From owner-freebsd-current@FreeBSD.ORG Wed Aug 4 16:56:55 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 A195E16A4CE; Wed, 4 Aug 2004 16:56:55 +0000 (GMT) Received: from lakermmtao05.cox.net (lakermmtao05.cox.net [68.230.240.34]) by mx1.FreeBSD.org (Postfix) with ESMTP id 0DBCE43D1F; Wed, 4 Aug 2004 16:56:55 +0000 (GMT) (envelope-from conrads@cox.net) Received: from dolphin.local.net ([68.11.71.51]) by lakermmtao05.cox.net (InterMail vM.6.01.03.02.01 201-2131-111-104-103-20040709) with ESMTP <20040804165654.HYRK28993.lakermmtao05.cox.net@dolphin.local.net>; Wed, 4 Aug 2004 12:56:54 -0400 Received: from dolphin.local.net (localhost.local.net [127.0.0.1]) by dolphin.local.net (8.13.1/8.13.1) with ESMTP id i74Gurdu096795; Wed, 4 Aug 2004 11:56:53 -0500 (CDT) (envelope-from conrads@dolphin.local.net) Received: (from conrads@localhost) by dolphin.local.net (8.13.1/8.13.1/Submit) id i74GurXK096794; Wed, 4 Aug 2004 11:56:53 -0500 (CDT) (envelope-from conrads) Message-ID: X-Mailer: XFMail 1.5.5 on FreeBSD X-Priority: 3 (Normal) Content-Type: text/plain; charset=us-ascii Content-Transfer-Encoding: 8bit MIME-Version: 1.0 In-Reply-To: <20040804140103.69280.qmail@web50601.mail.yahoo.com> Date: Wed, 04 Aug 2004 11:56:53 -0500 (CDT) Organization: A Rag-Tag Band of Drug-Crazed Hippies From: "Conrad J. Sabatier" To: Kenneth Stailey cc: Mike Jakubik cc: freebsd-current@freebsd.org cc: freebsd-amd64@freebsd.org Subject: System load as crash factor with preemption??? X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.1 Precedence: list Reply-To: conrads@cox.net List-Id: Discussions about the use of FreeBSD-current List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Wed, 04 Aug 2004 16:56:55 -0000 On 04-Aug-2004 Kenneth Stailey wrote: > http://www.freebsd.org/releases/5.3R/todo.html > > << Problems persist with crashes and hangs under heavy load, > especially under SMP. The recent introduction of full-scale > preemption exasperated the problem, though it has been mitigated by a > recently committed hack. Speculation on the source of the problem > seems to center around races in critical section and spinlock > handling and nesting. This needs to be driven to root cause and > fixed in order for 5.3 to be considered STABLE. >> One thing I've been wondering about in all the postings I've seen re: the preemption-related crashes: reference is made repeatedly to crashes occurring under heavy load, yet my box (AMD Athlon 64, single-processor) was crashing under practically no load at all. With the no-preemption hack, the box has been quite stable, by the way. Also, my other machine, an AMD Athlon (32-bit), single-processor, running an i386 build, never experienced any crashes, yet it bears much more load most of the time than the 64-bit box, as I use it as my gateway, cvsup daemon, NFS server, etc, etc. Just makes me question how much of a factor load really played in this problem, and thought it might be worth mentioning. -- Conrad J. Sabatier -- "In Unix veritas"