path: shelby.stanford.edu!nntp.stanford.edu!saga2.Stanford.EDU!mrawashd
From: Moh'd Saleem Saleem Alrawashdeh 
Newsgroups: su.class.cs244a
Subject: Re: PS3 #2b Solution
Date: Mon, 18 Mar 2002 09:34:10 -0800
Lines: 41
Distribution: su
Message-ID: 
References:  
  
 
NNTP-Posting-Host: saga2.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
To: Matthew Jonathan Holliman 
In-Reply-To: 
Xref: nntp.stanford.edu su.class.cs244a:3625

Hi,

I stated in my solution that the host should keep sensing the line for
2*PROPmin before it decided if there was a collision or not. For the case
mentioned below (where there is no min boundary on the size of packet), If
Host A sends 1 byte, and senses the line for 2*PROPmin and found it idle,
then I argue that there was no collision.

> If you use 2*(PROP-PROPMIN), what happens in the case where you have two
> nodes?
>
> A-----B (so PROP=PROPMIN).  This is even simpler than Martin's example.


Even if in above network, packet size can be virtually zero, but if a
collision occurs, the then host will know about it. The host has to keep
monitoring the line for 2*PROPmin, which in above case is the same as
2*PROP. I wasn't able to come with counter example in which 2*PROPmin is
not enough to detect a collision for the upper case. Even for worest case
scenario, where A sends a 1 byte at t0, B sends at t0+PROP-epsilon, they
collided very near to B. Then the collision news will reach A at
t0+2*PROP-epsilon, which is shorter than its listening period, and it will
detect that there was a collision.

> You'd have no requirements on TRANSP (short of TRANSP > 0)--which means if
> there's a collision, a host might have absolutely no idea which packet
> collided!

If ths host hears any collision in the 2*PROPmin after sending, then it
will know that it was its packet who collided.

I agree with Martin that in a star Ethernet, the above might not work.
However, the question has a graph shows a conventional shared bus
Ethernet, on which the result (2*PROP - 2*PROPmin) should work.

Good luck for you all,

Moh'd



.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine40.Stanford.EDU!holliman
From:  (Matthew Jonathan Holliman)
Newsgroups: su.class.cs244a
Subject: Re: PS3 #2b Solution
Date: 18 Mar 2002 18:58:31 GMT
Lines: 55
Distribution: su
Message-ID: 
References:        
NNTP-Posting-Host: elaine40.stanford.edu
X-Newsreader: NN version 6.5.4 (NOV)
Xref: nntp.stanford.edu su.class.cs244a:3626


If A only sends one packet, you're correct--in the example I gave, it
could detect that its packet collided.  But what if it sent two packets
in that time?  Then I don't see how it can distinguish them.  (Imposing
that A cannot transmit again until time 2*PROP would be a rather inefficient
use of the network!)

I think the figure was intended to clarify the meaning of PROPMIN, not
necessarily indicate the topology to assume--after all, TRANSP >= 2*PROP
holds independent of the precise topology in CSMA/CD.


Moh'd Saleem Saleem Alrawashdeh  writes:

>Hi,

>I stated in my solution that the host should keep sensing the line for
>2*PROPmin before it decided if there was a collision or not. For the case
>mentioned below (where there is no min boundary on the size of packet), If
>Host A sends 1 byte, and senses the line for 2*PROPmin and found it idle,
>then I argue that there was no collision.

>> If you use 2*(PROP-PROPMIN), what happens in the case where you have two
>> nodes?
>>
>> A-----B (so PROP=PROPMIN).  This is even simpler than Martin's example.


>Even if in above network, packet size can be virtually zero, but if a
>collision occurs, the then host will know about it. The host has to keep
>monitoring the line for 2*PROPmin, which in above case is the same as
>2*PROP. I wasn't able to come with counter example in which 2*PROPmin is
>not enough to detect a collision for the upper case. Even for worest case
>scenario, where A sends a 1 byte at t0, B sends at t0+PROP-epsilon, they
>collided very near to B. Then the collision news will reach A at
>t0+2*PROP-epsilon, which is shorter than its listening period, and it will
>detect that there was a collision.

>> You'd have no requirements on TRANSP (short of TRANSP > 0)--which means if
>> there's a collision, a host might have absolutely no idea which packet
>> collided!

>If ths host hears any collision in the 2*PROPmin after sending, then it
>will know that it was its packet who collided.

>I agree with Martin that in a star Ethernet, the above might not work.
>However, the question has a graph shows a conventional shared bus
>Ethernet, on which the result (2*PROP - 2*PROPmin) should work.

>Good luck for you all,

>Moh'd



.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: "Mudassir I Sheikha" 
Newsgroups: su.class.cs244a
Subject: Midterm Q 3
Date: Mon, 18 Mar 2002 12:03:52 -0800
Lines: 17
Distribution: su
Message-ID: 
NNTP-Posting-Host: pakistan.stanford.edu
X-Priority: 3
X-MSMail-Priority: Normal
X-Newsreader: Microsoft Outlook Express 6.00.2600.0000
X-MimeOLE: Produced By Microsoft MimeOLE V6.00.2600.0000
Xref: nntp.stanford.edu su.class.cs244a:3627

tcp congestion control question in the midterm:

the ack for the 9th packet is lost. the solution says we need to retransmit
packet 9 and everything after that.

i am not sure why we need to do that since we have received acks with higher
sequence numbers. can't we assume that the 9th packet was successfully
received?

or is it that we decide to timeout packet 9 before we see the other acks?

mudassir





.

Path: shelby.stanford.edu!nntp.stanford.edu!not-for-mail
From: Arun Upadhyaya Kishan 
Newsgroups: su.class.cs244a
Subject: Re: Midterm Q 3
Date: 18 Mar 2002 21:39:31 GMT
Lines: 26
Distribution: su
Message-ID: 
References: 
NNTP-Posting-Host: elaine9.stanford.edu
User-Agent: tin/1.4.4-20000803 ("Vet for the Insane") (UNIX) (SunOS/5.8 (sun4u))
Xref: nntp.stanford.edu su.class.cs244a:3628


Yes, right at the start of Window #7 there is no ACK for 9 received, 
resulting in a timeout.

Arun

Mudassir I Sheikha  wrote:


: tcp congestion control question in the midterm:

: the ack for the 9th packet is lost. the solution says we need to retransmit
: packet 9 and everything after that.

: i am not sure why we need to do that since we have received acks with higher
: sequence numbers. can't we assume that the 9th packet was successfully
: received?

: or is it that we decide to timeout packet 9 before we see the other acks?

: mudassir





.

Path: shelby.stanford.edu!nntp.stanford.edu!elaine35.Stanford.EDU!dhawal
From: Dhawal Kumar 
Newsgroups: su.class.cs244a
Subject: Best assignment
Date: Wed, 20 Mar 2002 23:37:09 -0800
Lines: 5
Distribution: su
Message-ID: 
NNTP-Posting-Host: elaine35.stanford.edu
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Xref: nntp.stanford.edu su.class.cs244a:3629

If Martin Casado doesn't have any problem can I have a sample code from
him - I am just curious to know what really got the TAs impressed.

Dhawal Kumar

.