xfs
[Top] [All Lists]

Re: Speeding up XFS

To: linux-xfs@xxxxxxxxxxx
Subject: Re: Speeding up XFS
From: Nathan Scott <nathans@xxxxxxx>
Date: Tue, 27 Apr 2004 11:53:40 +1000
Cc: lists-linux-xfs@xxxxxxxxxxxxxxxxxxxxxxxx
In-reply-to: <200404261756.i3QHuhr2327870@xxxxxxxxxxxxxxxxxxxxxxx>; from overby@xxxxxxx on Mon, Apr 26, 2004 at 12:56:43PM -0500
References: <20040426165155.GA19148@xxxxx> <200404261726.i3QHQ8Aj298772@xxxxxxxxxxxxxxxxxxxxxxx> <20040426173300.GA23256@xxxxx> <200404261756.i3QHuhr2327870@xxxxxxxxxxxxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.2.5i
On Mon, Apr 26, 2004 at 12:56:43PM -0500, Glen Overby wrote:
> On April 26, Bruce Guenter wrote:
> > 
> > Is there any way to avoid this situation?
> 
> On Irix: mount -o inode64

And on Linux.

> On Linux: rewrite linux to use 64 bit inode numbers :-)

On a 64 bit platform, 64 bit inode numbers are available.
The inode64 option is enabled on such a system.  IA64 was
broken for some time in this regard, but we fixed that a
long time ago.


If your directories are indeed small, and directory operations
are your bread and butter for this fs, I expect going to the
smallest XFS inode size that fits all the directory entries
inline in the inode (largest XFS inode size is 2k iirc, default
is 256 bytes) to make the most difference.  But, as you can tell
from all the advice, there are lots of options, so fiddle with
all the knobs till you find your ideal configuration. :)

cheers.

-- 
Nathan


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