0

After a hard drive died on a RAID5 setup on a legacy system, the Vmware Virtual Machine on it kept on running for a period of 6 months. Needless to say, finally it crashed bad. Now, the dead hard drive has been replaced and RAID5 restoration tools have been used to recover the lost data. Yet, the logical drive is still giving IO errors. Is this normal? Should we be worried about further data loss?

Thanks in advance

Uku Loskit
  • 160
  • 2
  • 10

2 Answers2

2

You aren't telling us enough:

Are you using a hardware RAID controller? Which one? If not, software RAID on what operating system? Did the recovery have finished without reporting any errors? Did the RAID management tools for the RAID solution you are using report the volume as working fine? What kind of I/O errors do you get? Are they reported by the host OS or the guest machine?

Also, I can't help but to remark that you should have started to worry six months ago when the drive died. If a drive fails in a RAID setup you have to replace it immediately, especially with a RAID5. Any error on one of the remaining disks will fail the recovery, so there is no time to loose.

Sven
  • 97,248
  • 13
  • 177
  • 225
  • "I can't help but to remark that you should have started to worry", man, are you a life-teacher? :-) He asks clearly what he asks. – poige Feb 26 '11 at 11:20
  • Also, there're 90 % of such fuzzy questions on the serverfault.com so I wonder how you did gain all your nearly 10k rep. :) – poige Feb 26 '11 at 11:27
0

Yep, you should for sure but that's not 100 % warranty, you know. I'd use Linux Software RAID as a tool to check how many stripes have had bad CRCs. That would give overall damage picture.

poige
  • 9,171
  • 2
  • 24
  • 50
  • For the "some stupid with the flare gun" having had minused my answer I'd comment additionaly. *Yep*, Linux Soft RAID allows you assembling others RAID-types disks providing lots of knobs to specify theirs params. It also allows checking parity with quite a simple commands being echoed to corresponding /sys file. You can then find out how many mismatches it did encountered during its check. – poige Feb 26 '11 at 11:30