xfs
[Top] [All Lists]

Re: logfile size

To: thomas <thomas@xxxxxxxx>
Subject: Re: logfile size
From: Nathan Scott <nathans@xxxxxxx>
Date: Sun, 18 Nov 2001 16:07:30 +1100
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <9429832176.20011118051715@huno.net>; from thomas@huno.net on Sun, Nov 18, 2001 at 05:17:15AM +0100
References: <9429832176.20011118051715@huno.net>
Sender: owner-linux-xfs@xxxxxxxxxxx
User-agent: Mutt/1.2.5i
hi,

On Sun, Nov 18, 2001 at 05:17:15AM +0100, thomas wrote:
> hi there.
> 
> i just made a 75GB XFS partition using the default values etc. but now i
> read that a larger logfile will result in increased performance. i tried
> to find out the logfile size that mkfs.xfs gave me but the output from

Try the xfs_info command (ie. xfs_growfs -n).  By default, mkfs will
use 1200 blocks, unless you have a very large filesystem.

> +[~]: cat /proc/fs/xfs/stat
> remains a mystery to me.

The log size is not in there - these are log statistics.

> since i can change the logfile size with
> xfs_growfs i'd really like to optimize it as much as possible. this is a

The growfs man page -l option is a bit of wishful thinking - there is
no kernel code behind it as yet.  You will need to re-mkfs with the -l
option, something like: "-lsize=12000b".

> p133/96mb box running xfs 1.01/2.4.13 that mostly serves 10mb+ files
> through samba 2.2.2 and also processes some of them from time to time.
> what is the tradeoff for a larger logfile size? more memory usage? more
> cpu usage?

Less disk space - I'm pretty sure memory use and cpu use will remain
exactly the same.  You can tradeoff memory with the logbufs/logbsize
options.

cheers.

-- 
Nathan


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