[mythtv-users] Hard Drive reliability esp. RAID issues

[Michael Heironimus]

On Mon, Mar 08, 2010 at 03:56:54PM -0500, Ronald Frazier wrote:
> devices. For example. A tivo is actually recording live tv 24x7, so
> it's reading/writing 50+ TB/year. Thus it should be getting bad
> sectors several times per year. Has anyone ever pulled a drive from a

There are errors you don't see. Single-bit errors reading from the
physical disk will generally be corrected (ECC) by the firmware on
modern drives. If a sector gets enough failures the drive firmware will
mark it bad and remap that block to a spare. Current drives come with
some bad sectors already remapped (older drives shipped with a bad block
list on the label, sometimes even hand-written). A drive will continue
to accumulate and remap bad sectors throughout its lifetime.
Manufacturers usually give you statistics on this error rate because
it's the one they can measure, but it's not very useful to most of us
because it's too hard to draw valid conclusions.

Then there are errors you do see. They make it through to the OS as
either an I/O error or corrupted data. That often means that either the
number of bad sectors is so high that the drive is out of spares or that
some spot on the disk has started generating new errors quickly enough
that the firmware didn't have a chance to remap it. However, it can also
mean a bad cable or controller, or any number of other things.

-- 
Michael Heironimus