xfs
[Top] [All Lists]

Re: Repeating fs corruption

To: ViNiL <vladimir.linek@xxxxxxxxxxxxx>
Subject: Re: Repeating fs corruption
From: Martin Steigerwald <Martin@xxxxxxxxxxxx>
Date: Tue, 20 Jun 2006 21:44:42 +0200
Cc: linux-xfs@xxxxxxxxxxx
In-reply-to: <200606201758.20400.vladimir.linek@xxxxxxxxxxxxx>
References: <200606201758.20400.vladimir.linek@xxxxxxxxxxxxx>
Sender: xfs-bounce@xxxxxxxxxxx
User-agent: KMail/1.9.1
Am Dienstag 20 Juni 2006 17:58 schrieb ViNiL:
> Hello!
>
> We run several systems of this configuration, here:
>
> OS: Debian Sarge
> Kernel: vanilla 2.6.16+ (+ Areca driver patch)

[...]

> Can anyone explain to me, what (and why is that :-) is going on with
> the filesystem?

Hello ViNiL,

do you use write caching? With kernel 2.6.16 XFS got corrupted three times 
in one week here with writecache enabled.

I filed a bug report about this, and Nathan Scott recommended to 
deactivate the write cache or to enable the barrier functionality, which 
should be enabled by default in kernel 2.6.17.

http://bugzilla.kernel.org/show_bug.cgi?id=6380

I found no way to query the current status of the write cache.

hdparm -i /dev/yourdevice should show the harddisk manufacturers default, 
for example: WriteCache=enabled

hdparm -W0 disabled write caching when device, driver and controller 
understand the command used to do that.

I had 2.6.16 running for a while with disabled write cache and it seemed 
to work. Currently I run 2.6.15 with disabled write cache again as 2.6.16 
had non working sound after resume from disk (sws2).

Write barriers seems to be available as mount option "barrier". I did not 
yet find any documentation. I did not try the write barrier functionality 
yet. I want to try it with 2.6.17, but I want to wait a little bit 
longer. Maybe in two weeks or so when some more minor releases for 2.6.17 
are out and there a no reports on XFS corruption with 2.6.17 that apply 
to my situation.

Regards,
-- 
Martin 'Helios' Steigerwald - http://www.Lichtvoll.de
GPG: 03B0 0D6C 0040 0710 4AFA  B82F 991B EAAC A599 84C7


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