Recover only used part of failing hard drive

[maximus]

I fix computers on the side, mostly for friends, and on occasion I have the need to recover data from a failing hard drive. I have used ddrescue in the past, but that is getting to be more of a problem as hard drives are getting bigger. I might get a 500gb or bigger drive that only needs 20gb recovered. I am looking for a way in LINUX (not windows) to do a fast copy of chosen files/folders while ignoring errors and skipping damaged files, and also creating a log of the files (with full path) that did not copy so that I can make a determination about further attempts to recover with other software. I say linux because we all know how much windows likes any sort of failing drive plugged into it. Does such software or commands in linux exist?

[dick]

You might want to take a look at Dmde and R-studio.
They both have Linux versions.

There would be recovery limitations with the demo versions. I think it would be single files of any size with Dmde and only very small files with R studio.

I would try them both out on an old failing drive where you have already recovered the data in the past. This is because when a drive is failing it is likely to fail even faster when the heads have to continuously seek and copy the data from various locations rather than sequential sectors whilst imaging.

[maximus]

The paid version of those would probably work, but I am really looking for freeware for as often as I have to do it.

The basic cp command in linux actually more or less does what I am looking for, but painfully slow. In my tests I used the gui drag and drop to copy files from a damaged cd, and when it hit the first error it gave me the option to autoskip bad files. It took 40 seconds to copy 600mb and skip 3 bad files. The cp command took 5 minutes or so. If the gui could provide a log file and the option to autoskip from the beginning it would be perfect (forgot which ubuntu I was using for that).

I also tried unstoppable copier, but that is even slower and sounds like it is hammering the cd drive. I found ultracopier which looks like it would do what I want, but I can't get it to work on any of my systems. Sigh....

And I do have an old failing hard drive to test on. But until I find something worth testing I am using a scratched cd, very easy to make those

[lcoughey]

If the data is important enough to recover, always do the clone first...even if you are only looking for 20GB of data on a 500GB drive. Here are a few reasons:

• You can clone at faster speeds than with a file level copy
• A file copy will cause the heads to thrash all over the place...the constant access to the MFT is enough wear to cause a failing drive to completely crash
• When using a clone app like ddrescue, it can better deal with "bad sectors" than a file level copy
• When using a good clone tool with a log file, you need only ever read a sector from the bad drive once. As mentioned above, if you attempt to access the file system, you can find yourself re-reading the same MFT records over and over and over again as you attempt to copy the files off
• Cloning prevents the heads from thrashing (or at least minimizes it) simply going from sector 0 to the end of the drive.
• By accessing the file system directly, you are assuming that some there are no issues within the file table itself. By getting the cleanest clone possible, you can always scan the clone with a data recovery application and find the lost files

I'm sure my list is not complete...and I'm sure that there are those will disagree.

[maximus]

I do understand that cloning would be the best way to recover all important data. However, in a few instances a fast attempt recovery can be a good first start when the data isn't critically important. I explain the pros and cons and cost (sometimes free for friends, the cost is my time plus hardware if needed) and let them decide how important their data is. I am just looking for the best first fast attempt. It also has something to do with me not having drive space for large drive clones. I do this on the side, and am on a budget. I may not find what I am looking for, but at least I try.

I did manage to get ultracopier to work, but it also seems slow. The fastest one so far was Konquerer file explorer on backtrack 4r2. But with no failed file list, and also only tried on a scratched cd and not a hard drive. The ideal software would read the file table for the files needed to be copied first, and then go copy without having to read the file table again. It would seem that Konquerer might do that with a cd at least, for as fast as it goes. Too much testing and so little time...