Received:  by oss.sgi.com id <S554034AbRBOT0U>;
	Thu, 15 Feb 2001 11:26:20 -0800
Received: from tux.mkp.net ([130.225.60.11]:17674 "EHLO tux.mkp.net")
	by oss.sgi.com with ESMTP id <S554031AbRBOTZu>;
	Thu, 15 Feb 2001 11:25:50 -0800
Received: from tux.mkp.net ([130.225.60.11] helo=jaguar.mkp.net)
	by tux.mkp.net with esmtp (Exim 3.16 #1)
	id 14TU1v-0007Em-00; Thu, 15 Feb 2001 20:25:48 +0100
Received: (from mkp@localhost)
	by jaguar.mkp.net (8.9.3/8.9.3) id JAA01049;
	Thu, 15 Feb 2001 09:24:39 -0500
X-Authentication-Warning: jaguar.mkp.net: mkp set sender to mkp@mkp.net using -f
To:     james rich <james.rich@m.cc.utah.edu>
Cc:     linux-xfs@oss.sgi.com
Subject: Re: build failure update and raid 0
References: <Pine.GSO.4.05.10102142249390.16516-100000@pipt.oz.cc.utah.edu>
From:   "Martin K. Petersen" <mkp@linuxcare.com>
Organization: Linuxcare, Inc.
Date:   15 Feb 2001 09:24:39 -0500
In-Reply-To: <Pine.GSO.4.05.10102142249390.16516-100000@pipt.oz.cc.utah.edu>
Message-ID: <yq18zn8aus8.fsf@jaguar.mkp.net>
Lines:  29
User-Agent: Gnus/5.0808 (Gnus v5.8.8) XEmacs/21.1 (Canyonlands)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Sender: owner-linux-xfs@oss.sgi.com
Precedence: bulk
Return-Path: <owner-linux-xfs@oss.sgi.com>
X-Orcpt: rfc822;linux-xfs-outgoing

>>>>> "james" == james rich <james.rich@m.cc.utah.edu> writes:

james> I also thought I should mention that I am running xfs on a raid
james> level 0 device (using the md driver).  The XFS FAQ mentions
james> that xfs hasn't been tested with raid and that people using it
james> should let folks know how it works. 

XFS works fine on MD RAID0.  Some people have reported funnies with
RAID1, and that's on my list of things to look into.  We have fixes
for RAID5, but performance suffers.  Especially during resync.  Fixing
that will require major surgery to the stripe cache.


james> The FAQ also says that xfs will not work with the md driver and
james> the kiobuf option enabled.  Now I don't remember if the kiobuf
james> option is enabled, but xfs is working beautifully on md0.  

You enable kiobuf I/O by giving -o kio as mount option.  I.e.

        mount -o kio /dev/sdb2 /foo

That will currently fail miserably with MD and LVM, though.  I'm very
close to having kiobuf based LVM ready to go into the tree.  MD RAID0
is next on my list.

-- 
Martin K. Petersen, Principal Linux Consultant, Linuxcare, Inc.
mkp@linuxcare.com, http://www.linuxcare.com/
SGI XFS for Linux Developer, http://oss.sgi.com/projects/xfs/