From owner-freebsd-isp  Fri Aug 30 06:53:17 1996
Return-Path: owner-isp
Received: (from root@localhost)
          by freefall.freebsd.org (8.7.5/8.7.3) id GAA20295
          for isp-outgoing; Fri, 30 Aug 1996 06:53:17 -0700 (PDT)
Received: from brasil.moneng.mei.com (brasil.moneng.mei.com [151.186.109.160])
          by freefall.freebsd.org (8.7.5/8.7.3) with ESMTP id GAA20286
          for <isp@freebsd.org>; Fri, 30 Aug 1996 06:53:15 -0700 (PDT)
Received: (from jgreco@localhost) by brasil.moneng.mei.com (8.7.Beta.1/8.7.Beta.1) id IAA08372; Fri, 30 Aug 1996 08:52:08 -0500
From: Joe Greco <jgreco@brasil.moneng.mei.com>
Message-Id: <199608301352.IAA08372@brasil.moneng.mei.com>
Subject: Re: Max users/max processes in FreeBSD??
To: froden@bigblue.no
Date: Fri, 30 Aug 1996 08:52:08 -0500 (CDT)
Cc: isp@freebsd.org
In-Reply-To: <199608292312.BAA14425@login.bigblue.no> from "Frode Nordahl" at Aug 30, 96 01:13:51 am
X-Mailer: ELM [version 2.4 PL24]
Content-Type: text
Sender: owner-isp@freebsd.org
X-Loop: FreeBSD.org
Precedence: bulk

> login:~# limit [vt220]
> cputime         unlimited
> filesize        unlimited
> datasize        65536 kbytes
> stacksize       8192 kbytes
> coredumpsize    unlimited
> memoryuse       14988 kbytes
> descriptors     64
> memorylocked    9994 kbytes
> maxproc         40
> login:~# uname -a
> FreeBSD login.bigblue.no 2.1.0-RELEASE FreeBSD 2.1.0-RELEASE #0: Wed Aug  7 14:5
> 4:17 MET DST 1996     root@login.bigblue.no:/usr/src/sys/compile/LOGIN  i386
> login:~#
> 
> This is what I get on FreeBSD 2.1.0-STABLE (Cannot run 2.1.5 on this machine due to incompatibility of some 
> nature)
> 
> So, the MAXUSER directive in the kernel confiugration can be set to whatever I want?  You know where one 
> might find the proc directive?

The MAXUSERS directive can be set to whatever.  However, it is NOT a limit
to the number of simultaneous users you can have online!  It is simply a
guideline that helps the kernel scale a number of things to the size you
expect to need.  Obviously running 200 users on a box with MAXUSERS 10 may
fail... probably due to proc table overflow...  but running 20 is quite
possible and reasonable.

Last I checked, and it was a while ago, MAXUSERS directly or indirectly
contributes to the sizes of things like the proc and maxfiles and MBCLUSTERS
variables.  These are all things you want larger on a busy system.

As for "maxproc" - type "unlimit; limit" :-)  There is a way to change 
the 40 default in the kernel but it is generally considered the wrong thing
to do so.

Also note, MAXUSERS has a warning if you go > 64.  This is mainly to 
warn people who are doing silly things like setting MAXUSERS=2048 on a 
16MB box.  The setting of MAXUSERS does affect kernel resource consumption,
so be reasonable with it.

The old Sun guidelines are still pretty good:

Start with 16
Add one for every logged in user you expect
Add one for every NFS client you expect
Add one for every xterm window you expect
Round up to an even power of 2

Side note: Usually you are doing something wrong if MAXUSERS > Memory Size
in MB..

... JG