Read errors on Parity 2

[Marv]

Hi everyone,

 

just got home and my server reported 168 read errors on my second parity drive.

As I have no idea where to look at at the smart report maybe someone can help me out here.

 

Something to worry about?

unraid-diagnostics-20180215-1748.zip

unraid-smart-20180215-1746.zip

[JorgeB]

SMART looks fine but the first error on the syslog looks like a disk error, though since the problem sectors were successfully written back it might be good now, still recommend doing an extended SMART test, or a non correcting parity check.

[Marv]

will do, thanks a lot!

[Marv]

So I ran an extended SMART test but it got aborted because errors occured.

Should I replace the disk?

unraid-smart-20180216-2228.zip

[JorgeB]

Yes, the extended test confirms the error was caused by the disk, despite all attributes being OK, SMART is usually a good indication but likes most things it's not perfect.

[JorgeB]

Let me correct the "all attributes being OK" above, there's one attribute not OK, not usually one of the main ones monitored, but should always be 0 on WD drives:

 

200 Multi_Zone_Error_Rate   ---R--   200   200   000    -    5

And it was still 0 on the first diags, so besides the failed test there's still a clue on SMART, though it should be showing some pending sectors also.

[pwm]

8 minutes ago, johnnie.black said:

Let me correct the "all attributes being OK" above, there's one attribute not OK, not usually one of the main ones monitored, but should always be 0 on WD drives:

 

200 Multi_Zone_Error_Rate   ---R--   200   200   000    -    5

And it was still 0 on the first diags, so besides the failed test there's still a clue on SMART, though it should be showing some pending sectors also.

Second thread with these issues recently - WD drive that fails extended SMART and shows multi-zone error rate but does not flag any sector as bad.

 

I wonder if the multi zone here means the drive couldn't even correctly locate one or more tracks or calibration meta-data for a zone. So it gave read errors but couldn't pin-point actual sectors to flag as pending. I don't remember having seen the extended test fail without having seen any bad sectors flagged.

[JorgeB]

It's not common but I've seen it a few times before, failing the extended test without pending sectors and either Raw_Read_Rate errors, Multi_Zone, or both, on the other hand it's much more common for WD drives to show "false positives" pending sectors, I remember some time ago one with like 150 pending sectors reported and it passed the extended test, that's why I usually recommend the extend test.