Bad sector

[jdow]

From: "Bruno Wolff III" <bruno at wolff.to>

> On Mon, Jun 19, 2006 at 16:25:09 -0400,
>  Justin Piszcz <jpiszcz at lucidpixels.com> wrote:
>> I had a bad sector and used the methods below, 2 days later the drive 
>> croaked!  RMA'd it today.  Bad sector? RMA it!
> 
> Bad sectors are expected on consumer drives and you are unlikely to be
> able to RMA a drive because of one bad sector. Modern drives are designed
> to be able to have some bad sectors.

Corrected bad sectors leave the drive technically good. If the drive
is growing bad sectors it is a good idea to take that as a warning
and make sure you back it up and have a spare drive. I have one
cold spare for every RAID array. I don't place critical data of
any sort on single disks. I do backup the single disks for such
"important" data as might exist on the drives. I keep some other
spare drives around for their raw utility value.

A drive that is growing bad blocks, even corrected bad blocks, rapidly
will fail soon. Keep it exercised. Keep it backed up to another drive,
perhaps ad-hoc mirrored. Then when it does die you send it back to the
manufacturer. (Note, Maxtor HAS replaced a pair of drives for me that
had developed errors which tested out on the PowerMax scan even though
a subsequent reformat cleared the errors. As it turns out the drives
were not at fault (!). Something in a Promise RAID controller was bad
and created the bad blocks somehow - at the same absolute disk block
on both drives. I shamefacedly admit it took me too long to figure out
that was just way too big a coincidence even for a manufacturing
defect. But when your data is having on the line in a three disk RAID 5
arrangement with one disk claiming it's failed rationality starts to
depart quickly. <sigh>)

{^_^}