From owner-freebsd-stable@FreeBSD.ORG Thu Aug 5 06:20:02 2010 Return-Path: Delivered-To: freebsd-stable@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id 0754A1065676 for ; Thu, 5 Aug 2010 06:20:02 +0000 (UTC) (envelope-from jdc@koitsu.dyndns.org) Received: from qmta08.westchester.pa.mail.comcast.net (qmta08.westchester.pa.mail.comcast.net [76.96.62.80]) by mx1.freebsd.org (Postfix) with ESMTP id A2DB28FC15 for ; Thu, 5 Aug 2010 06:20:01 +0000 (UTC) Received: from omta24.westchester.pa.mail.comcast.net ([76.96.62.76]) by qmta08.westchester.pa.mail.comcast.net with comcast id qWK11e0031ei1Bg58WL1fB; Thu, 05 Aug 2010 06:20:01 +0000 Received: from koitsu.dyndns.org ([98.248.41.155]) by omta24.westchester.pa.mail.comcast.net with comcast id qWL01e0013LrwQ23kWL1BB; Thu, 05 Aug 2010 06:20:01 +0000 Received: by icarus.home.lan (Postfix, from userid 1000) id 829639B423; Wed, 4 Aug 2010 23:19:59 -0700 (PDT) Date: Wed, 4 Aug 2010 23:19:59 -0700 From: Jeremy Chadwick To: "Alex V. Petrov" Message-ID: <20100805061959.GA92106@icarus.home.lan> References: <201008042008.25679.alexvpetrov@gmail.com> <201008051247.43260.alexvpetrov@gmail.com> <20100805053504.GA91196@icarus.home.lan> <201008051409.57751.alexvpetrov@gmail.com> MIME-Version: 1.0 Content-Type: text/plain; charset=utf-8 Content-Disposition: inline Content-Transfer-Encoding: 8bit In-Reply-To: <201008051409.57751.alexvpetrov@gmail.com> User-Agent: Mutt/1.5.20 (2009-06-14) Cc: freebsd-stable@freebsd.org Subject: Re: zpool - low speed write X-BeenThere: freebsd-stable@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Production branch of FreeBSD source code List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Thu, 05 Aug 2010 06:20:02 -0000 On Thu, Aug 05, 2010 at 02:09:57PM +0800, Alex V. Petrov wrote: > В сообщении от 5 августа 2010 13:35:04 вы написали: > > Write performance here is abysmal, agreed. This is very odd. > > > > I hate to say this, but can you remove ahci.ko (ahci_load="yes") from > > your loader.conf and reboot? You may need to change filesystem names > > around in /etc/fstab for your OS disk (assuming it's on ada0), but for > > ZFS it should just magically find the disks on adXX. > > > > If you could also provide pciconf -lvc output that would be helpful. > > Thanks. > > dd if=/dev/zero of=/tank/test.zero bs=3M count=1000 > 1000+0 records in > 1000+0 records out > 3145728000 bytes transferred in 485.431690 secs (6480269 bytes/sec) Can you please remove use of the zpool entirely (e.g. zpool destroy tank) and do a write test to each disk itself? E.g.: dd if=/dev/zero of=/dev/ad8 bs=64k count=1000000 dd if=/dev/zero of=/dev/ad10 bs=64k count=1000000 dd if=/dev/zero of=/dev/ad12 bs=64k count=1000000 I don't recommend using large block sizes (e.g. bs=1M, bs=3M). If all of the above dds show good/decent throughput, then there's something strange going on with ZFS. If this is the case, I would recommend filing a PR and posting to freebsd-fs about the problem, pointing folks to this thread. If all of the dds show bad throughput, then could you please do the following: - Provide vmstat -i output - Install ports/sysutils/smartmontools and run smartctl -a /dev/ad8, smartctl -a /dev/ad10, and smartctl -a /dev/ad12 If only one of the dds shows bad throughput, then please: - Install ports/sysutils/smartmontools and run smartctl -a /dev/XXX, where XXX is the disk which has bad throughput - Try making a ZFS pool with all 3 disks, but then do "zpool offline tank XXX" and then re-attempt the following dd: dd if=/dev/zero of=/tank/test.zero bs=64k count=1000000 And see what throughput looks like. Thanks. -- | Jeremy Chadwick jdc@parodius.com | | Parodius Networking http://www.parodius.com/ | | UNIX Systems Administrator Mountain View, CA, USA | | Making life hard for others since 1977. PGP: 4BD6C0CB |