[Slugnet] Raid 5 Recovery

[desire at gmail.com]

Resend...

On 10/11/07, desire at gmail.com <desire at gmail.com> wrote:
>
> On 10/11/07, P.V.Anthony <pvantony at singnet.com.sg> wrote:
>
> > On this day, 11-October-2007 10:06 AM,  desire at gmail.com wrote:
> > >
> > > If you had a RAID5 array with 3 drives, and one of them crashed, you
> > > should be able to continue using the array in a degraded mode with
> > just
> > > 2 drives.  Once you have a replacement drive, you should be able to
> > > rebuild the array by using your usual software raid management tools
> > > (eg. mdadm).
> >
> > In total there are 15 drives. One drive was marked as faulty. I just
> > re-added it back into the array.
> >
> > Then it was syncing. Then another drive was marked faulty.
> >
> > Now all the drives are marked as spare.
>
>
> If 2 drives have failed, you're pretty much out of luck using standard
> tools.  Even if you could cobble together some specialist software to read
> what is left of the array, you'd have gaps of missing data blocks.
>
> Depending on how the failed drives failed, you may wish to try to
> resuscitate them instead.  If the failed disks are still detectable by the
> controller, you can try using dd_rescue and dd_rhelp on a separate system or
> a livecd to recover as much as possible to spare disks, and using the spare
> disks in place of the failed disks..  You may also have some luck by going
> to a data recovery company.
>
> I think the best practice is to use no more than 6-8 drives in a single
> raid5 array, to control the risk of 2 drives failing before recovery can
> happen.  Sorry to be telling you this after disaster struck.
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://www.lugs.org.sg/pipermail/slugnet/attachments/20071011/82ba489d/attachment.htm