[Durham] SMART errors

[Oliver Burnett-Hall]

On Mon, 10 Nov 2014 13:21:03 +0000
Richard Mortimer <richm at oldelvet.org.uk> wrote:
> 
> SMART at times isn't that smart! The self-tests only report on errors
> found during the self-test and not based on the general health of the
> disk at other times!
> 
> In particular if the bits of the disks that are involved in the
> selftests aren't faulty then it reports all is fine.

That's what was confusing me. I had assumed that the self-tests would
be picking up the errors that were occurring in normal operation.

> If you suspect that there is trouble on the disk look at the -a output
> and you will see lots of error counters. If any of those have any
> significant number of errors (especially if they are increasing
> rapidly) then that indicates problems. But note that some level of
> errors are to be expected due to cosmic rays, read/write alignment or
> whatever the magic disk fairy decided to do.

There are two attributes where this disk differs from its twin in the
array: Raw_Read_Error_Rate with a raw value of 43, and
Current_Pending_Sector which has a raw value of 14 (this value
corresponds directly to the error messages I've been receiving).

Thanks for the guidance on this, I feel like I understand it a bit more
than before.

- olly