From nobody@hyperreal.com  Tue Dec 24 06:49:02 1996
Received: by taz.hyperreal.com (8.8.3/V2.0) id GAA00719; Tue, 24 Dec 1996 06:49:02 -0800 (PST)
Message-Id: <199612241449.GAA00719@taz.hyperreal.com>
Date: Tue, 24 Dec 1996 06:49:02 -0800 (PST)
From: Jeff Finkelstein <finkels@bsdi.com>
Reply-To: finkels@bsdi.com
To: apbugs@hyperreal.com
Subject: Connections left in FIN_WAIT_2 state
X-Send-Pr-Version: 3.2

>Number:         57
>Category:       protocol
>Synopsis:       Connections left in FIN_WAIT_2 state
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    apache
>State:          closed
>Class:          sw-bug
>Submitter-Id:   apache
>Arrival-Date:   Tue Dec 24 06:50:01 1996
>Last-Modified:  Fri Jan 23 08:08:56 PST 1998
>Originator:     finkels@bsdi.com
>Organization:
>Release:        1.2b2
>Environment:
BSDI 2.1 and 3.0
gcc 2.7.2
>Description:
Possibly starting in Apache 1.2b2, something changed that causes connections to
be left in the FIN_WAIT_2 state (we have some customers with >4000 connections
left in this state). This did not happen in previous releases of Apache 
(at least it was not reported to us). I noticed in the code that there was
quite a bit added dealing with KEEPALIVE that may be affecting this. I do not
rule out that something may be wrong with our code that is being tickled by the
recent changes to Apache.
>How-To-Repeat:
One customer who is willing to try things is Luc Croteau 
<lcroteau@ls.barrhead.ab.ca> who runs an ISP in Canada. He may be able to setup
access to his site if you desire to see the problem live.
>Fix:
We have tried changing some TCP timer values to get the connections to close more
quickly, but are looking for a better solution
>Audit-Trail:
State-Changed-From-To: open-analyzed
State-Changed-By: marc
State-Changed-When: Fri Jan 24 08:04:28 PST 1997
State-Changed-Why:
Known problem.  Unfortunately, the solution is not equally
known.  Part of the problem is that Netscape is broken and
we can't do anything about that, but there are other issues
(with the lingering_close function) we are trying to resolve.
Responsible-Changed-From-To: gnats-admin (GNATS administrator)-marc
Responsible-Changed-By: marc
Responsible-Changed-When: Fri Jan 24 08:04:28 PST 1997
Responsible-Changed-Why:
I, among others, have been looking into it
State-Changed-From-To: analyzed-closed
State-Changed-By: dgaudet
State-Changed-When: Sun Apr 27 14:26:04 PDT 1997
State-Changed-Why:
FIN_WAIT_2 stuff is documented... and hopefully dealt with
as of 1.2b8.

Dean
Responsible-Changed-From-To: marc-apache
Responsible-Changed-By: coar
Responsible-Changed-When: Fri Jan 23 08:08:55 PST 1998
Responsible-Changed-Why:
Putting back into mainstream bugdb.
%0
>Unformatted: