[Pc_Support] Smart Self-test errors, fatal?

[Jason Boxman]

On Wednesday 29 March 2006 21:35, Phil Barnett wrote:
> On Wednesday 29 March 2006 19:41, Jason Boxman wrote:
> > If an ATA drive is coughing up SMART errors, is it completely fatal?
>
> Generally, smart errors are logged when a sector is found to not read
> correctly, the checksum may or may not have allowed the drive to correct
> the data and the data sector is shifted to a spare sector (remapped beneath
> the covers, so to speak).

Yeah, UNC means uncorrectable error after trying to use ECC according to the 
`smartctl` man page.

> Once this happens, the old sector is locked out and the new sector is used
> in it's place.
>
> If you know why this is happening and it stabilizes, you might not have a
> fatality on your hands.

I think it might be because I kinda dropped the case.  Actually, it fell 
sideways, hit another case, then hit the floor after another ten seconds or 
so.  It's a big heavy tower case with the drives mounted both at the top and 
two in the bottom of six 5.25" bays.  I'm surprised I haven't lost more 
drives yet...  (four aren't in warranty for years now, sigh.)

> I'd run spinrite on it to see how bad the damage is.

Excellent.  I'll look into it.  I think Seagate has a utility that'll scan for 
and remap bad sectors.  I doubt the recovery options are as extensive, but 
it's free.  Times some hours to run, though.  Last time I let it remap bad 
sectors on a different Seagate, which later failed again in a nonviolent way, 
but I decided to RMA instead lest I allow a perpetual heal-cycle to begin.

-- 

Jason Boxman
http://edseek.com/ - Linux and FOSS stuff