Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1424322AbWKJBKn (ORCPT ); Thu, 9 Nov 2006 20:10:43 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1424326AbWKJBKn (ORCPT ); Thu, 9 Nov 2006 20:10:43 -0500 Received: from omx2-ext.sgi.com ([192.48.171.19]:57314 "EHLO omx2.sgi.com") by vger.kernel.org with ESMTP id S1424322AbWKJBKn (ORCPT ); Thu, 9 Nov 2006 20:10:43 -0500 Date: Fri, 10 Nov 2006 12:10:18 +1100 From: David Chinner To: Russell Cattelan Cc: "Igor A. Valcov" , linux-kernel , xfs@oss.sgi.com Subject: Re: XFS filesystem performance drop in kernels 2.6.16+ Message-ID: <20061110011018.GP8394166@melbourne.sgi.com> References: <1163095715.5632.102.camel@xenon.msp.redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1163095715.5632.102.camel@xenon.msp.redhat.com> User-Agent: Mutt/1.4.2.1i Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org Content-Length: 1640 Lines: 41 On Thu, Nov 09, 2006 at 12:08:35PM -0600, Russell Cattelan wrote: > On Thu, 2006-11-09 at 20:30 +0300, Igor A. Valcov wrote: > > Hello, > > > > For one of our projects we have a test program that measures file > > system performance by writing up to 1000 files simultaneously. After > > installing kernel v2.6.16 we noticed that XFS performance dropped by a > > factor of 5 (tests that took around 4 minutes on kernel 2.6.15 now > > take around 20 minutes to complete). We then checked all kernels > > starting from 2.6.16 up to 2.6.19-rc5 with the same unpleasant result. > > The funny thing about all this is that we chose XFS for that > > particular project specifically because it was about 5 times faster > > with the tests than the other file systems. Now they all take about > > the same time. > > > > I also noticed that I/O barriers were introduced in v2.6.16 and > > thought they may be the cause, but mounting the file system with > > 'nobarrier' doesn't seem to affect the performance in any way. > > > > Any thoughts on the matter are appreciated. > I would try verifying the problem on a non ide disk just > to confirm the write barrier theory. > > Also file a bug. > http://oss/sgi.com/bugzilla > include test case and hard description if possible. and cc xfs@oss.sgi.com on XFS bug reports ;) Cheers, Dave. -- Dave Chinner Principal Engineer SGI Australian Software Group - To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.kernel.org More majordomo info at http://vger.kernel.org/majordomo-info.html Please read the FAQ at http://www.tux.org/lkml/