From owner-freebsd-questions  Tue Oct  8 20:55:37 1996
Return-Path: owner-questions
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id UAA01978
          for questions-outgoing; Tue, 8 Oct 1996 20:55:37 -0700 (PDT)
Received: from root.com (implode.root.com [198.145.90.17])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id UAA01967
          for <questions@freebsd.org>; Tue, 8 Oct 1996 20:55:30 -0700 (PDT)
Received: from localhost (localhost [127.0.0.1]) by root.com (8.7.5/8.6.5) with SMTP id UAA23496; Tue, 8 Oct 1996 20:55:44 -0700 (PDT)
Message-Id: <199610090355.UAA23496@root.com>
X-Authentication-Warning: implode.root.com: Host localhost [127.0.0.1] didn't use HELO protocol
To: "doc" <doc@zynetwc.com>
cc: questions@freebsd.org
Subject: Re: TCB's 
In-reply-to: Your message of "Mon, 08 Jul 1996 18:21:30 PDT."
             <199610090112.SAA11671@usr1.zynetwc.com> 
From: David Greenman <dg@root.com>
Reply-To: dg@root.com
Date: Tue, 08 Oct 1996 20:55:44 -0700
Sender: owner-questions@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

>I have a firly high volume www server runnin under FreeeBSD 1.5 and Apache.
                                                             ^^^
   I assume you mean 2.1.5?

>My question simply is.. Is thre any way I can increase the TCB's ( Transmit
>Control Buffers ). 

   You're probably running out of mbuf clusters or possibly reaching the
listen() backlog limit. If you see any "out of mbuf clusters" messages in
your /var/log/messages file, then you should either increase the maxusers
parameter in your kernel config file or add:

options "NMBCLUSTERS=4000"

   ...to it. As for the listen() backlog limit, it's difficult to tell in
2.1.5 if this is your problem. We added a statistic to '-stable' called
"listen queue overflows" and can be found in the 'netstat -s' stats.

>Durring especially high volume periods my and my users browser reports "web
>site found waiting for reply" for more than a fiew seconds at times. I
>assume it's because of waiting for a new buffer to open for the connection.

   Probably the listen queue limit. You'll have to look at the Apache source
to see what limit they're using by default. If it is >128, then you're hitting
the system maximum. You can increase the system maximum in 2.1.5 by changing
SOMAXCONN in /sys/sys/socket.h to something larger.

-DG

David Greenman
Core-team/Principal Architect, The FreeBSD Project