On Wed, 4 May 2005, Stephen Borrill wrote:
> On Wed, 4 May 2005, Charles Swiger wrote:
>> On May 4, 2005, at 11:20 AM, Stephen Borrill wrote:
>>> Got a RAID 1 set of identical Maxtor 6Y080M0 80GB S-ATA drives. One seems 
>>> to have developed a bad sector or two:
>>> 
>>> wd2a: error reading fsbn 845648 of 845648-845663 (wd2 bn 847664; cn 840 tn 
>>> 14 sn 62)wd2: (uncorrectable data error)
>>> [ ... ]
>>> Is there any way to mark bad sectors in the underlying components so that 
>>> RAIDframe will ignore them? Is doing such a thing a sensible move? 
>>> bad144/badsect don't seem appropriate.
>> 
>> No, it's not a sensible move.  Modern ATA drives already use ECC and 
>> migrate bad sectors to the spare sectors automaticly.  You don't see errors 
>> until the drive has had so many bad sectors appear that it has used up all 
>> of the replacement spare sectors.
>
> I thought that might be the case.
>
>> You should replace this drive ASAP, because it may well fail completely, 
>> and soon...
>
> It's odd, I've had 5 of these same drives fail in the last few months (with 
> the same mobo model). It got me almost thinking it was a bug in the driver 
> for the chipset.

Just had another 2 of these drives fail in the same way at a different 
site, thus killing the RAID array completely (what is it with Maxtor 
6Y080M0 drives?). With a regular FFS partition, some data recovery is 
possible, but it appears that as soon as bad sector is encountered with 
RAIDframe, the whole component is failed. While this is generally a good 
thing, if 2 drives die at roughly the same time (with just a couple of bad 
sectors), it means data recovery is not easy. Any suggestions how to 
recover anything from this?

-- 
Stephen