xfs
[Top] [All Lists]

Re: [LNX-XFS] Re: large file problems with 2.4.25-pre4

To: Nathan Scott <nathans@xxxxxxx>
Subject: Re: [LNX-XFS] Re: large file problems with 2.4.25-pre4
From: Kelsey Cummings <kgc@xxxxxxxxx>
Date: Mon, 12 Jan 2004 16:51:50 -0800
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <20040113003039.GB969@frodo>
References: <20040112233619.GF98119@xxxxxxxxx> <20040113003039.GB969@frodo>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.4.1i
On Tue, Jan 13, 2004 at 11:30:39AM +1100, Nathan Scott wrote:
> On Mon, Jan 12, 2004 at 03:36:19PM -0800, Kelsey Cummings wrote:
> > I'm new to using xfs, so please forgive me if I'm missing somehting
> > obvious.  I've heard good things about xfs' performance, and I have to
> > admit I'm quite impressed with what I've seen so far.  However, I've run
> > into some confusing problems.
> > 
> > With all of the XFS kernels I've build (2.4.24-pre1, 2.4.25-pre4) I've had
> > trouble creating large files.  
> > 
> > dd if=/dev/zero of=/mnt/blah bs=1024k count=10000
> > 
> > locks up at app 6gigs with bdflush and kswapd consumming all available CPU.
> > Once this occurs any thing attempting to access the filesystem gets stuck
> > waiting on the kernel.
> 
> Hmm... I'm not seeing this behaviour on my test system -
> using the 10G case you have above it completes for me in
> a few minutes (and a bit more quickly on 2.6 than 2.4).
> 
> What is free space like on your /mnt filesystem & how much
> memory do you have?  I tried both a too-small and a plenty-
> large filesystem, and didn't see lockups on either 2.4 or
> 2.6 in a couple of attempts.

plenty, it does it on the empty 800gig file system.  2 gigs of ram, dual
Xeon, HT enabled.  'defau;t' file system creation args.

Just verified again, this time on small filesystems:

/dev/sdb1            359652544   4675564 354976980   2% /mnt/vol0
/dev/sdc1            359652544    239820 359412724   1% /mnt/vol1

It's a redhat 7.3 box, if perhaps the system libs could be causing the
problems.

> If you consistently see lockups, your best bet is to drop
> into kdb and start with backtraces on the stuck processes.

Aie, I'm afraid I don't have that kind of foo. :)

-- 
Kelsey Cummings - kgc@xxxxxxxxx           sonic.net, inc.
System Administrator                      2260 Apollo Way
707.522.1000 (Voice)                      Santa Rosa, CA 95407
707.547.2199 (Fax)                        http://www.sonic.net/
Fingerprint = D5F9 667F 5D32 7347 0B79  8DB7 2B42 86B6 4E2C 3896


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