Failed WD5000KS

[benhanson]

First off, I'm familiar with drive recovery to the extent that I've done plenty of recoveries from minor head crash, partition corruption, etc using ddrescue to image a drive, then repairing the image or recovering data from the image. I've never done a pcb swap to recover, since I've not had a customer willing to go to the expense, and I've never done any firmware/rom recovery mostly because I'm a Mac user at home and all the manufacturer tools tend to be DOS or Windows based. That said, I'm actually a network/server admin for a Windows shop, so I'm more than comfortable in the Windows realm.

I'm working on a WD50000KS-00MNB0 that appears to have a PCB or Firmware failure. The drive spins up, is recognized by my system(OS X 10.5.x). Disk geometry appears to be correct, OS actually appears to have proper folder and file count, and at least once in the last 10 attempts I've actually been able to read a small amount of data from the disk. That said, most of the time I don't get anything, no directory listing, nothing via ddrescue. ddrescue ends up logging the entire drive as one giant error, and it cruises through the entire 500gb drive in no time, such that it doesn't actually seem like it's accessing the platters.

The drive sounds like it's doing about 10-15 short repetitive seeks as it starts up, but it doesn't really sound like a head crash click. I'm thinking the mechanicals are OK. Anyone have any advice for the next step?

[drc]

Use MHDD to get a better look at what's going on.

Does the drive identify? (F2)
How does the SMART info look? (F8)
Can you scan it? (F4)

Having said that, the vast majority of these drives that I have seen are suffering from weak or failed head(s). I have not ever seen one that was only a PCB failure. They usually make obvious clicking sounds when powered on though, so maybe your case is different.

[benhanson]

Wow, very nice tool!

Drive Identifies(F2):
WDC WD5000KS-00MNB0 LBA:976,773,168 BIOS: 80H
SN:CORRECT# FW: 07.02E07 CACHE: 16384KB Size = 476940MB

SMART Info(F8):
#: Name: Val: Worst: Raw
1: Read Error Rate: 200 : 200 : 0
3: Spin up time: 225 : 218 : 5741
4: Number of spin-up times : 70 : 70 : 30847
5: Reallocated sectors count : 200 : 200 : 0
7 : Seek error rate : 122 : 118 : 1581
9: Power-on time : 96 : 96 : 3212
10: Spin-up retries : 100 : 100 : 0
11: Calibration retries : 100 : 100 : 0
12: Start/stop count : 98 : 98 : 2319
194: HDA Temperature : 253 : 253 : 34
196: Reallocate event count : 200 : 200 : 0
197: Current pending sectors : 200 : 200 : 11
198: Offline scan UNC sectors : 200 : 200 : 0
199: Ultra ATA CRC Error Rate : 200 : 200 : 0
200: Write error rate : 200 : 200 : 0

It does scan, looks like it has a few bad sectors(UNC). UNC count is currently 9 with 1% complete. I'll let it finish out the drive.

[BlackST]

The symptoms are strongly related to a weak head. Next step before it's too late : get what is REALLY relevant before it's too late UNLESS you have a professional tool for data extraction that can disable reallocation and maybe extract data selectively head by head. Note that every attempt can add +1 to G-list or make head weaker, until the drive won't respond anymore and will be recognised with his alias at startup.
What I would do : proper diagnose, imaging , if not applicable because imaging won't work and every bit of data is important AND the customer will pay for it , head swap in controlled environment, then imaging. It requires HW tools like PC3000 and DE or alike. I trust no SW solution in this case.

P.S. you should know everything about the alignment issue, don't you ?

[BlackST]

I didn't see the SMART report when I was posting , it's not "too bad" , except for attribute 197 and 7 . Get data before it's too late .

[benhanson]

Any suggestions for a tool to image the drive? I usually use ddrescue to do a block level copy to an image, but that is failing me in this case.

[benhanson]

And what is "the alignment issue"?

Still holding at UNC=9 with 12% complete.

[benhanson]

Output from mhdd is pretty interesting. Is it normal to see 99.2% at <3ms with .8% at <10ms?

530,000 <3ms
4,000 <10ms
68 <50ms
1 <150ms

[benhanson]

OK, the scan completed, only marked 9 blocks as UNC. Then tried doing a scan with remap turned on, and it seems like the drive gets kicked offline as it tries to remap. After escaping out of the scan, the drive is no longer visible(can't ID, get smart, etc), then if I reboot, the drive is back and still shows 9 UNC, scanning the first 1,000,000 blocks. Is there a trick to remapping bad sectors?

[drc]

If you can scan it to the end with minor issues then a smart software imager should be fine. I've had issues with ddrescue from time to time. Try booting into linux (try a livecd like knoppix or something) with as few things touching the drive as possible (auto fstab, udev, whatever). Remember that ddrescue has a dumber older brother called dd_rescue that is more controllable. Some people here always shout me down when I mention it but it usually works better for me.

[BlackST]

No way to remap the bads without special tools or recert the drive unless you want to spend more than 10'000 $. Alignment: remove the screws, open the cover and see what happens (DON'T DO IT).