From owner-freebsd-fs@FreeBSD.ORG Tue May 8 22:48:25 2012 Return-Path: Delivered-To: freebsd-fs@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id B5E031065680 for ; Tue, 8 May 2012 22:48:25 +0000 (UTC) (envelope-from freebsd@grem.de) Received: from mail.grem.de (outcast.grem.de [213.239.217.27]) by mx1.freebsd.org (Postfix) with SMTP id 1486F8FC16 for ; Tue, 8 May 2012 22:48:24 +0000 (UTC) Received: (qmail 98385 invoked by uid 89); 8 May 2012 22:48:23 -0000 Received: from unknown (HELO ?192.168.250.164?) (mg@grem.de@80.137.83.22) by mail.grem.de with ESMTPA; 8 May 2012 22:48:23 -0000 Mime-Version: 1.0 (Apple Message framework v1084) Content-Type: text/plain; charset=us-ascii From: Michael Gmelin In-Reply-To: Date: Wed, 9 May 2012 00:48:22 +0200 Content-Transfer-Encoding: quoted-printable Message-Id: <1CEFF50A-4CD5-4947-8A38-2EEAE3311E67@grem.de> References: <73F8D020-04F3-44B2-97D4-F08E3B253C32@grem.de> <180B72CE-B285-4702-B16D-0714AA07022C@grem.de> <44759017-6FAC-4982-B382-CE17DED83262@grem.de> To: freebsd-fs@freebsd.org X-Mailer: Apple Mail (2.1084) Cc: Subject: Re: ZFS resilvering strangles IO X-BeenThere: freebsd-fs@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: Filesystems List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 08 May 2012 22:48:25 -0000 On May 9, 2012, at 00:42, Bob Friesenhahn wrote: > On Wed, 9 May 2012, Michael Gmelin wrote: >>>>=20 >>>> Setting vfs.zfs.vdev_max_pending=3D"4" in /boot/loader.conf (or = whatever >>>> value you want). The default is 10. >>=20 >> Do you think this will actually make a difference. As far as I >> understand my primary problem is not latency but throughput. Simple >> example is dd if=3D/dev/zero of=3Dfilename bs=3D1m, which gave me = 500kb/s. >> Latency might be an additional problem (or am I mislead and a shorter >> queue would raise the processes chance to get data through?). >=20 > The effect may be observed in real-time on a running system. Latency = and throughput go hand in hand. The 'dd' command is not threaded and is = sequential. It waits for the current I/O to return before it starts the = next one. If the wait is shorter (fewer pending requests in line), then = throughput does increase. System total throughput (which includes the = resilver operations) may not increase but the throughput observed by an = individual waiter may increase. >=20 > The default for vdev_max_pending on Solaris was/is 32. If FreeBSD = uses a default of 10 then reducing from the default may be less = dramatic. >=20 That makes sense. I will run more sophisticated I/O tests next time to get a more complete picture. --=20 Michael > Bob > --=20 > Bob Friesenhahn > bfriesen@simple.dallas.tx.us, = http://www.simplesystems.org/users/bfriesen/ > GraphicsMagick Maintainer, http://www.GraphicsMagick.org/