Remapping Media Errors when Rebuilds Fail

[Greg Carlson]

Hello All,

I wanted to get some opinions as this has been a heated topic as of late:
Scenario is a hardware Raid 5 with 3 or more hdds.
Ch 0
ID 0 Online A0-0 10 media errors
ID 1 Online A0-1 no errors
ID 2 Failed A0-2 no errors

One hdd fails and the rebuild fails at the same percentage on multiple hdds. Upon closer inspection one of the hdds that is still in an Online State has Media Errors.

Some techs recommend to run a disk verify off the SCSI controller and remap the bad blocks. Doing so usually finds the bad blocks at the same percentage where the rebuild fails and once remapping is complete the rebuild will complete 100%. Seems to be a fix.

However the question is what is the status of the data on the bad blocks?? If the data was corrupt on the bad blocks are you introducing corruption into the array because you essentially have two failed blocks in the Data Stripe?? If the data is ok you should be ok but how do you know? Any thoughts on this??

I believe that if you are in this situation you are better off replacing the hdd that had the media error and recreating the array.

Cheers,
Greg

[Chris Blumenshine]

So you suggest recreating the array with 2 new drives, and restoring from backup, I assume. If the bad block contains live data that cannot be read, then when you ran a backup, that data has the same chance of being corrupt on the tape. If the bad block contained parity data, which was not backed up before the failure, then you have a chance of being better off than you were if you replace both drives and restore.