Having trouble recovering data from disk img

hyjalsoul · 01-19-2011, 03:17 PM

I have 2 RAID1 hard drives with possibly hardware errors, (when I tried to mount them in a degraded array, they won't start, throwing some Buffer I/O errors)
So I used ddrescue to make a disk image out of one drive, ran losetup to use the image file as loopback device:

Code:

losetup /dev/loop0 imagefile.img

but when I tried to assemble a raid array including the /dev/loop0 device like:

Code:

mdadm --assemble /dev/md0 --force /dev/loop0

it will complain that no superblock is found on /dev/loop0 device.

With desperation, I tried to create a legacy raid array with following command, of course, including one of the bad drives:

Code:

mdadm -B /dev/md0 --level=1 -n2 /dev/sdb1(bootable partition on the bad drive) /dev/loop0

I successfully created a new RAID1 array, but when I tried to mount it

Code:

mount /dev/md0 /mnt/tmp

it failed because of bad drive sdb
and after I stopped raid array, the disk image file got contaminated too, because I don't see something like anymore:

Code:

   Device Boot      Start         End      Blocks   Id  System
         /dev/loop0p1   *           1       29664   238267392   83  Linux raid autodetect
         /dev/loop0p2           29664       30402     5928961    5  Linux raid autodetect

it only left me one partition with ext4 system on it

Now the only thing I haven't tried is to clean the superblock by doing --zero-superblock, and not sure if that will solve the problem.

Should I get a second drive to hold my broken drive image so that I might be able to assemble a good RAID1 array or should I continue working on the only disk image file I recovered from one of the broken drives?

kbp · 01-19-2011, 03:30 PM

I would have just tried to recover the data from the disk image, not tried to put it back into the array

xeleema · 01-19-2011, 03:59 PM

Greetingz!

I agree with kbp, pull the data out of the disk image. However, if you don't have enough disk space for that, and want to use the other (known-good) drive from your RAID1 array, then rebuild it with the --missing option.

Example:
mdadm --assemble /dev/md0 --level=1 -n2 missing /dev/sdb1

That should bring up the RAID1 as degraded and let you mount it.

jefro · 01-19-2011, 04:03 PM

I'm thinking that I might have looked at replacing one drive and then re-image it. If fails then swap original drives. If fails then the controller or other part may be bad.

The image may be used in a VM but I doubt the data is any good.

hyjalsoul · 01-19-2011, 04:36 PM

What would be the best way to pull the data out of the image, I suspect because I recovered the image out of a bad drive, the file system might also be corrupted on there. I tried mounting the image as loopback device, it wouldn't let me because it had raid partitions...
I ran something like

Code:

mount -o loop, offest=32256 imagefile.img /mnt/tmp

hyjalsoul · 01-19-2011, 04:48 PM

And the fact is both of my RAID1 drives were in bad state... before I had a chance to back up the data out of it.