Re: Timeout Status Code

• Mail folder: WWW Talk Apr 94-present
• Next message: Jeremy Wertheimer: "WIT"
• Previous message: Achim Jung: "freewais and WWWWais"
• Maybe in reply to: John C. Mallery: "Re: Timeout Status Code"

Errors-To: listmaster@www0.cern.ch
Date: Tue, 14 Jun 1994 02:29:24 +0200
Errors-To: listmaster@www0.cern.ch
Message-id: <199406132128.OAA01872@nic.cerf.net>
Errors-To: listmaster@www0.cern.ch
Reply-To: miked@CERF.NET
Originator: www-talk@info.cern.ch
Sender: www-talk@www0.cern.ch
Precedence: bulk
From: miked@CERF.NET (Michael A. Dolan)
To: Multiple recipients of list <www-talk@www0.cern.ch>
Subject: Re: Timeout Status Code
X-Listprocessor-Version: 6.0c -- ListProcessor by Anastasios Kotsikonas
X-Mailer: <Windows Eudora Version 2.0.2>
X-Mailer: <Windows Eudora Version 2.0.2>
Content-Type: text/plain; charset="us-ascii"
Content-Type: text/plain; charset="us-ascii"
Mime-Version: 1.0
Mime-Version: 1.0

At 03:10 PM 6/13/94, John C. Mallery wrote:
> timeouts after the connection is idle for 5 minutes.

"Idle" is a funny thing.  In systems with lots of socket buffering, many
small to medium document transfers will allow it to be completely written
to the socket.  The httpd will then hang up in the socket close waiting for
the socket buffers to drain.

If you're fortunate enough to hang in a write (or in a select waiting for it
to be writable), then a reasonable timeout is a function of the socket
buffer size.  Don't forget to factor in the socket buffer size of the client
which, of course, there's no way to know.

Also, what if your httpd is proxying a slow PH or WAIS server ?  These
can easily take 5 minutes with no traffic.  The user may be Buddha :-)

If you get a socket connect with no GET (or similar), then timeout real fast.
If you're already responding to a GET (or similar) then it's a hard question.

        Mike

------------------------------------
Michael A. Dolan - miked@cerfnet.com 
TerraByte Technology (619) 445-9070