What should I make of a drive for which S.M.A.R.T. reports 8 bad sectors
(a single 4K block, I assume), but badblocks reports no errors on a
read-write test?

Perce

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 7/2/19 10:41 AM, Percival P. Cassidy wrote:

What should I make of a drive for which S.M.A.R.T. reports 8 bad sectors
(a single 4K block, I assume), but badblocks reports no errors on a read-write test?

I think it depends on the drive is.

I suspect that any drive that has S.M.A.R.T. also has spare sectors that
aren't visible to the OS.

As such, it's entirely possible that the 8 sectors are indeed bad and
have been swapped out with spares that are good.

Running a bad block check from the OS (outside of the drive & integrated controller) on such a will likely not see any problem(s).



--
Grant. . . .
unix || die

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 7/2/19 2:31 PM, Grant Taylor wrote:

What should I make of a drive for which S.M.A.R.T. reports 8 bad
sectors (a single 4K block, I assume), but badblocks reports no errors
on a read-write test?

I think it depends on the drive is.

I suspect that any drive that has S.M.A.R.T. also has spare sectors that aren't visible to the OS.

As such, it's entirely possible that the 8 sectors are indeed bad and
have been swapped out with spares that are good.

Running a bad block check from the OS (outside of the drive & integrated controller) on such a will likely not see any problem(s).

Now, after just one more run of badblocks, S.M.A.R.T. reports no errors.
So I assume that now a spare block is in use.

Perce

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 7/2/19 2:49 PM, Percival P. Cassidy wrote:

Now, after just one more run of badblocks, S.M.A.R.T. reports no errors.
So I assume that now a spare block is in use.

So, the disk likely did have a problem. But I'm guessing it has worked
around the problem.

Now the question becomes is the problem stable and not going to grow?
Or is the problem going to spread. Only time and monitoring will tell.



--
Grant. . . .
unix || die

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 7/2/19 5:44 PM, Grant Taylor wrote:

Now, after just one more run of badblocks, S.M.A.R.T. reports no
errors. So I assume that now a spare block is in use.

So, the disk likely did have a problem.  But I'm guessing it has worked around the problem.

Now the question becomes is the problem stable and not going to grow? Or
is the problem going to spread.  Only time and monitoring will tell.

This is a brand-new Seagate drive, so do I return it for replacement, or
assume that the problem is now solved (false alarm perhaps), or wait
till it acts up again and get it replaced by Seagate by a "refurbished"
(i.e., tested more thoroughly than a brand-new one) drive?

BUT, since S.M.A.R.T. now reports no errors, why should either the
vendor or Seagate replace it?

Even if it craps out altogether, it's part of a RaidZ2 pool (two drives
can fail without data loss), and I do have a spare with which I can
replace it.

Perce

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)

On 7/2/19 4:17 PM, Percival P. Cassidy wrote:

This is a brand-new Seagate drive, so do I return it for replacement, or assume that the problem is now solved (false alarm perhaps), or wait
till it acts up again and get it replaced by Seagate by a "refurbished" (i.e., tested more thoroughly than a brand-new one) drive?

Depending how convenient it is to exchange, I'd be inclined to feign
ignorance and say it's not working and ask for a replacement.
Especially if it's a moderately convenient brick and mortar store.

An online acquisition would depend on their return policies, shipping,
time frames, etc.

BUT, since S.M.A.R.T. now reports no errors, why should either the
vendor or Seagate replace it?

Ya. Of course it does. That's going to make it even harder to get a replacement from Seagate.

Even if it craps out altogether, it's part of a RaidZ2 pool (two drives
can fail without data loss), and I do have a spare with which I can
replace it.

Depending on how annoying it will be to exchange, I'd likely go ahead
and put it in the ZFS pool and deal with the drive in the future.

I will say that I'd be inclined to run SpinRite on the drive before
putting it into the pool. I'd probably crank it up to a level 4 or 5.
The high level being a desire to exercise the crap out of the drive for
fear that it's might have infant mortality. I'd rather it die before
putting it into service than after.



--
Grant. . . .
unix || die

--- SoupGate-Win32 v1.05
* Origin: fsxNet Usenet Gateway (21:1/5)