From owner-freebsd-amd64@FreeBSD.ORG Thu Apr 27 22:24:09 2006 Return-Path: X-Original-To: freebsd-amd64@freebsd.org Delivered-To: freebsd-amd64@freebsd.org Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125]) by hub.freebsd.org (Postfix) with ESMTP id BD38716A400 for ; Thu, 27 Apr 2006 22:24:09 +0000 (UTC) (envelope-from taob@luxography.ca) Received: from as2.dm.egate.net (shell1.dm.egate.net [216.235.15.210]) by mx1.FreeBSD.org (Postfix) with ESMTP id 693DD43D46 for ; Thu, 27 Apr 2006 22:24:08 +0000 (GMT) (envelope-from taob@luxography.ca) Received: by as2.dm.egate.net (Postfix, from userid 8159) id AFB344B44; Thu, 27 Apr 2006 18:24:05 -0400 (EDT) Received: from localhost (localhost [127.0.0.1]) by as2.dm.egate.net (Postfix) with ESMTP id A1CFE4B3C; Thu, 27 Apr 2006 18:24:05 -0400 (EDT) Date: Thu, 27 Apr 2006 18:24:05 -0400 (EDT) From: Brian Tao X-X-Sender: taob@as2.dm.egate.net To: FreeBSD AMD list In-Reply-To: Message-ID: <20060427181602.F43350-100000@as2.dm.egate.net> MIME-Version: 1.0 Content-Type: TEXT/PLAIN; charset=US-ASCII Cc: Subject: Re: Disk I/O-related panics in 6.0-RELEASE? X-BeenThere: freebsd-amd64@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Porting FreeBSD to the AMD64 platform List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 27 Apr 2006 22:24:09 -0000 On Wed, 5 Apr 2006, Vivek Khera wrote: > > All hardware is not created the same.... Last year I went through 5 > motherboards, two full systems, lots of RAM sticks, yada yada yada, > all to get *one* stable server out of it. Needless to say, I don't > buy that vendor anymore. Just wanted to followup on this thread. I initially tried Kris Kennaway's software-based suggestions (don't use ULE, don't use QUOTAS, don't run bg fsck) before going the hardware route. Those had no effect. I suspected flakey RAM to be the most likely culprit, so I replaced the two sticks of OCZ DIMMs with equivalent ones from Kingston. Although I should not have been anywhere close to the capacity of the power supply (should have been plenty left on both the 5V and 12V rails), I took all but two drives offline, and also pulled out the Promise TX-4 SATA card, just in case. Now, so far so good with both 6.0p4 and 6.1-RC. The machine easily gets through a 48-hour period of continuous make buildworlds and buildkernels, whereas before it would panic during the first or second iteration. I'm fairly sure it is the RAM and not the TX-4 or power draw from the drives, but I'll have to schedule a maintenance window to test that. Thanks all for pointing me in the right direction! -- Brian Tao, Luxography http://www.luxography.ca/ (main) http://blog.luxography.ca/ (blog) "The art of light"