[Linux-PowerEdge] 2 predicted failure disks and RAID5

Grzegorz Bakalarski bakalarski.g at gmail.com
Tue Nov 14 10:52:21 CST 2017

I have a server (R815) with PERC H710. I have 4 disks and RAID-5 on them.
The server have 2 empty disk slots.

After night I noticed 2 disks got predicted failure state - LEDs on disk
flash yellow (not green) on them (2 of them). MegaCLI shows that 2 disks
have high "Predictive Failure Count:" - some few thousands. Otger 2 disks
are just fine.
For any sake I've put 2 new disks in empty slots and set 2 hot spares.
However I wonder if I could do more?
Is there any safest way to secure data?

I consider:
 A) reconfiguring raid-5 on 4 disks into raid-6 on 5 disks (and then
replacing pre-failing disks one-by-one) - how with MegaCLI - is it good way?
 B) somehow (how???) copying data from one pre-failing disk onto hot spare
and in final stage do switch of good for bad (pre-failing. If possible
repeat it with second disk. In case of media failure data could be
recovered from raid parity. No idea if possible?
 C) replacing disk by marking one of pre-failing disk as missing which
should initiate rebuild on hot spare. This could be dangerous. However is
there any way to judge which of pre-failing disks is in better state (so I
could  first replace and rebuild disk which is in worse condition)??

Any other suggestions?

The VD is used (via lvm) by many Xen virtual machines so I hope it is
possible to do it online.
 (I know: do/check backups, request long [for 1.5TB] maintenance window,
build new raid from new disks and recover from backups - lots of time and
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.us.dell.com/pipermail/linux-poweredge/attachments/20171114/6150b40a/attachment-0001.html>

More information about the Linux-PowerEdge mailing list