From owner-freebsd-current@FreeBSD.ORG  Tue Mar  9 11:39:52 2004
Return-Path: <owner-freebsd-current@FreeBSD.ORG>
Delivered-To: freebsd-current@freebsd.org
Received: from mx1.FreeBSD.org (mx1.freebsd.org [216.136.204.125])
	by hub.freebsd.org (Postfix) with ESMTP id 7DA8B16A4CE
	for <current@freebsd.org>; Tue,  9 Mar 2004 11:39:52 -0800 (PST)
Received: from critter.freebsd.dk (critter.freebsd.dk [212.242.86.163])
	by mx1.FreeBSD.org (Postfix) with ESMTP id 97B5143D41
	for <current@freebsd.org>; Tue,  9 Mar 2004 11:39:51 -0800 (PST)
	(envelope-from phk@phk.freebsd.dk)
Received: from critter.freebsd.dk (localhost [127.0.0.1])
	by critter.freebsd.dk (8.12.11/8.12.11) with ESMTP id i29JdniS034955
	for <current@freebsd.org>; Tue, 9 Mar 2004 20:39:49 +0100 (CET)
	(envelope-from phk@phk.freebsd.dk)
To: current@freebsd.org
From: Poul-Henning Kamp <phk@phk.freebsd.dk>
Date: Tue, 09 Mar 2004 20:39:49 +0100
Message-ID: <34954.1078861189@critter.freebsd.dk>
Subject: MD(4)/vnode hangs resolved (?)
X-BeenThere: freebsd-current@freebsd.org
X-Mailman-Version: 2.1.1
Precedence: list
List-Id: Discussions about the use of FreeBSD-current
	<freebsd-current.freebsd.org>
List-Unsubscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>,
	<mailto:freebsd-current-request@freebsd.org?subject=unsubscribe>
List-Archive: <http://lists.freebsd.org/pipermail/freebsd-current>
List-Post: <mailto:freebsd-current@freebsd.org>
List-Help: <mailto:freebsd-current-request@freebsd.org?subject=help>
List-Subscribe: <http://lists.freebsd.org/mailman/listinfo/freebsd-current>,
	<mailto:freebsd-current-request@freebsd.org?subject=subscribe>
X-List-Received-Date: Tue, 09 Mar 2004 19:39:52 -0000


Those of you who have experienced that md(4) backed by vnodes hang your
systems, please try this patch:

	http://phk.freebsd.dk/patch/md.patch

It changes so that vnode backed md(4) devices use IO_SYNC by default
(can be overridden with "async" option to mdconfig(8)).

This solves the case where the system get stuck in "wdrain" because
a md(4)/vnode device is relatively large compared to the buffer cache.

Another benefit is that this makes the md(4)/vnode devices behave like
a proper disk with no write-caching.

Performance sucks though.

But slow performance is still better than hung systems.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk@FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe
Never attribute to malice what can adequately be explained by incompetence.