From owner-freebsd-isp Tue Apr 15 11:28:09 1997 Return-Path: Received: (from root@localhost) by freefall.freebsd.org (8.8.5/8.8.5) id LAA27608 for isp-outgoing; Tue, 15 Apr 1997 11:28:09 -0700 (PDT) Received: from mail.webspan.net (mail.webspan.net [206.154.70.7]) by freefall.freebsd.org (8.8.5/8.8.5) with ESMTP id LAA27579 for ; Tue, 15 Apr 1997 11:28:05 -0700 (PDT) Received: from orion.webspan.net (orion.webspan.net [206.154.70.5]) by mail.webspan.net (WEBSPAN/970116) with ESMTP id OAA18154; Tue, 15 Apr 1997 14:27:15 -0400 (EDT) Received: from orion.webspan.net (localhost [127.0.0.1]) by orion.webspan.net (WEBSPN/970116) with ESMTP id OAA01905; Tue, 15 Apr 1997 14:27:14 -0400 (EDT) To: Alan Judge cc: Seppo Kallio , isp@freebsd.org From: "Gary Palmer" Subject: Re: Is someone running heavy loaded inn server? In-reply-to: Your message of "Tue, 15 Apr 1997 15:55:30 BST." <199704151455.PAA23521@indigo.ie> Date: Tue, 15 Apr 1997 14:27:14 -0400 Message-ID: <1903.861128834@orion.webspan.net> Sender: owner-isp@freebsd.org X-Loop: FreeBSD.org Precedence: bulk Alan Judge wrote in message ID <199704151455.PAA23521@indigo.ie>: > I'm running a large FreeBSD based server. It's not heavily loaded yet > (two full streaming feeds in and maybe 100 readers at peak times). > Load rarely climbs about 0.3. > > Setup: PPro 200, 256MB RAM, 8*4GB Quantum Atlas II Ultra Wide, 3940UW, > SMC EtherPower 10/100. > > Software: FreeBSD 2.2-STABLE (cvsuped kernel on 2.2.1 system), > INN-1.5.1, nntplink, newsxd. > > All works fine, as far as I can see, but the load is not that high > yet. We are taking a full feed (around 300K articles/day) without > problems, and the disks are ticking over nicely. > > One warning though. If you are using Adaptec disk controllers, make > sure to get the very latest kernel. Both 2.2.1-RELEASE and > 2.1.7.1-RELEASE have Adaptec problems under load and we had some hangs > and crashes before upgrading. root@news:~> uname -a FreeBSD news.webspan.net 2.2-CURRENT FreeBSD 2.2-CURRENT #0: Sat Nov 2 13:36:08 EST 1996 gpalmer@news.webspan.net:/usr/src/sys/compile/NEWS i386 root@news:~> uptime 2:22PM up 63 days, 1:26, 10 users, load averages: 2.59, 2.41, 2.43 200MHz PPro (256k cache) (one processor), 256Mb 60ns (true) parity ram, 5 4gig quantum atlas drives, 3940uw, smc 9332 10/100, and a 2 gig HP drive for root/newlib/etc. One note: apply the shared memory / mmap patches available. This will DRASTICALLY decrease the memory footprint of the nnrpd process. Pre patching, 120 readers would fill ram and 200Mb of swap. Post patching, I've gone past 150 readers and still had 150Mb swap left.... I'm running inn1.4unoff4 with about 6 full inbound feeds and 3 or 4 partials, although for all but 3 of those I send out more than I get. Also, more disks==good. CCD works fine. The only downside is when a drive dies, there is no recovering the ccd'd partition. However, news isn't THAT valuable. Gary -- Gary Palmer FreeBSD Core Team Member FreeBSD: Turning PC's into workstations. See http://www.FreeBSD.ORG/ for info