xfs
[Top] [All Lists]

Re: Problems with many processes copying directories on XFS

To: Simon Matter <simon.matter@xxxxxxxxxxxxxxxx>, jflandry@xxxxxxxxxx
Subject: Re: Problems with many processes copying directories on XFS
From: Seth Mos <knuffie@xxxxxxxxx>
Date: Fri, 14 Sep 2001 08:56:16 +0200
Cc: linux-xfs <linux-xfs@xxxxxxxxxxx>
In-reply-to: <3BA19DA9.E5B81574@xxxxxxxxxxxxxxxx>
References: <3BA08004.A2FBC956@xxxxxxxxxxxxxxxx> <3BA0A6BA.84D64FE1@xxxxxxxxxxxxxxxx> <20010913165214.A4685@xxxxxxxxxx>
Sender: owner-linux-xfs@xxxxxxxxxxx
At 08:03 14-9-2001 +0200, Simon Matter wrote:
Jean-Francois Landry schrieb:
>
> On Thu, Sep 13, 2001 at 02:29:46PM +0200, Simon Matter wrote:
>
> > Well since nobody seems interested I'm replying myself. I didn't realize
> > that I created a filesystem with more than 1.8 mio files with my 'stress
> > test'.
> <snip some lines>
>
> >Okay something goes wrong here as well but now I want to
> > see the ReiserFS deletion speed. 'rm -rf 0*' should go fast but -
> > surprise - it took more than 30min!
>
> >-Simon
>
> OK, this is kind of off-topic, but that behavior is to be expected if
> your ReiserFS fs (hmm, can I say that?) uses the tea hash. Since you
> mention a 2.2 kernel it's quite possible your fs was created while the
> tea hash was the default. You might want to retry with the r5 hash,
> since tea bogs down after say, about 500k files or so.
>
More OT: dmesg says
Checking ReiserFS transaction log (device ...
Using r5 hash to sort names

It tells me r5 whenever a filesystem gets mounted so it is r5. The point
is that XFS is known to be weak when deleting files but in my test
ReiserFS was even slower.

Weird, better tell them on the reiserfs list what is happening with detail which will help them.

Cheers
--
Seth
Every program has two purposes one for which
it was written and another for which it wasn't
I use the last kind.


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