From owner-freebsd-current@FreeBSD.ORG Tue Jul 3 01:35:34 2007 Return-Path: X-Original-To: freebsd-current@freebsd.org Delivered-To: freebsd-current@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [69.147.83.52]) by hub.freebsd.org (Postfix) with ESMTP id 85A0E16A46B for ; Tue, 3 Jul 2007 01:35:34 +0000 (UTC) (envelope-from n-butcher=freebsd-current=freebsd.org=sbibybnr@fusiongol.com) Received: from smtp02.dentaku.gol.com (smtp02.dentaku.gol.com [203.216.5.72]) by mx1.freebsd.org (Postfix) with ESMTP id 5CF2213C469 for ; Tue, 3 Jul 2007 01:35:34 +0000 (UTC) (envelope-from n-butcher=freebsd-current=freebsd.org=sbibybnr@fusiongol.com) Received: from pat.gol.co.jp ([203.216.1.191] helo=[127.0.0.1]) by smtp02.dentaku.gol.com with esmtpa (Dentaku) id 1I5XIi-0004mc-F9; Tue, 03 Jul 2007 10:35:32 +0900 Message-ID: <4689A7DC.1070604@fusiongol.com> Date: Tue, 03 Jul 2007 10:35:24 +0900 From: Nathan Butcher User-Agent: Thunderbird 2.0.0.4 (Windows/20070604) MIME-Version: 1.0 To: =?UTF-8?B?RGFnLUVybGluZyBTbcO4cmdyYXY=?= References: <4679CE35.8080907@fusiongol.com> <86wsxibxc2.fsf@dwp.des.no> In-Reply-To: <86wsxibxc2.fsf@dwp.des.no> X-Enigmail-Version: 0.95.1 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit X-Abuse-Complaints: abuse@gol.com Cc: freebsd-current@freebsd.org Subject: Re: Promise SATA300 TX4 card issues X-BeenThere: freebsd-current@freebsd.org X-Mailman-Version: 2.1.5 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: Tue, 03 Jul 2007 01:35:34 -0000 Dag-Erling Smørgrav wrote: > Nathan Butcher writes: >> I've been having problems with my ZFS pool consisting of 4 drives hooked >> up to my Prmoise SATA 300TX4 controller. It had been working fine until >> to around the 19th of this month, and then I started seeing error >> messages whenever I tried to write lots of data to ZFS..... followed by >> a nasty lockup with no other debugging information of which to speak of. > > This is a recurring problem with ZFS and ata disks. It appears that the > ata driver doesn't like the ATA_FLUSHCACHE command very much, and ZFS > uses it a lot. Turning off atime helps, as does turning off ata write > caching (hw.ata.wc=0 in loader.conf) - I still get timeouts and > transient hangs once in a while, but my server no longer crashes: Thanks for your input. I had my suspicions about the ata driver. Perhaps I should clarify my situation however. I rebuilt my kernel and world on the 15th, and my server was working fine until the 19th when I rebuilt my kernel and world again and then started seeing issues. It seems to me that some changes must have been made in the ata driver (assuming that's what it is) between updates which caused this borkage to happen. When I rolled back to the 200706 snapshot sources, once again my server is stable. Mind that my server has a core2 duo running amd64 with 2GB of memory. I noticed some updates to the ata driver code during the past month, so I will try updating to the next snapshot sources (200707) when it is released to see if this issue has been fixed.... otherwise I'm rolling back to 200706 sources.