lcoughey
Well-Known Member
- Reaction score
- 2,341
- Location
- Ontario, Canada
There is only one software program that I would consider using for cloning a drive with read errors and that is GNU ddrescue. All the other software cloning programs read in blocks and don't have the ability to log and re-clone the unread sectors only.
For me, we clone in this way using our DeepSpar Imagers:
Disable SMART and auto sector remap (if possible)
Generate a head map
Run a head/media test
Disable questionable heads
Pass 1 - Read timeout 100ms, on read timeout, jump to next head
Pass 2 - Read timeout 100ms, try to read unread sectors
Pass 3 - Read timeout 1000ms, PIO mode, reverse, attempt to read what was missed
Repeat the process with all heads enabled
If we can and it makes sense, we'll jump into DeepSpar's DRE, read the file system and tell the drive to only clone the sectors associated with the targeted files and folders.
What most software programs do is read in a block (ie 256 sectors). If any of the sectors does not read, it skips the entire block and moves on. There is no way to go back and complete the clone after the process is done.
What ddrescue does is goes forward in blocks and skips a block on read errors and goes back and fills in the holes on a second pass. With the use of a log file, you can keep digging at the drive, only reading the unread sectors until you get the best clone possible. This program is not simple to use and if you make a mistake, the damage can be severe.
