[Wylug-discuss] Help needed with 'failed' Linux software RAID 10

[John Hodrien]

On Mon, 21 Jun 2010, Christopher Brown wrote:

> Yes - some say it causes additional stress on the drive, causing it
> fail earlier than it normally would.
> I prefer a shorter but more predictable life to a longer less
> predictable one - for my arrays that is. :)

Ah, but that's only if you're running the offline tests.  If you're not
running regular offline tests then you're putting absolutely no extra load on
the disks.

Logging of things like soft read errors is often a very good indicator of
likely disk failure.  This is doubly relevent given SATA's unfortunate
behaviour of not remapping sectors on repeated soft read failures, unlike
SCSI.

Run smartctl -a /dev/sda

That does absolutely nothing to the disk hardware, merely interrogates the
SMART data already collected.  There really is no downside that I'm aware of.

Running full scans on disks you can run as often or as rarely as you like.
Certainly hardware RAID controllers seem to run them monthly or so.  If you're
only running them that often, I really can't believe it's going to shorten the
life of your disks measurably.

jh