[Top] [All Lists]

Terrible xfs performance on current cvs kernel with sw RAID

To: <linux-xfs@xxxxxxxxxxx>
Subject: Terrible xfs performance on current cvs kernel with sw RAID
From: Jani Jaakkola <jjaakkol@xxxxxxxxxxxxxx>
Date: Wed, 18 Jul 2001 17:19:30 +0300 (EEST)
In-reply-to: <Pine.LNX.4.33.0107171934150.3797-100000@xxxxxxxxxxxxxxxxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
On Tue, 17 Jul 2001, Jani Jaakkola wrote:

> On Tue, 17 Jul 2001, Steve Lord wrote:
> >
> > Could you possibly try the cvs tree, Linus was still working deadlocks
> > out of the memory allocation/reclaim end of things up until 2.4.7-pre2.
> > XFS and ext2 will almost certainly push things in different directions.
> OK I'll try it..
> Right, it has now been running longer than ever before without a lockup.
> However, the performance is very bad. But it just might be caused by the
> simultaneous RAID resync I am doing at the same time. I'll get back to
> this after the resync is done (or the machine has crashed).

OK, no more lockups, but the horrible performance was not caused by RAID
resync. It seems that the current cvs kernel has bad problems with sw
RAID and xfs combined (and maybe SMP too). When untarring archive with
lots of small files or using 'rm -r' I get 5 files created/removed per
second, which is the performance of my last 286 with MSDOS :(

The performance problem goes away if I don't use sw RAID. It also seems
to be journal related, since there is no problems with just
reading/writing big files. Bonnie tells me my block output is 37940K/sec
and input is 133400K/sec with 4GB data.

I must now leave this thing for a few days, but I guess I will try to find
out what is wrong (however, I'm not a competent kernel hacker).

- Jani

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