[BlueOnyx:13736] Re: old raq550 question

[George F. Nemeyer]

On Tue, 24 Sep 2013, Stephanie Sullivan wrote:

> Yes - look how many times you see "pre-fail"... This is a strong warning in
> my experience that a drive failure will be happening. If I were you I'd
> remove the drive raid thinks has failed and put a new one in. Then when the
> rebuild/clone is completed, replace the other one too. Both look very iffy
> to me.

The 'pre-fail' is just the sort condition that the test being performed is
looking for...there will always be lots of those kinds of tests.  The
important part is the numbers in the columns that show if the drive really
in good or bad shape for that particular test.

Not all manufacturers use the same test methods or criteria for every
test, so you have to look at the 'threshold' value and compare that with
what the current value is to make any conclusion.   Usually, the numbers
start big and work *DOWN* toward the threshold.   Anything below the
threshold is a warning.

Some numbers can be huge...like seek and data correction ones.  That's not
always anything to worry about.   Also, some of those may be 'snapshots'
and may go lower or higher over time. i.e. a drive that starts out really
cold may have a higher number than one that's just been powered up after
being in operation and is still warm.

The best way to use the information is to make a periodinc run of
'smartctl' and save the info, then compare that to older runs to spot any
trends.  You can do that with a cron job maybe every month or so.

I generally make a run with a new drive, and store that away as a
'baseline' for tests run later.

=^_^=  Tigerwolf