On 02/25/2010 02:25 PM, m.roth@xxxxxxxxx wrote:
> Ok, I saw more sectors on a drive yesterday, so this morning, no one was
> running on it, and I took it out of use, then bounced it onto a DVD, and
> ran fsck -c (check for bad blocks). It finished. I bounce the server.
>
> And SMARTD reports the sectors as "currently unreadable (pending)
> sectors", and "offline uncorrectable sectors".
>
> Does smartd cache its info somewhere, or is it reading what fsck already
> marked as bad? And this has happened before, under 5.3, and under
> continuing current updates of 5.4....
smartd queries the drive directly. Running "fsck -c" will have caused the
drive to discover all of the unreadable sectors within that file system and
mark them as "pending reallocation". Those sectors will remain in the
"pending" state and visibly bad to the OS until the next time they are
written. If you were able to copy all of the files from that FS without
error, that suggests that all of the bad sectors are in free space, and
you could stimulate reallocation by filling all of the free space with
zeros:
mount /dev/{whatever} /mnt
dd if=/dev/zero of=/mnt/xxxxxx bs=64k
sync
rm /mnt/xxxxxx
umount /mnt
Then run "smartctl -A" on the drive and see if it still reports pending
sectors.
Of course if that drive is continuing to develop new bad sectors you
should get rid of it immediately.
--
Bob Nichols "NOSPAM" is really part of my email address.
Do NOT delete it.
_______________________________________________
CentOS mailing list
CentOS@xxxxxxxxxx
http://lists.centos.org/mailman/listinfo/centos
