Barracuda 7200.11 1.5TB ABRT uncorrectable error NOT BUSY

[aussiechippie]

Hello genius types...

I have a fairly old Seagate Barracuda 7200.11 1.5TB drive that has become borked. After a bit of early learning, for a long time I thought was doing the well publicised BUSY behaviour. However, having used MHDD to try scanning the surface, I've found that at a particular block the drive goes to ABRT mode and will no longer respond to any read requests. The drive WILL actually boot, indicating that the problem block isn't actually used for a core OS file, but MY GOD does it take a long time.

I haven't dug TOO far into the drive yet, such as hooking into the serial interface, although I do have the means to do so fairly easily (USB to 3.3v TTL serial interface, for example).

I have tried scanning with MHDD from an LBA well above the problem block, and the scan progresses, albeit slowly, with some very slow read times in clusters of blocks that are widely (though regularly) spaced. I have been trying to minimise the spindle time now that I'm aware that the drive has a serious problem, and I'm trying to find a cloning tool that can accept a "skip" list. I've found the article: viewtopic.php?f=1&t=7850&p=47953 which seems to give a good basic approach (mine would be type 2), and I'm hoping that since 2007 either someone has produced a tool that automates the steps, or that a tool exists to make it fairly easy to do by hand.

I have a 2TB WD blue drive sitting here ready to accept the cloned data, to give me a recovery master. Unfortunately the dinky old HP/Compaq dc7100 corporate desktop machine I have available to do this work with only has two SATA interfaces (and I have a basic USB/SATA external case) but I have been able (a little unreliably) to boot the machine from a USB drive to run MHDD from DOS. Should I be looking for some better hardware?

So, I'm not a complete noob, but I feel like one in this field because my background many years ago is tinkering with OS internals, not peripherals so much. One thing I DO know is to tread very very carefully!

What advice can you far more knowledgeable folks give me to start making some progress here, because I think I've hit the limit of what I can do alone (at any decent rate of progress). I'm hoping (pretty confident) I can get this done with the hardware I have, and really just hoping to get a pointer to some suitable software.

TIA

[fzabkar]

See http://www.hddoracle.com/viewtopic.php?f=83&t=557

[maximus]

I would like to suggest HDDSuperClone free version (Linux based) as a possible cloning tool that could meet your requirements.

http://www.sdcomputingservice.com/hddsuperclone

It is designed to exit when a device fault is detected, which is very possibly the case you are explaining. A device fault requires the drive to be power cycled before it will read again, so you can power cycle the drive and then continue by restarting HDDSuperClone. HDDSuperClone has a built in head skipping algorithm and can be stopped and resumed as it uses a progress log file to keep track of all read attempts. In your case it would be best to run it with the --ata option as that will detect a device fault directly. Plus it has many options to manually control the cloning, although in many cases the default settings are good.

Now, if your drive does suffer from a "slow read" issue where the drive is trying do something in the background like do a scan or attempt to reallocate bad sectors, this can be an issue as it could cause HDDSuperClone to think there are many more bad areas then really exist. But if it is just going south and locking up when reading a specific block/sector, then you may be in luck. If there are only a few of these areas then you should be able to clone the rest of the drive. But if there are a lot of them then you would be constantly power cycling the drive to continue, and without an expensive hardware cloning tool that would not be feasible.

[maximus]

I forgot to add the normal disclaimer. If this data is valuable then seek professional data recovery, and stop using or messing with the drive! But if you insist on trying it yourself, then just understand that everything you do can cause further damage which could make it less likely for successful recovery.

The drive WILL actually boot, indicating that the problem block isn't actually used for a core OS file, but MY GOD does it take a long time.

And most definitely QUIT USING THE DRIVE! STOP THIS NOW! YOU WILL KILL IT! DO NOT BOOT INTO THE DRIVE'S OS!

Any why will this kill it? Because the OS will just hammer on the drive trying to read the bad areas, and also possibly try to repair bad spots in whatever way it thinks is best. These are very bad things.

[aussiechippie]

See http://www.hddoracle.com/viewtopic.php?f=83&t=557

Thanks fzabkar, I'll look into that. I have the interface ready to set up, so it's time to start learning the commands and the level structure, I guess. From what I gather, the intention is to tell the drive to not attempt to re-map bad blocks any more, and just return whatever it gets even if the ECC retries get exceeded? I only had time to quickly look at the page from my phone earlier. I'll read it in more detail this weekend.

The drive WILL actually boot, indicating that the problem block isn't actually used for a core OS file, but MY GOD does it take a long time.

And most definitely QUIT USING THE DRIVE! STOP THIS NOW! YOU WILL KILL IT! DO NOT BOOT INTO THE DRIVE'S OS!

Believe me, I know It was only an accidental discovery when I was attempting to boot my host machine from a USB stick. It booted from the drive instead ! It had also been in use for a long time prior to this hard fault appearing, and had been getting progressively slower over the past 6-8 months (I know, I know, SMART aware bios etc etc).

That software will be the go, I think. I'll just have to set up a USB linux boot, or maybe a CD-ROM, and get it going that way. I only have two SATA ports and I'd prefer to dump the clone data via SATA for speed. The machine doesn't have USB 3.0 AFAIK. If HDDSuperClone has some sort of way to command a power cycle I'm sure I can build a device to do that (a relay, a serial chip handshake line, bit banging a parallel port, how hard can it be <tm>). I'm fairly sure that the drive isn't trying to reallocate blocks, because that seems to be what got us into this trouble in the first place. I'm not completely confident of how I'm interpreting the data in the SMART results, but MHDD seems to be suggesting that the drive is out of swap blocks. The slow reads are probably just ECC retries.

Thank you both for your help so far.

[maximus]

If HDDSuperClone has some sort of way to command a power cycle I'm sure I can build a device to do that (a relay, a serial chip handshake line, bit banging a parallel port, how hard can it be <tm>).

HDDSuperClone does not currently have a direct method of calling for a power cycle event. However, it will exit with specific exit codes (see the documentation) that will allow for the end user to write a shell (bash) script to perform actions based on the exit code. And by this I mean you write a script that calls HDDSuperClone, and when it returns check the exit code, if a certain value perform whatever operation to power cycle the drive, and then restart HDDSuperClone.