xfs
[Top] [All Lists]

Re: mkfs.xfs options with Linux RAID

To: AndyLiebman@xxxxxxx
Subject: Re: mkfs.xfs options with Linux RAID
From: Chris Wedgwood <cw@xxxxxxxx>
Date: Mon, 1 Mar 2004 06:03:53 -0800
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <1ed.1a502952.2d749761@xxxxxxx>
References: <1ed.1a502952.2d749761@xxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
On Mon, Mar 01, 2004 at 08:40:49AM -0500, AndyLiebman@xxxxxxx wrote:

> I have read in a couple of places that, in addition to setting the
> block size with the -b size=XXXX flag, when using xfs under Linux
> along with RAID, it's a good idea to set an xfs option with the -d
> flag (is it the suint, the swath, or something like that?) so that
> it's equal to the block size. Can somebody explain this?  Which is
> the correct parameter to use?  Why should these things be equal? And
> what happens when they're not equal?

For software raid (md device) mkfs.xfs will usually detect this and
set these options correctly for you.  For hardware you probably want
to set swidth to be the hardware stripe chunk size.

I assume you would want to do this to spread the superblocks out
evenly across all your spindles (I can't really think of any other
good reason).



<Prev in Thread] Current Thread [Next in Thread>