smartd: 3 Offline uncorrectable sectors

[Paul Howarth]

Bruno Wolff III wrote:
> On Sun, Aug 27, 2006 at 09:54:08 -0400,
>   Kanwar Ranbir Sandhu <m3freak at rogers.com> wrote:
>> On Sat, 2006-26-08 at 14:01 -0700, Wolfgang S. Rupprecht wrote:
>>>     Aug 26 13:16:15 arbol smartd[2284]: Device: /dev/sdb, 3 Currently unreadable (pending) sectors 
>>>     Aug 26 13:16:15 arbol smartd[2284]: Device: /dev/sdb, 3 Offline uncorrectable sectors 
>>>
>>> I'm at a loss as to how to proceed here.  (Or is smartd reporting a
>>> problem that occurred in antiquity and I need to somehow acknowledge
>>> this to the disk and let it clear an error flag?)
>> I've been getting a similar error for a few months:
>>
>> smartd[2602]: Device: /dev/hda, 1 Offline uncorrectable sectors
>>
>> Nothing else appears to be wrong with the hard drive.  Googling for the
>> error message didn't help me to understand what it means.  I've just
>> been hoping it doesn't turn into something worse.
>>
>> I know that doesn't help, but I thought I would let you know you're not
>> alone.
> 
> The disk isn't going to do anything with the bad sectors until you either
> get a good read or write over them. If you do that then the disk may either
> not doing anything further or it may remap the sector to a spare depending
> on whether or not it thinks the problem is permanent.
> 
> If you are running a raid 1 mirror you can copy the sectors from the mirror
> using dd (you probably want to use direct flag). To be a bit safer fail
> out the bad drive (this should be doable under any raid option but 0), write
> over the blocks to clear the errors (this can take a couple of tries if the
> problem is permanent but the drive thinks it might be permanent) and then add
> it back into the raid set.

What is the purpose of writing over the blocks here? Doesn't the 
addition of the drive back into the set cause it to get re-written anyway?

Paul.