I recently had a hard disk go bad. It had a single NTFS partition where the partition was no longer recognized by Windows or Linux.
I managed to recover (I think) all of my files by reading the data off of the disk using TestDisk. It discovered my partition without issue, and the actual file recovery went relatively smoothly.
I believe the way TestDisk was primarily designed to work was by attempting to actually repair the partition. This should be fine if you’re working on a copy of the raw disk bits (you can probably use dd for Windows for this, or write a program that accesses ), but I’d generally be averse to making any further changes to a bad disk. I chose to recover my data just by copying the files directly out. You can do this as follows:
- go to (your disk) > Intel > Analyse > (your dynamic partition)
- highlight your partition
- press P to list files
- press H to hide deleted files
- press C to copy the current directory (the directory from which you started TestDisk)
One thing to watch out for when using this approach is to first make sure you press H to hide deleted files. Otherwise, you end up recovering deleted files as well, and sometimes these deleted files (for whatever reason—probably because my data’s sufficiently corrupted) end up pointing back into the root of the file system, leading to cycles in the tree and thus an infinitely recursive file copying procedure.
Follow me on Twitter for stuff far more interesting than what I blog.