HDD failure

On Tue, 26 Feb 2008 00:09:57 -0800 (PST), rowan194 put finger to keyboard and composed:

My 13GB Seagate drive reported 119 reallocated sectors just before I took it out of service (because it was growing new defects on a weekly basis):

Its S.M.A.R.T. status was still OK.

I think the figure of 99 you are referring to may be the "normalised" value, not the raw value. I believe my 13GB drive would have been allowed to develop several thousand bad sectors before its S.M.A.R.T. status would have been reported as bad.

- Franc Zabkar

Here's the diagnostics on the dud reconditioned drive that Seagate sent me

Short test FAILS, long test finds 80 errors but is able to remap them and therefore SUCCEEDS, a subsequent short test FAILS

Shortly after this I rebooted and the drive reported a S.M.A.R.T. event.

Note "POH 1" which means it's only been powered on for a total of 1 hour! (Not entirely true, since it's a reconditioned drive the mechanicals could have been run for who knows how long. The firmware was just reset to 0)

This was a replacement for a faulty drive and it turned out to be even worse... may as well have just kept the first one...

On Wed, 27 Feb 2008 03:36:51 -0800 (PST), rowan194 put finger to keyboard and composed:

You might like to try smartctl:

Here is the log for my 13GB Seagate drive:

- Franc Zabkar

Yeah, I'm aware of smartctl, but unfortunately the Windows RAID driver presents the array as a single SCSI drive. Have tried many diagnostic utils with the same result. It's likely that even in another OS (such as *nix) there may still be issues with detecting and reporting data for individual drives. The only other thing I can think of is to buy a standalone PCI SATA card and selectively swap drives between the onboard RAID and the standalone SATA controller, but doing so will invalidate/degrade the array, requiring rebuilding between each drive swap. May as well just do the Seatools tests...

On my FreeBSD servers I'm moving over to software RAID1, which has the advantage that every HD util can still see the individual drives. I've run smartctl there. (Easy to see if a webhost is using second hand drives!)

On Wed, 27 Feb 2008 14:22:15 -0800 (PST), rowan194 put finger to keyboard and composed:

If you drop back to DOS, then you could try SMARTUDM. It reports raw SMART attributes. You may not need to run a complete Seatools diagnostic. In any case, if the bad sectors have already been reallocated, then there'll be nothing for Seatools to report.

See

SMARTUDM is claimed to support "external UDMA/SATA/RAID controllers (HPT/Promise/FastTrack)".

- Franc Zabkar