WD Blue < 2 years old died this evening.

[Goobley]

Well this was a surprise. My main drive that I purchased in Sept/Oct 2013 decided to go corrupt this evening. I was playing the Witcher 3, and some textures hadn't streamed in off the drive. I then alt-tabbed out and went to browse for a few minutes.

Everything was fine for a couple of minutes until the machine locked up and then blue-screened with something along the lines of KERNEL_PAGE_ERROR. After rebooting the windows boot loader kicked in before saying something along the lines of "missing required material" and suggesting that windows be reinstalled.

My main linux install, which is on a second hard drive, also didn't want to boot properly and just chucked me into a root prompt. So I booted off a linux usb key I had - which took over 5 minutes due to the number of segment reallocation errors on the main drive. Windows installer repair just hangs on scanning the system.
`ntfsfix` just said "corrupt, try running chkdsk". I couldn't think of any way to run chkdsk without getting windows past the boot loader.
Then I remembered S-F10 on the installer screen to get a cmd prompt. So I now have chkdsk /F running and I've never seen the like of it. In ~20 minutes it's checked 22 sectors (of 451328) of which 4-22 have been unreadable (i.e only 2 or 3 readable). I'm gonna leave this running into the morning and see what happens. I don't expect much, but my last full backup was 6 months ago (*wristslap*, "I've been busy" is not a good enough excuse).
Most of my big linux projects are fine on the other drive, but I've probably lost a fair amount of day-to-day stuff.

If/when this doesn't get me anywhere, does anyone have any other suggestions? Am I correct in assuming that I should be able to rma the drive for this?
Cheers!

[Splash]

If it's still in warranty, yep.

[DanceswithUnix]

Have you checked the power and data cables to the drive?

Once you are sure of those, the next command should be from your Linux usb key of (assuming it was your primary drive):

smartctl -a /dev/sda

and see what your smart stats say.

If the drive is failed, then bad luck, but 2 years should be within warranty.

[Goobley]

I'll get onto rma'ing it once I'm sure there's nothing else I can do. I left chkdsk /F running for about 6 hours off my windows install usb key. It had made it to volume sector ~440 of 451k. So I stopped that.
After a bit of hassle I got my linux USB key online, (hooray for lying on the floor in the hall next to the router), and installed smartmontools.

The SMART status itself says OK, but the tests seem to show otherwise. From what I can see I'd say "he's dead Jim, but I'm a physicist, not a hard drive expert". Can someone with some more knowledge than I have a look at this log and give me their opinion please?

Cheers!

[DanceswithUnix]

Just one line looks worrysome:

197 Current_Pending_Sector 0x0032 199 199 000 Old_age Always - 213

Sounds like 213 sectors couldn't be read, and if you write to those sectors it will re-allocate new ones to replace them. That is quite a lot, when there should be zero. I ma guessing from the grief you had, that some critical directory information was hit.

A destructive surface scan should "fix" the drive, but I would run several of them and see if the total allocated sectors (ID #5) climbs.

[Goobley]

Thanks very much for the advice so far!
I'll look into it, but I'm hoping to avoid a destructive scan for now. I've managed to get my linux booting again (on my second hard disk) by modifying fstab to not mount the damaged drive at boot. Currently badblocks is running non-destructively on the partition that interests me the most. Do you think there's any chance of getting anything from ddrescue if I get another drive to clone onto?

[DanceswithUnix]

Sounds like most of the data should be there somewhere, just depends on how much time/money you want to invest in it. Stuff like running a filesystem check against it may have moved stuff around and made it harder though. I usually try and recover any data from a drive mounted read only, but that time was long gone before your first post here. One to remember for next time though.

[Goobley]

Ah, the joys of experience. I swear I'm cursed with hard drive failures far more than is statistically likely, I've been through 5 in the past 10 years, and I've never owned more than 3 hard drives at a time or had them running 24/7 etc...

[DanceswithUnix]

Ah, the joys of experience. I swear I'm cursed with hard drive failures far more than is statistically likely, I've been through 5 in the past 10 years, and I've never owned more than 3 hard drives at a time or had them running 24/7 etc...

Ah, but they like running 24/7

Well, some do.

[chrestomanci]

I had the main hard drive on my Linux system fail about a year ago. (Strangely enough it was also a WD one).

