xfs
[Top] [All Lists]

Re: data corruption after hard halt

To: linux-xfs@xxxxxxxxxxx
Subject: Re: data corruption after hard halt
From: Nicolas Kowalski <Nicolas.Kowalski@xxxxxxx>
Date: Wed, 01 Oct 2003 22:54:09 +0200
In-reply-to: <20031001200039.GB6236@xxxxxxxxxxxxxxxxxxxx> (Iustin Pop's message of "Wed, 1 Oct 2003 23:00:39 +0300")
References: <vqoisn9uywd.fsf@xxxxxxxxxxxxxxx> <20031001200039.GB6236@xxxxxxxxxxxxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Gnus/5.1002 (Gnus v5.10.2) XEmacs/21.4 (Common Lisp, linux)
Iustin Pop <iusty@xxxxxxxxx> writes:

>> some files that were
>> modified a few seconds before the outage on the NFS clients have been
>            ^^^^^^^^^^^^^^^^^^^^
> As far as I know, you can't protect yourself from this - at least not
> with XFS. It journals meta-data, not data. Maybe ext3's data journalling
> can help you, though.

This is exactly what I was thinking about. Hm. In normal state , XFS
is the best filesystem I know (speed, reliability). In the case of
hardware/power failures , this is another story; however, these are
very very rare events, perhaps one or two max per year, and they do
not cause so much damage. So, XFS is still the best choice for us, I
hope.

By the way, backups are always goods...This saved a lot of my users
work.


>> What can I do to prevent this ? 
> Except for doing sync (or fsync, from programs), I don't know another
> software solution. A hardware one is an UPS, coupled to the server,
> which signals the server to shutdown BEFORE losing power (battery).

Actually, this is our UPS which failed...

Thanks for your reply.

-- 
Nicolas


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