From owner-freebsd-current@FreeBSD.ORG Thu Aug 12 19:44:47 2004 Return-Path: Delivered-To: freebsd-current@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id 2C85116A4CE; Thu, 12 Aug 2004 19:44:47 +0000 (GMT) Received: from gw.catspoiler.org (217-ip-163.nccn.net [209.79.217.163]) by mx1.FreeBSD.org (Postfix) with ESMTP id A360543D1F; Thu, 12 Aug 2004 19:44:46 +0000 (GMT) (envelope-from truckman@FreeBSD.org) Received: from FreeBSD.org (mousie.catspoiler.org [192.168.101.2]) by gw.catspoiler.org (8.12.11/8.12.11) with ESMTP id i7CJib4f021202; Thu, 12 Aug 2004 12:44:41 -0700 (PDT) (envelope-from truckman@FreeBSD.org) Message-Id: <200408121944.i7CJib4f021202@gw.catspoiler.org> Date: Thu, 12 Aug 2004 12:44:37 -0700 (PDT) From: Don Lewis To: rwatson@FreeBSD.org In-Reply-To: <200408112123.i7BLNZfX018166@gw.catspoiler.org> MIME-Version: 1.0 Content-Type: TEXT/plain; charset=us-ascii cc: mb@imp.ch cc: freebsd-current@FreeBSD.org Subject: Re: SCHEDULE and high load situations X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 12 Aug 2004 19:44:47 -0000 On 11 Aug, To: rwatson@freebsd.org wrote: > I've noticed in fork()/exec() intensive loads like buildworld and > building ports on a UP box that ULE favors a CPU-bound but niced process > like setiathome over the software build. Even the longer c++ compile > steps seem to get less CPU time, than the niced CPU-bound process, at > least according to top. Buildworld times with ULE on my Athlon XP box > are about 145 minutes with ULE and 82 minutes with 4BSD when competing > with setiathome. Top shows setiathome consistently getting about 55% of > the CPU with ULE. Without setiathome running, buildworld takes about 65 > minutes. Here's a case where there is not much fork()/exec() activity (last pid is stable), and there doesn't appear to be much I/O, but the niced process is getting more than 50% of the CPU. I noticed this during a portupgrade of editors/openoffice-1.1. The CPU% for setiathome seems to stay right around in this range +/- a percent or so, with spikes higher when the other processes normally competing for CPU time are waiting for I/O. 57 processes: 3 running, 53 sleeping, 1 stopped CPU states: 42.8% user, 56.0% nice, 1.2% system, 0.0% interrupt, 0.0% idle Mem: 99M Active, 628M Inact, 156M Wired, 31M Cache, 111M Buf, 83M Free Swap: 2055M Total, 84K Used, 2055M Free Seconds to delay: PID USERNAME PRI NICE SIZE RES STATE TIME WCPU CPU COMMAND 504 setiathome 139 15 17744K 16884K RUN 719:22 54.69% 54.69% setiathome 64008 root 139 0 4232K 3832K RUN 11:50 41.41% 41.41% dmake 568 dl 76 0 6084K 2324K select 5:08 0.00% 0.00% sshd 483 uucp 76 0 1336K 988K select 2:36 0.00% 0.00% newapc 16995 root 76 0 1284K 760K select 2:21 0.00% 0.00% script 12823 root -8 0 1200K 552K piperd 2:03 0.00% 0.00% tee 485 uucp 76 0 1308K 944K select 0:33 0.00% 0.00% upsd 53855 root 8 0 4956K 4212K wait 0:13 0.00% 0.00% perl 489 uucp 8 0 1316K 1012K nanslp 0:12 0.00% 0.00% upsmon 410 root 76 0 2876K 1500K select 0:10 0.00% 0.00% ntpd 59764 dl 76 0 2400K 1604K RUN 0:07 0.00% 0.00% top 437 root 76 0 3436K 2300K select 0:05 0.00% 0.00% sendmail 12822 root 8 0 21348K 20792K wait 0:02 0.00% 0.00% ruby18 490 uucp 8 0 1272K 908K nanslp 0:01 0.00% 0.00% upslog 1101 dl 76 0 6084K 2324K select 0:01 0.00% 0.00% sshd 454 root 8 0 1356K 1004K nanslp 0:01 0.00% 0.00% cron 657 root 20 0 2440K 1940K pause 0:01 0.00% 0.00% csh