mdadm RAID5 degraded, recreated, partition problems??

tunixman · 01-08-2013, 01:13 PM

Well. It's definitely different geometry. And I think this means that the RAID system has moved the drives around in the RAID, so all of the stripe/parity information will be invalid. You've overridden the original slot information, so you may need to run create again, but with the drives specified in the right order and leaving slot 2 missing. But there's been quite a bit of creative troubleshooting here and we're rapidly leaving what I've done with mdadm, so I may not be able to do much except read man pages to you past this point.

codemastermm · 01-08-2013, 03:14 PM

Yup, that was my thought - having to recreate the array with everything in the exact same position as mdadm had them previously (which is why I was starting to think that the disk slot # was an issue).

I've so far tried creating a RAID5 device purposely degraded (as to leave disk 1 and 2 out and add them in as 4 and 5), but I am receiving a "RUN_ARRAY failed: Input/output error" which isn't too comforting. Here's what I ran:

Code:

sudo mdadm --create --assume-clean /dev/md2 --level=5 --raid-devices=4 missing /dev/sde3 missing /dev/sdg3
mdadm: /dev/sde3 appears to contain an ext2fs file system
    size=728756224K  mtime=Thu Jan  1 00:00:00 1970
mdadm: /dev/sde3 appears to be part of a raid array:
    level=raid5 devices=4 ctime=Tue Jan  8 19:07:39 2013
mdadm: /dev/sdg3 appears to be part of a raid array:
    level=raid5 devices=4 ctime=Tue Jan  8 19:07:39 2013
Continue creating array? yes
mdadm: Defaulting to version 1.2 metadata
mdadm: RUN_ARRAY failed: Input/output error