If you have somewhere to dump the drive's contents, dump the entire
thing out (insofar as is possible - ddrescue is a useful tool for
this, though I prefer GNU ddrescue to dd_rescue), note sector realloc
count, zero entire drive, check if pending sector count is nonzero. If
so, tell them that on the phone, and complain repeatedly that the
drive is misbehaving, either by miscounting pending or by not
reallocating on zeroing (possibly due to being out of spares), and in
either case, the drive is bad (and really should fail a short test,
"Pending Reallocation" sectors are sectors the drive has failed to
read enough times that it's internally marked them as "when this
sector gets rewritten, or when I ever successfully read it, reallocate
from spare sectors and mark as bad". Consequently, for obvious
reasons, zeroing the drive does the right thing. But make no mistake,
it's not a transient state - pending sector means "I have a count of
failed reads on this sector large enough that it's terrifying me and I
never want to trust it again". It's also not authoritative, which is
why marking a drive as "failed" for having pending sectors is a bad
idea - this is A) just the sectors the drive has noticed are bad, and
B) could be a transient problem, not an actual critical failure (e.g.
a giant scratch running down a platter or horrible particulates
bouncing around inside, versus a few sectors that happened to fail
shortly after leaving the factory).
*shrug* I don't know what else to say. I've never, ever had any
problem with telling Dell "your test passes. i promise, the drive is
failing.", and not because I have some hilarious Tier 1 relationship
with Dell or something. Just keep complaining if they say it's fine
and you have evidence that it's clearly not.
On Wed, Aug 18, 2010 at 3:18 AM, Xavier Chantry
> On Sun, May 16, 2010 at 5:07 PM, Rich <rincebrain@...> wrote:
>> I've never had Dell refuse to exchange a drive just because it passed
>> their sanity check. YMMV, but if you tell them "this drive
>> consistently has SMART errors and they only transiently go away", they
>> should take it.
>> That said, a few reallocated sectors isn't much to worry about - keep
>> backups, and if the reallocated count never stops growing, or if you
>> run out of spare sectors, then get worried.
>> - Rich
> The reallocated count never stopped growing to reach 145 today.
> I called DELL and as I expected they wanted me to run the sanity check
> to see if it fails.
> Since the short test pass, no exchange :)