Re: Confirmed premature hard disk failure

["David Nielsen" <gnomeuser@xxxxxxxxx>]

2008/12/31 Chuck Anderson <cra@xxxxxxx>

On Tue, Dec 30, 2008 at 07:21:26PM -0600, Callum Lerwick wrote:
>   5 Reallocated_Sector_Ct 0x0033 197 197 140 Pre-fail Always - 19

> 197 Current_Pending_Sector 0x0012 192 190 000 Old_age Always - 86

> off until now. 6 and 7 are from when I first noticed
> Current_Pending_Sector was nonzero, maybe a month or two ago. 8 is from
> my post in July. 9 and 10 are from when the drive was new.

> 1) The SMART health status is PASSED. Just like every other failed drive
> I've had...

Don't look at that alone.

> 2) None of the attributes are failed.

I consider > 0 on Current_Pending_Sector or Reallocated_Sector_Ct to
be failing.

> 3) By the time the self tests start failing, it is too late, you are
> already losing data.

If Current_Pending_Sector or Reallocated_Sector_Ct increases > 0,
replace the drive immediately.

> We can not trust the drive to assess itself properly. If we ever get a
> SMART desktop alert thing going, we MUST look directly at the
> attributes, and come to our own conclusion.

Yes, we should consider any drive to be failed if it gets any pending
or reallocated sectors.

> Any questions before I RMA the drive?

Why did you wait until you had many pending/reallocated sectors before
RMAing it instead of RMAing it as soon as you had 1 bad sector?

I wonder if this new wonder tool davidz is working on Devicekit-disks will be able to give the user some more understandable warnings in this area. I e.g. have in the past 1½ years lost 3 drives, probably all problems of which I could have gotten early warnings and taken action instead of weeping over spilled data.

- David

-- 
fedora-devel-list mailing list
fedora-devel-list@xxxxxxxxxx
https://www.redhat.com/mailman/listinfo/fedora-devel-list