From mboxrd@z Thu Jan 1 00:00:00 1970 From: bugzilla-daemon@bugzilla.kernel.org Subject: [Bug 20902] High IO wait when writing to ext4 Date: Wed, 24 Nov 2010 05:13:22 GMT Message-ID: <201011240513.oAO5DMoG002507@demeter1.kernel.org> References: Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" To: linux-ext4@vger.kernel.org Return-path: Received: from demeter1.kernel.org ([140.211.167.39]:53025 "EHLO demeter1.kernel.org" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751326Ab0KXFNZ (ORCPT ); Wed, 24 Nov 2010 00:13:25 -0500 Received: from demeter1.kernel.org (localhost.localdomain [127.0.0.1]) by demeter1.kernel.org (8.14.4/8.14.3) with ESMTP id oAO5DNC6002508 (version=TLSv1/SSLv3 cipher=DHE-RSA-AES256-SHA bits=256 verify=NO) for ; Wed, 24 Nov 2010 05:13:23 GMT In-Reply-To: Sender: linux-ext4-owner@vger.kernel.org List-ID: https://bugzilla.kernel.org/show_bug.cgi?id=20902 --- Comment #17 from Eric Sandeen 2010-11-24 05:13:16 --- Created an attachment (id=38012) --> (https://bugzilla.kernel.org/attachment.cgi?id=38012) graph of IO Here's a graph of the IO. Does the trace cover the entire test starting at mount? Looks like it. Were the 10x128MB writes in series or parallel? Looks like it spends about 20s seekily sucking up 28MB of metadata off the disk (mostly done in the [flush-253:0] process) and then doing 1.3G of horrible writeback ;) What exact commands did you use for the test; I assume it was buffered IO, all in the same subdir? Sync afterwards or not? It might be interesting to see the tune2fs -l output to correlate the initial reads to disk metadata but I suppose we can have a pretty good guess at that. Anyway looks like the first issue is the giant seeky read of disk metadata before any writing even gets underway. -- Configure bugmail: https://bugzilla.kernel.org/userprefs.cgi?tab=email ------- You are receiving this mail because: ------- You are watching the assignee of the bug.