From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S266422AbUA3Gpw (ORCPT ); Fri, 30 Jan 2004 01:45:52 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S266440AbUA3Gpw (ORCPT ); Fri, 30 Jan 2004 01:45:52 -0500 Received: from fw.osdl.org ([65.172.181.6]:44422 "EHLO mail.osdl.org") by vger.kernel.org with ESMTP id S266422AbUA3Gpv (ORCPT ); Fri, 30 Jan 2004 01:45:51 -0500 Date: Thu, 29 Jan 2004 22:46:29 -0800 From: Andrew Morton To: "Curt" Cc: linux-kernel@vger.kernel.org Subject: Re: Raw devices broken in 2.6.1? AND- 2.6.1 I/O degraded? Message-Id: <20040129224629.702e9eca.akpm@osdl.org> In-Reply-To: <015301c3e6fb$2dbc22b0$0300000a@falcon> References: <01c501c3e6b9$67225f70$0700000a@irrosa> <20040129163852.4028c689.akpm@osdl.org> <020d01c3e6d0$acd78f60$0700000a@irrosa> <20040129205605.5bd140b2.akpm@osdl.org> <015301c3e6fb$2dbc22b0$0300000a@falcon> X-Mailer: Sylpheed version 0.9.4 (GTK+ 1.2.10; i686-pc-linux-gnu) Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: 7bit Sender: linux-kernel-owner@vger.kernel.org X-Mailing-List: linux-kernel@vger.kernel.org "Curt" wrote: > > > Or you can put 2.6 on par by setting > > /proc/sys/vm/dirty_background_ratio to 40 and dirty_ratio to 60. > > Okay will do, is there a good comprehensive resource where I can read up on > these (and presumably many other I/O related) variables? We've been relatively good about keeping the in-kernel documentation up to date. For this stuff, see Documentation/filesystems/proc.txt and Documentation/sysctl/vm.txt. > > Longer-term, if your customers are using scsi, you should ensure that the > > disks do not use a tag queue depth of more than 4 or 8. More than that > and > > the anticipatory scheduler becomes ineffective and you won't get that > > multithreaded-read goodness. > > I've heard-tell of tweaking the elevator paramter to 'deadline', again could > you point me to a resource where I can read up on this? And forgive the > newbie-question, but is this a boot-time parameter, or a bit I can set in > the /proc system, or both? It's boot-time only. We were working on making it per-disk but that was quite complex and we really didn't get there in time. So add `elevator=deadline' to your kernel boot command line. From my (brief) testing, it was a significant lose. It needs more work though: 2.6+deadline shouldn't be slower than 2.4.x > > Please stay in touch, btw. If we cannot get applications such as yours > > working well, we've wasted our time... > > I'll do what I can to provide real-world feedback, I want this to work too. Thanks.