From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S1759430AbZAMNO2 (ORCPT ); Tue, 13 Jan 2009 08:14:28 -0500 Received: (majordomo@vger.kernel.org) by vger.kernel.org id S1755340AbZAMNOT (ORCPT ); Tue, 13 Jan 2009 08:14:19 -0500 Received: from atrey.karlin.mff.cuni.cz ([195.113.26.193]:51879 "EHLO atrey.karlin.mff.cuni.cz" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1754822AbZAMNOT (ORCPT ); Tue, 13 Jan 2009 08:14:19 -0500 Date: Tue, 13 Jan 2009 14:14:18 +0100 From: Pavel Machek To: kernel list , jack@suse.cz, tytso@mit.edu Subject: ext2 + -osync: not as easy as it seems Message-ID: <20090113131418.GD30352@atrey.karlin.mff.cuni.cz> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline User-Agent: Mutt/1.5.13 (2006-08-11) Sender: linux-kernel-owner@vger.kernel.org List-ID: X-Mailing-List: linux-kernel@vger.kernel.org Hi! I tried to use ext2+ -osync, and was surprised that my data do not hit the (SATA) disk. Fortunately I soon realized that I need hdparm -W0 /dev/sda. Could the ext2 be teached to use barriers? -W0 hurts performance more than neccessary... Plus it has a nasty behaviour where it reverts to -W1 if disk connection is momentarily lost. (If you unplug/replug SATA disk, linux will happily rediscover and use it, but -W0 was already forgotten at that point, right?) Pavel -- (english) http://www.livejournal.com/~pavelmachek (cesky, pictures) http://atrey.karlin.mff.cuni.cz/~pavel/picture/horses/blog.html