Hi all,

Question is in bold below.
I've been asked if I can recover data from a 1TB SATA Toshiba HD (P/N 9F13180) which is failing.
The drive is currently in a HDD dock connected to my PC via USB. Windows assigns a drive letter but states the drive needs to be formatted.

My issue is that when scanning the drive or copying sectors in DMDE the drive tends to disconnect at certain points (the drive disappears in windows). It looks to be disconnecting at specific sectors as opposed to after a certain amount of time. On disconnecting the drive will sometimes reappear again, other times I'll need to power off the dock and power it back on in order for the drive to show. Once back on I can hit reset in order to reconnect to the device and continue with the scan.

This happens way too regularly to make an entire drive scan/copy practical but I can see from the limited amount I've scanned that DMDE is finding plenty of files it can recover.

Another side effect is that there are many other read errors which I have to manually deal with because I can't click "Ignore all". DMDE doesn't check whether the drive letter exists on error so If I click ignore all and the hard drive disappears DMDE will rapidly start skipping through the scan ignoring the errors generated by the drive being disconnected.

Has anyone experienced anything like this? Could the drive be disconnecting due to something in DMDE which can be changed or is it likely the hard drive is crashing and restarting itself?

Some things worth mentioning:
• All USB/HDD power settings in windows have been changed to ensure the drive doesn't disconnect because of windows.
• When connected directly to the PC via SATA cable the PC fails to boot.

Thanks for your time.

Smith

Ddrescue is a much better tool for cloning a drive with bad sectors.

Could we see the drive's SMART report? You could use CrystalDidkInfo or smartmontools for this.

Could we see DMDE's partitions window?

_________________
A backup a day keeps DR away.

What you are experiencing is absolutely normal with a failing drive. The more it is tampered with using software based tools, the more degraded the internal disks will become, eventually leading to severe scratch like damage.

While the ideally the best option is to use a professional hardware based imaging equipment (expensive), your best software based option is ddrescue as already suggested. But, pay attention to it. If the drive still chocks, then more than likely the damage is a lot more severe then it allows to show. Likely failing heads and so on.

_________________
Hard Disk Drive (HDD), Solid State Drive (SSD, SATA, NVMe, etc), USB Flash Drive and RAID Data Recovery Specialist in Massachusetts

Thanks for the replies.

I've attached the CrystalDisk and DMDE partition screenshots as requested (hope this is what you were after).

I appreciate the advice about ddrescue. I'm going to download a linux OS and attempt this in virtualbox or from a live CD (fingers crossed). I'd love to think ddrescue could image the drive without the same disconnect that DMDE runs into; But I admit I've no idea how that could be possible.

Regarding "professional hardware based imaging equipment", just out of interest do you have any recommendations?

Thanks again,

Smith

@Smith, your USB dock is part of the problem. Its firmware is configured for a sector size of 4096 bytes wherease your drive's file system was created in a 512-byte environment. This means that sector 0 (the MBR and partition table) is still in the correct place, but every other sector is out by a factor of 8. Windows thinks the drive is unformatted because it is looking for the boot sector in the wrong place. You need to connect the drive to a SATA port.

The other problem with the dock is that it uses a JMicron bridge IC. It appears that CrystalDiskInfo is unable to retrieve SMART from a drive that is behind a JMicron bridge. Once again, this problem will go away when you connect the drive to a SATA port.

_________________
A backup a day keeps DR away.

_________________
Hard Disk Drive (HDD), Solid State Drive (SSD, SATA, NVMe, etc), USB Flash Drive and RAID Data Recovery Specialist in Massachusetts

Scott Dwyer has written HDDSuperClone:
http://www.sdcomputingservice.com/hddsuperclone
http://www.sdcomputingservice.com/hddlivecd

This tool purports to be similar to but better than ddrescue.

_________________
A backup a day keeps DR away.

Thanks fzabkar, really interesting. I now have the drive hooked up via SATA.

I decided to go with your first suggestion of ddrescue (via SystemRescueCD).

Currently:

Running time of about 26 hours, Pass 1
Rescued: 27605 mb
ipos: 47124 mb
errors: 0

So probably many more days of scanning but it looks to be doing the trick and thankfully no micromanagement .

If the drive is slowing down and chocking, then that is a sign of it becoming severely degraded. Keep an eye on it as they go fubar unexpectedly.

