xfs
[Top] [All Lists]

Re: 2.6.11-rc3: 80-95% systime when serving files via NFS

To: Sonny Rao <sonny@xxxxxxxxxxx>
Subject: Re: 2.6.11-rc3: 80-95% systime when serving files via NFS
From: Dave Chinner <dgc@xxxxxxx>
Date: Wed, 9 Feb 2005 07:20:03 +1100
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <20050208165456.GA15252@xxxxxxxxxxxxxxxxxx>; from sonny@xxxxxxxxxxx on Tue, Feb 08, 2005 at 11:54:56AM -0500
References: <222BE5975A4813449559163F8F8CF5030C22D2@xxxxxxxxxxxxxxxxxxxx> <20050208153427.G147390@xxxxxxxxxxxxxxxxx> <20050208165456.GA15252@xxxxxxxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.2.5.1i
On Tue, Feb 08, 2005 at 11:54:56AM -0500, Sonny Rao wrote:
> On Tue, Feb 08, 2005 at 03:34:27PM +1100, Dave Chinner wrote:
> <snip> 
> > Seeing as you are running 2.6.11-rc3, you're in luck. Nathan's
> > inode hash tweaks were merged into -rc3, so you can increase the
> > size of the inode hash with a mount option. The mount option
> > is "ihashsize", and it specifies the width of the hash table.
> > i.e. `mount -o ihashsize=xxxxx /dev/sdb1 /mntpt`
> 
> 
> Hmm, I've seen this before too running something which resembles
> SPEC-SFS where we have millions of inodes in cache.
> 
> I'm guessing that increasing the hash-size will help us here as well?

Depends on how many filesystems you're using. If this is from a small
number of filesystems, then yes. If you are using typical specsfs
benchmark techniques using many tens or hundreds of filesystems,
then it probably wont make any big difference because each filesystem
has it's own hash....

Cheers,

Dave.
-- 
Dave Chinner
R&D Software Engineer
SGI Australian Software Group


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