6
I have a Western Digital Elements Desktop USB-connected hard drive. Basically, it's a glorified package for a WD30EZRX drive. I use it to hold system backups (old now, at that), so there's nothing critical on it. I could probably wipe it and throw it away right now and not miss anything.
A few months ago, I effectively replaced the above-mentioned drive with a different one, but I've been keeping this old one (partly in case I need to refer to an older revision of some file, partly because I intended to convert it to an off-site backups drive). It has been connected and powered up, but unused; the file system has not been mounted, so the only activity it should have seen is a partition table scan on boot and possibly ZFS looking its way a few times to see if there's any partition on it that's part of a pool.
Also, about a month ago, I set up smartd for monitoring the health of the various disks connected to my system. It immediately screamed bloody murder about this drive reporting a pending (unreadable) sectors count of 5. Knowing that by themselves, pending sectors are managable, I kept the drive connected but unused.
This afternoon, the email report from smartd suddenly indicates that there are 6 pending sectors, as well as 1 offline uncorrectable sector (this is new).
Here's the weird part: the most recent reboot, and thus when the drive should have seen any activity most recently, was almost four days ago.
The drive logically holds a single partition spanning the entire disk, which houses a LUKS container that has not been started in the period since I set up smartd for monitoring the drive's health. It has never been part of any RAID array or similar on any level (drive, partition, LUKS container, contained file system).
Checking with smartctl --all the drive reports that no self-test has been logged. That also confirms the pending sectors count of 6 as well as the offline uncorrectable count of 1.
What might have caused the increase in the pending sectors and offline uncorrectable sectors count when the drive should not even have seen any activity?
Note: I am not asking whether I should keep using this drive. It's obviously becoming unreliable at this point, and will be retired; I've suffered from data loss through hard drive crashes too many times to take any real chances with my data.
a CVn
Posted 2013-12-29T16:04:55.127
Reputation: 26 553
1This is the first time where I would consider an all caps word in the title. (in the "on an IDLE drive" part). As for the "obviously unreliable" part. A single pending sector increase does not mean it is about to die. If bad or pending sectors start to accumulate rapidly then throw it way, but an extra single sector does not mean you need to discard it. Keep it, use it to carry photos or movies around or other unimportant tasks. But I agree that it is not longer nice as a drive containing backups. – Hennes – 2013-12-29T16:06:48.950
@Hennes Yes, that's the puzzling part! I could easily see the pending sectors count going up if I did a read pass over the drive and some of the data on the platters was bad, but in this case, unless Linux is doing something really weird on its own it really has been sitting idle! – a CVn – 2013-12-29T16:07:52.313
Does the drive have an 'self test' mode which triggers every few days? Or is it on a RAID controller which occasionally does a patrol read (reading every sector on a drive during disk idle to make sure it the data is still intact) ? --- Edit: No self tests are logged. Darn, you already thought of self tests. – Hennes – 2013-12-29T16:11:04.507
@Hennes I'm not aware of either. I'd expect any such self-test to be reported in the SMART log, and the drive is not part of any storage array. It's just a plain external hard disk, hooked up via USB and with a single partition on it holding a LUKS container. LUKS isn't even started for that drive. – a CVn – 2013-12-29T16:15:10.810
@Hennes I'm not so concerned about the pending sectors count going up (that's easy enough to work around by using it only for non-critical tasks as you say and/or putting everything on it in duplicate; the odds of corruption hitting both copies of any single file would be small), as I am about the pending sectors count going up when the drive shouldn't really have seen any activity at all. – a CVn – 2013-12-29T16:18:50.927