From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1757157AbYERUEi (ORCPT ); Sun, 18 May 2008 16:04:38 -0400 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1754449AbYERUEZ (ORCPT ); Sun, 18 May 2008 16:04:25 -0400 Received: from smtp-out03.alice-dsl.net ([88.44.63.5]:35556 "EHLO smtp-out03.alice-dsl.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1753830AbYERUEY (ORCPT ); Sun, 18 May 2008 16:04:24 -0400 To: Eric Sandeen Cc: Andrew Morton , linux-ext4@vger.kernel.org, linux-kernel@vger.kernel.org, linux-fsdevel@vger.kernel.org Subject: Re: [PATCH 0/4] (RESEND) ext3[34] barrier changes From: Andi Kleen References: <482DDA56.6000301@redhat.com> <20080516130545.845a3be9.akpm@linux-foundation.org> <482DF44B.50204@redhat.com> <20080516220315.GB15334@shareable.org> <482E08E6.4030507@redhat.com> Date: Sun, 18 May 2008 22:03:55 +0200 In-Reply-To: <482E08E6.4030507@redhat.com> (Eric Sandeen's message of "Fri, 16 May 2008 17:21:26 -0500") Message-ID: <8763tbcrbo.fsf@basil.nowhere.org> User-Agent: Gnus/5.1008 (Gnus v5.10.8) Emacs/21.3 (gnu/linux) MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii X-OriginalArrivalTime: 18 May 2008 19:56:59.0121 (UTC) FILETIME=[55006E10:01C8B921] Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Eric Sandeen writes: > > Right, that was the plan. I wasn't really going to stand there and pull > the plug. :) I'd like to get to "out of $NUMBER power-loss events > under this usage, I saw $THIS corruption $THISMANY times ..." I'm not sure how good such exact numbers would do. Surely power down behaviour that would depend on the exact disk/controller/system combination? Some might be better at getting data out at power less, some might be worse. To get a good picture, you would probably need to do such tests over a wide range of systems and put all the data together, but even then you couldn't be sure you covered all well. For a distributor it would probably make more sense to do such tests as part of system certification and then modify the defaults to match all tested systems. But that doesn't really work for good mainline defaults. I suspect for mainline the only good default is to be very conservative. -Andi