Unrecoverable sector WD Green (EARS)

[eltoro]

Hi,

I suddenly get the following messages on my FreeBSD/FreeNAS server:

Mar 17 09:29:33 eltoro-storage smartd[1906]: Device: /dev/ada0, 1 Currently unreadable (pending) sectors
Mar 17 09:29:33 eltoro-storage smartd[1906]: Device: /dev/ada0, 1 Offline uncorrectable sectors

smartctl reports Healthy and when I run an extended sector test using Western Digitals diagnostic utility (on another server), it also reports healthy...Can someone explain this to me ? Shouldn´t smartctl on FreeBSD report not heathy when smartd has detected errors?

Also, since I upgraded to FreeNAS 8.2Beta2 yesterday, I get a lot of
Mar 17 09:26:52 eltoro-storage init: getty repeating too quickly on port /dev/ttyv0, sleeping 30 secs

I never saw those on 8.04REL

Thank you in advance.

 

[Thomas Blomberg Hansen]

I got starting yesterday with 18 Currently unreadable ... :
Mar 17 11:39:22 freenas smartd[1294]: Device: /dev/ada3, 17 Currently unreadable (pending) sectors
Mar 17 11:39:22 freenas smartd[1294]: Device: /dev/ada3, 2 Offline uncorrectable sectors

I asked on IRC and got the message that I should start considering a drive change and not to worry that much.
Its a "Your drive is starting to fail, still working but consider getting a replacement."

Advanced RMA was a suggestion.

 

[eltoro]

The problem is that the store will probably run WD Diagnostics which reports no errors at all. I think it is strange that smartd can detect errors so fast and WD Diagnostics does not find anything after 20 hours...

 

[survive]

Hi eltoro,

I would imagine that WD has an incentive to make their testing tool less "sensitive" to errors.

-Will

 

[eltoro]

Hi eltoro,

I would imagine that WD has an incentive to make their testing tool less "sensitive" to errors.

-Will

It appears so. HD Tunes reports C5 errors as well...

Thanks

 

[jgreco]

An unreadable sector is not necessarily an indication that a drive is starting to fail. It can be a symptom of a minor imperfection in a platter. All drives have minor defects. A typical SCSI disk, for example, might have several hundred blocks listed as it comes from the factory, and then also maintains a grown defect list, which are defects that develop later and require the use of additional spare blocks. On a relatively smallish 72GB enterprise drive, I see 438 blocks in the primary defect list and 4 in the grown. I have a 10 year old 36GB drive with 68 blocks in the grown list - almost 50% increase over the 154 in the primary table.

Modern SATA drives are moderately intelligent, and have some similar capability for coping with the occasional defect. The WD tool should be able to spot and correct a bad sector. A single bad sector is probably not cause for RMA and is likely to cause WD to return your drive to you. However, if it's developing lots of bad sectors rapidly, that would be a cause for concern, and a definite sign of problems ahead.