Return to

How to back up a failing hard drive?

So I have this 4TB Toshiba disk in my home server for a while now. There are some home video in it, which are sync’ed to my backblaze account, at least I thought so. Recently I noticed that no matter what I copy from it, the speed always drops to zero.

So finally I restarted the server, and when it booted up, it says hard drive would fail and must back up immediately. And it turns out some of the videos are not synced to backblaze.

My question is how do I back up the data on this drive? Right now no matter what I copy from it, it just hangs. And any program that touches the drive (like TeraCopy or system error checking) takes a long time to exit.

Thanks a lot in advance!

You may want to try TestDisk:
Afaik it has a file “recovery” function.

1 Like

ddrescue works well. First pass should read the easy parts and skip big sections of unreadable parts. The other passes will try to fill in any readable bits around the bad sections. Command-line assumes the failing drive is the second (sdb) drive in the system and you have enough space for a copy of the full 4TB:

ddrescue -nN -a 500KiB -K10MiB -n /dev/sdb sdb.raw sdb.log
ddrescue -d -M /dev/sdb sdb.raw sdb.log
ddrescue -r1 /dev/sdb sdb.raw sdb.log

heat may be an issue with this drive. one trick i use for recovery of data on dodgy drives is to put the drive in a ziplock bag (power and data cables attached ) and place the bagged drive between a pair of ice packs, wait about 20 minutes before powering it up.
run TestDisk or other recovery software.
ddrescue is very good software.

1 Like

I have a big silica gel pack I throw in aswell.


Thank you, looking into it

Thank you. Trying it now. Had to create a Ubuntu Live USB and run from there.

Make an image of the disk and use testdisk on copies of the image. Don’t try to do anything but image the disk with ddrescue.


Is there some switches to make ddrescue skip bad sectors on a first run, with a log file, to get most stuff off, in one day.
then another run with the log to get as much as possible, even if it takes a week or whatever?

Or does it just try hard to get what it can, then end?

I believe what @rcxb posted is probably the way to go.

Typically you’re in a race against time to get what you can off the bad disk. Its a matter of just hammering out whats readable first and then going back for the rest later.

I would personally avoid things like the freezer trick as thats a fix for a very specific issue and is often a death sentence for the drive later.

1 Like

Um, well, that’s embarrassing. I guess there is the info Just a couple of posts before.
I am going to blame mobile scrolling, and pretend I didn’t say nufffin.