From: Jim Jagielski <jim@jaguNET.com>
Message-Id: <199701261638.LAA13614@shado.jaguNET.com>
Subject: Re: SO_LINGER and SO_DONTLINGER
To: new-httpd@hyperreal.com
Date: Sun, 26 Jan 1997 11:38:44 -0500 (EST)
In-Reply-To: <32EB827A.167EB0E7@topsail.org> from "Chuck Murcko" at Jan 26,
 97 11:12:42 am
Content-Type: text
Sender: new-httpd-owner@apache.org
Precedence: bulk
Reply-To: new-httpd@hyperreal.com

Chuck Murcko wrote:
> 
> Jim Jagielski wrote:
> > 
> > My TCP/IP handbook is miles away... Anyone know offhand waht SO_DONTLINGER
> > does to process the request "as quickly as possible"?
> > 
> > The reason: I have 2 A/UX boxes which I'm testing lingering_close on.
> >   Every 10 mins I have a cron job check the number of WAIT_2 stats
> >   and log that. It's looking like lingering_close causes more of
> >   these than without (~20 on one server and ~40 on another, each
> >   ave about 50khits/day or so). I'm thinking about seeing what
> >   SO_LINGER does (I seem to recall that it was buggy on SysV4
> >   platforms mostly) to help, but am also curious what SO_DONTLINGER
> >   might do.
> > 
> If SO_LINGER is set, and the linger time is zero, any data remaining to
> be sent is discarded when the socket is closed. If the linger time is
> nonzero (positive 8^) the system attempts to deliver unsent data for the
> time in seconds specified.
> 

Yeah, that I remember... But I can't recall for the life of me
what SO_DONTLINGER does :/

I'm thinking that if the OS does have an internal WAIT_2 timer,
then setting SO_LINGER might cause problems. If it doesn't and
the OS isn't a nasty SysV4, then SO_LINGER might be a good thing
to try.

Maybe I'll grab some TCP/IP stack source and take a gander :)


-- 
====================================================================
      Jim Jagielski            |       jaguNET Access Services
     jim@jaguNET.com           |       http://www.jaguNET.com/
                  "Not the Craw... the CRAW!"