From owner-freebsd-questions Fri Jun 20 07:11:23 1997 Return-Path: Received: (from root@localhost) by hub.freebsd.org (8.8.5/8.8.5) id HAA04658 for questions-outgoing; Fri, 20 Jun 1997 07:11:23 -0700 (PDT) Received: from implode.root.com (implode.root.com [198.145.90.17]) by hub.freebsd.org (8.8.5/8.8.5) with ESMTP id HAA04649 for ; Fri, 20 Jun 1997 07:11:18 -0700 (PDT) Received: from localhost (localhost [127.0.0.1]) by implode.root.com (8.8.5/8.8.5) with SMTP id HAA13705; Fri, 20 Jun 1997 07:12:40 -0700 (PDT) Message-Id: <199706201412.HAA13705@implode.root.com> X-Authentication-Warning: implode.root.com: localhost [127.0.0.1] didn't use HELO protocol To: Josef Karthauser cc: questions@FreeBSD.ORG Subject: Re: Apache 1.1.3 and 1.2.0 problems under FBSD2.2.2 In-reply-to: Your message of "Fri, 20 Jun 1997 15:00:30 BST." <19970620150029.55055@pavilion.net> From: David Greenman Reply-To: dg@root.com Date: Fri, 20 Jun 1997 07:12:40 -0700 Sender: owner-questions@FreeBSD.ORG X-Loop: FreeBSD.org Precedence: bulk >On Fri, Jun 20, 1997 at 05:07:02AM -0700, David Greenman wrote: >> >Any suggestions, other than a full frontal labotamy, would gratefully >> >be received. >> >> What do you have NMBCLUSTERS set to? A busy WWW server needs a fairly high >> value for that - many times the default. >> > >It appears that: > NMBCLUSTERS (512 + MAXUSERS * 16) >therefore > NMBCLUSTERS=512 + 256*16. > >Can it be changed with sysctl? No, it must be changed with a kernel option. It determines the amount of virtual address space that is available to allocate mbuf clusters. Add: options "NMBCLUSTERS=8000" To your kernel config file. You can monitor the peak by looking at 'netstat -m'...for instance: [wcarchive:davidg] netstat -m 17687 mbufs in use: 14705 mbufs allocated to data 2968 mbufs allocated to packet headers 11 mbufs allocated to protocol control blocks 3 mbufs allocated to socket names and addresses this-> 13089/18678 mbuf clusters in use 39566 Kbytes allocated to network (-34% in use) 0 requests for memory denied 0 requests for memory delayed 0 calls to protocol drain routines The line marked above is current/peak. You don't ever want the peak to be anywhere near the maximum (as determined by NMBCLUSTERS). -DG David Greenman Core-team/Principal Architect, The FreeBSD Project