May also wanna put a fan on it to keep it cool.

Good luck.

_________________
Hard Disk Drive (HDD), Solid State Drive (SSD, SATA, NVMe, etc), USB Flash Drive and RAID Data Recovery Specialist in Massachusetts

So just a little update on this.

DDRescue has now been running for just over 15 days and only has 32 years and 190 days left until completion . This number jumps around hugely of course but it's clearly going to run for months if I let it. What does concern me is the results so far after pass 1.

I've attached a picture of the terminal info and a visual representation of the log file processed by ddrescueview.

What concerns me is the pattern; DDRescue has apparently rescued 341780Mb of the drive but these rescued blocks are regularly cut up by presumably failed blocks?

I was wondering if the pattern displayed had meaning to anyone here? For instance is this pattern common when a drive head fails?

Thanks again.

I would ask "maximus" (aka Scott Dwyer) whether he thinks HDDSuperClone may be able to help in your case. It does appear that there is a bad head, though.

FWIW, here is an encouraging report from a user of HDDSuperClone (1 head out of 4 was dead):
http://www.sdcomputingservice.com/hddsuperclone/success-stories

In your case there appear to be two heads, one of which is bad. I'm not sure how many intact files you will be able to recover in such a situation.

_________________
A backup a day keeps DR away.

@islamm3rouf
I assume you mean for the initial connection issues? I have the damaged hard drive hooked up via SATA now.

@fzabkar
Thanks, I have doubts about getting much intact data also. I've downloaded the HDDSuperClone live CD and will PM maximus.

Cheers

I have been contacted about this case and whether or not HDDSuperClone could help. My first impression is that it could help read the data from the good head faster than ddrescue (due to its ability to perform passthrough commands that don't do as much or any extra read retries by the OS), but would not likely produce more recovered data. The image shows that for the very first part of the drive, both heads were reading with one having some problem spots. But then a very consistent repetitive pattern starts, and from what I know about ddrescue I would say that one head is not returning any data at all. Without the actual log file it is difficult to know for sure. So assuming my analysis is correct (from only the picture to work from), either there is a major physical or firmware issue shortly after the beginning of the drive causing one head to not read, or the weak/damage head completely died during the recovery attempt.

_________________
http://www.hddsuperclone.com
Home of HDDSuperClone

Hi Maximus,

A link to the log file in case you're interested:
https://www.dropbox.com/s/dqma3zgkz0o3k ... 2.txt?dl=0

Initially I ran DMDE against the drive a few times (for very limited periods) before attempting to image it with DDRescue. Each time the start of the DMDE scans went okay before hitting what felt like damaged areas of the disk. It definitely "feels" like the beginning of the disk is less damaged so your first theory may be correct.

Regarding HDDSuperclone.
Really impressed with the speed so far although skip size is currently 15728, not sure if I should've lowered the max skip size?
To put things in perspective:

DDRescue Pass 2:
17/08/2016 9am ish: Rescued: 341147 iPos: 999653 -34.1147%
22/08/2016 9am ish: Rescued: 341685 iPos: 987437 -34.1685%
23/08/2016 9am ish: Rescued: 341780 iPos: 984341 -34.1780%

HDDSuperclone Phase 2:
Starting from where DDRescue left off and running for 1 day 1 hour:
Position: 1594464033 (around 760.3Gb?)
Finished: 768967400 (3381 areas 39.364010%) <-These numbers were changing as I typed so the maths may not work out perfectly =).

Although I can't be sure how much disk damage played a part (The end of the disk seems far worse than the rest of it going by scan speed) SuperClone definitely seems much better able to handle this damaged disk.

Thanks maximus, the tool looks great so far

That skip size is fine. It self adjusts to adapt to how much data is in each run of the bad head. The initial value of 4096 is conservative to accommodate most drives.

_________________
http://www.hddsuperclone.com
Home of HDDSuperClone

From the log I can confirm that it looks like the drive suddenly totally quit reading from one head very early in the recovery. If the drive can still now read that data from the bad head at the beginning then it would be a curious thing (possible firmware issue). But either way, it would almost certainly require professional hardware tools and a possible head swap to get any more data from the bad head, if it were even possible at this point. Continuing the recovery yourself will at most yield maybe just over 50%.

_________________
http://www.hddsuperclone.com
Home of HDDSuperClone