Is this disk about to fail?

[Emirzan]

I have a relatively old Western Digital 640GB hard disk in my computer. This is a disk I use for data storage only (not running applications or an OS) and it has given me no signs of failure until today.

I noticed when running my linux OS that the disk could not be read. I booted into Windows 7 and the disk was visible however in the Disk Management utility it says 'Healthy (At Risk)'. I ran the error scan utility in HD tune to scan for 'bad sectors' and one third of the way through the whole scan stopped marking the rest of the risk as bad and the disk was no longer mounted in windows. I could not remount it so restarted the PC.

Again using Windows the disk is now visible again and the Windows chkdsk utility can find no errors. The disk management screen now just says the hard drive is healthy again. I ran a test copy and paste operation on the drive of 50GB and it went fine. Was this just a temporary blip or would you assume the drive is at the point of failure? Hours powered on is at 32,000 (approx 3.5 years if at constant use) so it is quite old.

Everything on the hard drive in question is already backed up byte for byte so data loss is not an issue.

[peterb]

What file system do you use on the disk? If it is NTFS do you have the NTFS Drivers installed on the Linux box?

Is the drive S.M.A.R.T capable? (Most are) and if so can you run smartctl on your Linux box to read the S.M.A.R.T data?

[smargh]

Depends, but I'd say it's on life support judging from your description. There could be mitigating factors though - dodgy power supply, faulty RAM etc.

Download Speedfan. Click the button to run the online diagnostics, which will give you some human-readable details on the state of the various SMART health attributes.

[DarkflameZM]

Check SMART, bios generally warns when a drive is dying, i use Acronis Drive Monitor to keep an eye on my non SSD drives.

[Bonebreaker777]

Download HDD Scan so you can check the SMART plus you can preform a surface check to see if you have really some dodgy parts you have or will have problem with.

[DarkflameZM]

Thanks for the HDD Scan program suggestion, wasn't aware of that one

[noveltylamp]

Definitely check the SMART table for "Reallocated sector count", "Reallocated event count" and/or "Pending reallocated sector count" or similar as suggested, also try swapping (or at least re-seating) the SATA cable if the SMART values look good but you're getting file system corruption.

[Emirzan]

Apologies for the delayed response. Not had much time in front of a computer since writing the opening post. Thanks for the suggestions, below is the SMART status as reported by HDD Scan. One reported fail on 199 UltraDMC CRC Errors



I'm making my judgements using wikipedia http://en.wikipedia.org/wiki/S.M.A.R.T. and based on that reallocated sectors (item 005) seems high to me - does anyone know why HDD scan shows a green symbol when the value of 200 is over the 140 threshold? That seems a clear fail to me?

Similarly items 197 and 196 seem high, none of which looks good.

Back in the real world the drive is still behaving oddly, it keeps un-mounting and re-mounting without prompt in windows and when I tried to play music files stored on it there were occasional pauses (circa 1-2 seconds every couple of minutes).
(Edit) - and I meant to include that I would try scanning it with HDD scan but it fails each time because the drive unmounts itself.

I'll try replacing the cable to see if that makes a difference to anything but I think it is destined for the bin.

[DanceswithUnix]

CRC errors is the only one of those that looks bad. You could in Linux have just done a (assuming the drive is sdb) "smartctl -a /dev/sdb" and get all the info.

Most values start high and then go lower as the device fails. So looking at the raw values you have zero uncorrectable errors etc so the drive looks OK.

So I would replace the cable and see how it goes.

[abaxas]

Personally if any disk looks even a bit fishy, I replace.

The cost of effort/faf on failure is greater than the cost of buying new.

[smargh]

The online report function of Speedfan is significantly easier to read compared with looking at the SMART output itself.

Could be the cable. Could also be many several other things.

I use h2testw to write to free space & then verify that the stuff written is valid. It was designed to test for fake USB drives but works well with hunting for problematic HDDs.

[peterb]

You said it was a relatively old drive - is it a SATA drive? If so, I'd definitely look at the cable as the source of errors. The connectors are quite flimsy. I've had a couple fail on me.

[Emirzan]

It is indeed quite old, bought in 2009 I think.

I have changed the cable for a brand new one and so far the drive in question seems ok. I haven't had it unmount itself and I have been able to run a scan of the disk which didn't find any damaged sectors. And, not that it is a particularly scientific test, but I am able to play the same music files I previously tried without any pauses or interruption. So it does look as though the cable has failed, not something I've experienced before.

[peterb]

So it does look as though the cable has failed, not something I've experienced before.

Ive had two fail - in both cases it was a hairline crack in the side of the connector shell - hard to see, but when they were connected to the drive, a bit of vibration was enough to cause an intermittent or noisy connection, causing errors.

[Agrippa]

Personally if any disk looks even a bit fishy, I replace.

The cost of effort/faf on failure is greater than the cost of buying new.

In my experience, this is the only safe approach. SMART is a great feature, but hardly something to rely on. I've had numerous disk fail without the slightest warning. While any number of things could cause the problems, all too often there's just no way to be entirely sure and the only road to peace of mind is a new drive.

[DanceswithUnix]

In my experience, this is the only safe approach. SMART is a great feature, but hardly something to rely on. I've had numerous disk fail without the slightest warning. While any number of things could cause the problems, all too often there's just no way to be entirely sure and the only road to peace of mind is a new drive.

That is what backups are for.

In this case, the SMART data clearly showed comms errors talking to the drive so the cable should fix it.

It is however good practice to replace a drive when it goes out of warranty, which at 5 years old must be the case here.