Re: Data errors on clean, degraded Raid5. Help Please.

On Friday December 2, erich.newell@gmail.com wrote:

Thank you for the feedback Neil.

Although, your last comment did confuse me a little...run what in
parallel? Should I be running badblocks against the unassembled
components of the raid and then doing something like:
     >fsck -l /badblockfile_sda3.txt /dev/md0
     >fsck -l /badblockfile_sdb3.txt /dev/md0
     >fsck -l /badblockfile_sdc3.txt /dev/md0
     >fsck -l /badblockfile_sdd3.txt /dev/md0

type in haste, repent at leisure....

I was thinking that you wold be using 'badblocks' directly rather then
via fsck, and was suggesting
   badblocks -n /dev/sda3 & badblocks /dev/sdb3 & badblocks /dev/sdd3
   ....

Running 'badblocks' across a raid5 array isn't a good idea, as if
there are any bad block, the devices will be kicked from the array,
and the test will abort.

NeilBrown