The first thing you want to do with a disc that has bad blocks in a mirror
is to throw it out.
If a HDD developed bad blocks at run time when on a h/w RAID mirror then if
the h/w RAID is good (IE not a on motherboard cheapo SATA RAID) it is likely
to have the ability to reallocate the defective block at run-time - this is
a component of fault tolerance. If your RAID controller & its driver can't
do that (EG it is a low cost one) then it will likely fail the disc drive
immediately and drop it off line - this is when you should replace the HDD.
If a HDD had defective blocks remapped at run-time I would also consider
tossing it out. Depends on the quality of the controller and the "quality"
of the HDD. If the drive vendor will replace the drive under warrantee, then
it is stuffed, try the vendors diagnostics - most have them for download. If
its out of warrantee, then its old, needs replacement and will likely only
give you more trouble than its worth.
If you have hot standby discs and the rate of defects on the drive is other
than very low then a controller may not toss the disc offline and use the
hot standby - depends entirely on the controller and its driver s/w /
firmware. When drive fails - heads or surface or servo etc., the defect
count goes from nought to a large number in a flash so RAID controllers
rapidly toss them out.
In the past (this is no doubt still true, although many aspects of this have
changed with advances in technology and huge increases in volume size) with
SCSI drives, there were 2 types of defect lists recorded on a drive when you
purchased it: the vendor list where the vendors advanced diagnostics had
determined that a sector is falwed in manufacture or was likely to become
defective. The second type of defect is the type that the manufacturer is by
far best equipped to detect. This list is the "must preserve" list for SCSI
as if it is lost it (EG via a wrong type of low level format) you will have
never ending problems with the drive until you or your OS rediscovers all
the Vendor Defects. The other defect list is the list that is added to by
Fault Tolerant drivers or perhaps the HDD firmware itself. IE the "User"
defect list. This list is - last I read - for IDE / SATA drives supposed to
be empty when you buy a drive and to remain empty for the warrantee period -
check with the manufacturer and its warrantee.
If you really want to try to resurrect a drive with bad sectors / blocks
then try checkdsk or products like scanright (?) - some people swear by the
latter. Checkdsk will attempt to detect and reallocate bad blocks for you
( the /f switch is the thing to use IIRC). On a system drive you have to do
a reboot for this to happen after you specify checkdsk to run.
Personally, if you are trying to build a mirror, get a h.w controller,
verify your current new disc, checkdsk it, use it for a while, then backup /
restore to it from the current stuffed drive if you can. Then try whatever
you can to fix up the defective drive (warrantee replacement is a good idea
as it should have a zero defect warrantee) until it too tests 100% - if you
can't get it to test 100% then THROW IT OUT.
- Tim
