Re: bad block remapping

On Thursday May 23, kanoa@cfht.hawaii.edu wrote:

I have philosophical question about the RAID software.

I think you are right.
After a read error, the raid system would do well to:
  read a few other random blocks on the drive
    (first/last/middle/superblock) to see if the whole drive is dead
    or not
  Updates a failed-read-count in the superblock - abort if this fails
  regnerate the data from redundant info
  attempt a re-write
  On success assume the device has coped.

All we need now is code .... are you offering?

NeilBrown

Thanks Neil,

No I'm not offering; I don't have the skill to do something like that 
though I would offer emotional support and pizza to anyone who could.

-Kanoa

Neil Brown wrote:

On Thursday May 23, kanoa@cfht.hawaii.edu wrote:

quoted

I have philosophical question about the RAID software.

I think you are right.
After a read error, the raid system would do well to:
 read a few other random blocks on the drive
   (first/last/middle/superblock) to see if the whole drive is dead
   or not
 Updates a failed-read-count in the superblock - abort if this fails
 regnerate the data from redundant info
 attempt a re-write
 On success assume the device has coped.

All we need now is code .... are you offering?

NeilBrown