ddrescue making no progress on splitting bad blocks

Where are these bad block located?
I mean, saying that + means "good" and - means "bad", is it a "pattern", like
+++++—–+++++—–+++++——
or are they all at the beginning or the end (or in the middle) , like
+++++++++++++——————–
———————–++++++++++++
+++++++—————–++++++++
or are they "random" (or with no apparent pattern), like
++-+++–+–++++–+++—–+-+-++

What was the original issue (like "disk dropped" or "stopped working suddenly") etc.?

What exact make/model is the hard disk?

The typical causes can be

• head crash (this would normally create a "pattern")
• failed head (on a multi platter disk this would make a whole side or a whole platter unreadable)
• *something else* (like - as an example - a corrupted translation table)

If the bad sectors are 1 Tb (i.e. 1/3 of the whole disk) it is likely to be a failed head (but as said this may be confirmed by the actual location of the bad sectors and on the number of platter/heads the disk has).

jaclaz

Well, I looked at the log fie, and I dont find a very regular pattern. Seems that the second part has more errors than the first one. The harddrive is still getting recognized, I was able to even get off some files, but there are sections that cannot be acessed anymore. So I thought with an image I have more chances to recover some files than from a damaged drive that gives me constantly read errors. But yes, the imaging was only sucessful unto this point…

I dont know, if it is useful I can send you the logfile. But it seems to me it is not that much of a regular pattern. If you have some more hints, let me know.

I would suggest that after 20 days the drive is now dead.

When trying to recover data it is important not to stress the drive with too many retries. 20 days of retries can often mean that failing heads end up scratching the tracks.

Your position now is to recover what you can from the partial image you have.

You need a good data recovery program that can work with a partial image.

I have never used ddrescue but are the sectors on the disk you copied to in the correct location? Also, was the disk clean before you started the copy? If the disk was not clean, you will have major problems with 'data contamination'

You may be luck and get some files back logically, but otherwise you may need to use data carving and hope your data is not too fragmented.

FYI -the approach I would have taken with a drive like yours is

1) establish if it requires physical repair, eg new heads. Expensive but can be more reliable

2) If software approach taken, I always build up an image in sections. Some parts of the disk typically read better than others. Have an educated guess as to where the data will be and try and get 90% of the useful part of the drive imaged. At that stage you can afford to stress the drive more to hopefully get the final 10%.

Many disks have partial failures on the directory (MFT / CAT) areas. These are often the most useful sectors, but also the hardest to recover. However, they don't contain user data which can sometimes be recovered by other means.

Sure, recovering from an image residing on a surely good disk is much better/faster.

If there is no "pattern" or "whole zones" not readable it is likely to be the *something* else.

Almost *anything* (and the contrary of it 😯 ) may be the actual cause, including intermittent (and/or temperature connected) problems in the electronics.

I doubt you can do much more than let the drive cool completely and try again (sometimes a not-warmed up disk allows to read a bunch of sectors more), all the rest would need a pro examining and diagnosing properly the issue (and if possible repair the problem), though if it is a head or a translator this may be possible, if it is the actual platter surface the data is gone for good.

Which make/model is the disk?

jaclaz

I actually tried now to do a separate image of just the first 5 GB. It works fine in the first 2 phases of ddrescue, but as soon as it comes to the splitting phase, it is stuck again (Looks like the drive is not totally dead at least). Has within 5 GB some 800MB failed sections, and cannot split, not a single one. I tried already to let the drive cool down and start again, but no change. For your info, it was running 20 days recuperating with success, but then in the splitting phase 2-3 days stuck with no progress at all. Any ddrescue specialist can give me a tip on that?

Having had a quick look at the ddrescue manual splitting should be the final stage.

What you should have is a nearly complete image. Can you try data recovery on this image? Most good data recovery programs will skip, or improvise on missing sections.

Can you answer this simple question?

Which make/model is the disk?

Explanation a given make model of disk may be more subject than another to a given "typical" issue, and maybe there could be specific things to try (that may work on a given make/model but not on another one).

jaclaz

I have a Seagate Barracuda 3000GB Model ST3000DM001
By the way I mirrored to an identical drive, I could try swopping the controller board, if that could be of any help. But not sure if it would..

Anyways, there is still 1 TB that is not recovered, I wonder how we can improve on that value.

Swapping controller boards very rarely works. Most drives store drive specific parameters on the board, and so it may be necessary to also swap (ie unsolder) a ROM.

The most likely reason for your problem is failing heads. One typical symptom of this is very slow reading (due to retries).

For the failing 1TB - can you determine on which sectors your critical files are on? Can these sectors be read with a sector viewer? If they cannot be read singly, they will not help building up an image.