SMART errors on my disk

[juanjico]

HI! I?ve activated the SMART test on one of my disk, and now I get this on the log:

....
Sep 18 19:02:42 freenas smartd[87221]: Device: /dev/ada1, 1 Offline uncorrectable sectors
Sep 18 19:32:42 freenas smartd[87221]: Device: /dev/ada1, 1 Currently unreadable (pending) sectors
Sep 18 19:32:42 freenas smartd[87221]: Device: /dev/ada1, 1 Offline uncorrectable sectors
Sep 18 19:32:42 freenas smartd[87221]: Device: /dev/ada1, 1 Currently unreadable (pending) sectors
Sep 18 19:32:42 freenas smartd[87221]: Device: /dev/ada1, 1 Offline uncorrectable sectors
Sep 18 20:02:41 freenas smartd[87221]: Device: /dev/ada1, 1 Currently unreadable (pending) sectors
Sep 18 20:02:41 freenas smartd[87221]: Device: /dev/ada1, 1 Offline uncorrectable sectors
Sep 18 20:02:42 freenas smartd[87221]: Device: /dev/ada1, 1 Currently unreadable (pending) sectors
Sep 18 20:02:42 freenas smartd[87221]: Device: /dev/ada1, 1 Offline uncorrectable sectors
Sep 18 20:32:42 freenas smartd[87221]: Device: /dev/ada1, 1 Currently unreadable (pending) sectors
Sep 18 20:32:42 freenas smartd[87221]: Device: /dev/ada1, 1 Offline uncorrectable sectors
....

It's repeated until the infinity. It seems to be one bad sector. But, I can't find information on how to solve this. The disk is unused so it can be formated o something else.

Any idea ?

 

[paleoN]

The disk is unused so it can be formated o something else.

Do the full disk wipe with zeros. That should cause the drive to swap the bad sector out with a spare.

 

[cgoodwille]

I've got a similar situation, however my disk is part of a 4-disk RAIDZ that has important information on (yes, it's backed up elsewhere). Is there something that I can try to fix it without wiping the disk?
Cheers.

 

[bollar]

I've got a similar situation, however my disk is part of a 4-disk RAIDZ that has important information on (yes, it's backed up elsewhere). Is there something that I can try to fix it without wiping the disk?
Cheers.

IMO, I wouldn't do anything aside from have a spare drive available to replace this one in the event it does fail (and have backups that you know are good). This drive is statistically more likely to fail, but it will likely never fail. A lot of the solutions that you'll find here and elsewhere require you to take the zpool from a known, healthy state into one where the zpool is temporarily degraded. IMO, this isn't a situation you want to be in, unless 1) you understand this is what you want 2) you have a spare drive available 3) you have a backup that you're willing to restore from.

Sometimes, though, those errors drive me a little nuts. So, here's what I do:

1. Ensure there are backups that are known to be good.
2. Replace the failing drive with a new drive.
3. Allow the resilver to complete and ensure that the zpool is healthy again
4. Do diagnostics on the failing drive.
5. If you clear the errors, considering using it as a spare (but be aware that it is still statistically more likely to fail).

 

[fracai]

For what it's worth, after I built my system and moved over all my data, I started seeing a few "unreadable (pending) sectors" in the log. I think it started out as 5. From my past experience with bad sectors and some brief research into the message and ZFS's data recovery, I figured I'd just let it go. Then one day about a week later the pending count was down to 2, then 1. That last one stuck around for a while, but finally it too was cleared.

If a bad sector is found during a write, the data will be written elsewhere and the sector marked as bad. If it's bad during a read, you might lose data, but ZFS should be able to repair it.

At least, that's my understanding. For now, my spares are ready.

 

[cgoodwille]

Thanks - both good common-sense solutions! I shall resist the urge to be overly pro-active for the moment. One thing I'm not clear about - will a ZFS Scrub pick up on disk-level errors like this? Or, is that just looking at file system-level corruptions?

 

[fracai]

My understanding is that a scrub will pick up file errors (really block errors), so it will only detect bad blocks if a block of written data has gone bad. I assume that the list of pending sectors is handled by the drive hardware and FreeNAS is just more verbose in reporting that information than other OSs.

 

[paleoN]

I've got a similar situation, however my disk is part of a 4-disk RAIDZ that has important information on (yes, it's backed up elsewhere).

Oh, look a completely different situation.

Is there something that I can try to fix it without wiping the disk?

Try searching better next time. [post=43395]SMART error (CurrentPendingSector) detected on host[/post].

I assume that the list of pending sectors is handled by the drive hardware

Yes, it is along with sparing out the sectors.

FreeNAS is just more verbose in reporting that information than other OSs.

Actually, smartd is reporting on the SMART attributes.