Failed disk in Raid

[Ed Wilts]

On Mon, Dec 27, 2004 at 08:06:07AM -0500, Michael J. Pawlowsky wrote:
> On one my home machines I run Software RAID 0 ever since I had set it up 
> for a client of mine.
> I figured it is good insurance.

RAID 0 is striping - it is not good insurance and in fact doubles your
risk of losing data since if either drive dies, you lose data.  RAID 1 
is software mirroring.

> Well it looks like last night it paid off.
> I received a disk failed event.
> My primary / partition has a block that is not writeable on /dev/hda
> 
> But since this is my home machine, and my $ are a bit more precious to me, 
> I was thinking of removing all the partitions from the raid and trying 
> to fix it using chkdisk and then adding it back into the raid.
> 
> What do you think?

Bad idea.  Assuming you meant mirroring and not striping, you do have
your partitioned data on a mirror member but remember that the MBR is
not automatically mirrored.  You could try running the badblock utility
on the failing drive and try to map the bad block out, but in my
experience (going back about 25 years), once a drive starts to fail, it
is on its way down and you get a another day or month out of it, but you
will lose the drive - it's just a matter of when.

Bottom line:
1.  Check to make sure you're mirrored and not striped.
2.  Replace the drive.

-- 
Ed Wilts, RHCE
Mounds View, MN, USA
mailto:ewilts at ewilts.org
Member #1, Red Hat Community Ambassador Program