Anyway, I wrote up a post describing the steps I used to recover my data here:

http://forums.hexus.net/storage/3253...ml#post3305093

In my case, there where a large number of bad sectors, but the drive still mostly worked, so I was able to get most of my data back, bar a few hundred megabytes of mostly replaceable data.

One thing I would add, is that if at all possible, avoid writing to the failing drive. You should use recovery tools to read as much as possible from it, and then run filesystem repair tools on the coppied image.

In my case, I also made use of Linux LVM snapshots so that I could have multiple attempts at filesystem recovery, and if it failed, I could still revert to the original, damaged but not made any worse filesystem image.

[Goobley]

Thanks I'll look into that in more detail later. Before I can recover anything I'll have to get a new drive, because clever me put my documents on a 400GB partition, of course, I don't have anything else with anything like that much contiguous space to attempt a clone. All since I got up this morning I have only been using read-only software. Given that I have some work over the summer that'll bring in a trickle of money I'm probably going to get something RAID to backup on to. I was lax in backing up for a time and it's caught me out.

[Goobley]

Interestingly, whatever happened when this drive went down (and due to this I'm no longer convinced the drive was the root problem) also took out my usb WiFii adaptor. I've tried it in my normal Linux and off a fresh USB key. Both see the the adaptor and it sees the network it just won't accept the password, and actually connect to it. Tried it on my macbook (in Windows with the drivers), same deal.

Went and got the most likely one I could find in town this afternoon and as soon as I installed the correct kernel module WiFi works again. The old one still behaves the same way.

Now, I was trying to refresh pages when my machine blue-screened in the first place, so maybe it got frazzled somehow, but I'm struggling to see how...

For info:
old adaptor: Asus WL-167g v3 (about 1 year old, used for only 4 months of that time, when it's always been stable).
new adaptor: Asus N10 (hate these nano adaptors, but it was the only one I recognised that I knew was commonly used in Linux).

EDIT:: Hang about! I just remembered that almost as soon as my 'puter died people in the flat started to say that the network was down. I didn't pay too much attention because I couldn't boot at the time.
This would be a lot of coincidence. Thoughts?

[Goobley]

After one full pass of the partition that interests me (which isn't the boot partition, so probably/hopefully wasn't too mangled by chkdsk) ddrescue reports the following (it's started its backwards pass):

current pos: 344830 MB, current status: copying
logfile extent: 423484 MB, in 58226 area(s)
rescued: 124195 MB, in 1238 area(s) ( 29.32%)
non-tried: 297202 MB, in 28360 area(s) ( 70.18%)

errsize: 2086 MB, errors: 28628 ( 0.49%)
non-trimmed: 2086 MB, in 28628 area(s) ( 0.49%)
non-scraped: 0 B, in 0 area(s) ( 0%)
bad-sector: 0 B, in 0 area(s) ( 0%)

Do people with more experience of this than I feel like its worth my time continuing? I reckon it'll be another couple of days to do enough passes to try every block and 0.5% errors already sounds pretty unsalvageable to me.

[mikerr]

Interestingly, whatever happened when this drive went down (and due to this I'm no longer convinced the drive was the root problem) also took out my usb WiFii adaptor.

EDIT:: Hang about! I just remembered that almost as soon as my 'puter died people in the flat started to say that the network was down. I didn't pay too much attention because I couldn't boot at the time.
This would be a lot of coincidence. Thoughts?

Maybe the it was actually the mains electricity being flaky at that point ?

[Goobley]

Ended up running ddrescue for over a week on the partition that I hadn't tried to repair. It rescued 170GB but the rest (as far as it got) was errors, so there was no mountable filesystem in there :/ Gonna RMA the drive in a few weeks when I get back from staying with family. Luckily nothing majorly important lost, set my music collection organisation back a bit (since I organised it properly after my last full backup), but I'm pleased to say I haven't come across any important missing files.

I am backing up again now (learnt that lesson, eh?) and thinking of grabbing a small NAS or similar and just keeping a couple of cloned copies of my data partitions on separate drives in it (maybe just use an RPi 2). None of my data is worth much at the moment, but as I near the end of my degree and work on other things I get the feeling that that will start to change.

I am feeling that maybe some sort of power fluctuation was the cause, a little UPS probably wouldn't be a bad investment.

Thanks again to everyone who contributed their expertise here.