#25 Crashed RAID5

Version_2.0
open
Kevin Corry
Plugins (6)
5
2006-06-16
2006-06-16
cahlfors
No

My file server crashed (system freeze) the other day.
When restarting, evms logged that three out of four
disks in the set had their superblocks corrupted.

These are the messages:
MDRaid5RegMgr: Region md/md0: MD superblocks found in
objects [hdc1 hdd1 sde1] are not valid. [hdc1 hdd1
sde1] will not be activated and should be removed from
the region.
MDRaid5RegMgr: RAID5 region md/md0 is corrupt. The
number of raid disks for a full functional array is 4.
The number of active disks is 1.

This is really sad reading. Can anything be done, or is
all lost?

I am running Mandriva 2006.0 with rpm packaged evms
version 2.5.1. and kernel 2.6.13. The array was set up
with evms using compatibility mode. Mdadm and
/proc/mdstat does not seem to recognize that there is
an md0 array configured, but other arrays are there. In
evmsgui I can see that the superblocks were indeed
modified on the crash day on all three drives, but I
was not home. All known system activity during the
crash was a running tape backup from the RAID5 array.

Any help is *greatly* appreciated!

Best regards, /Chris

Discussion

  • Mike Tran
    Mike Tran
    2006-06-17

    Logged In: YES
    user_id=209052

    Please run evmsn -d debug, quit, then send
    /var/log/evms-engine.log to mhtran@us.ibm.com

    --
    Regards,
    Mike T.