Convert RAID1 to standard ext2

Convert RAID1 to standard ext2

[Timothy Weaver]

Hi,

I had 2x750GB drives in a RAID1 configuration in a D-Link NAS
(DNS-323) device with embedded Linux as an ext2 volume. The
configuration was corrupted and the NAS no longer saw the RAID.

I put the drives in another Linux box and was able to use mdadm to
scan the drives. It recognized the RAID1 configuration and created the
RAID device at /dev/md0. Unfortunately, it says it is corrupt and
cannot be mounted. I tried using e2fsck / fsck but it says the
superblock is corrupt. Trying to use copies of the superblock were
unsuccessful.

I am confident the data is still there and want to get to it. Is there
a way to take one of the drives and convert it from being in the RAID1
set to just a standard ext2 partition in a non-destructive way? I
figured this should be possible with the second drive just to be sure
not to destroy both copies of the data.

Thanks.

Re: Convert RAID1 to standard ext2

[Bill Davidsen]

Timothy Weaver wrote:
> Hi,
>
> I had 2x750GB drives in a RAID1 configuration in a D-Link NAS
> (DNS-323) device with embedded Linux as an ext2 volume. The
> configuration was corrupted and the NAS no longer saw the RAID.
>
> I put the drives in another Linux box and was able to use mdadm to
> scan the drives. It recognized the RAID1 configuration and created the
> RAID device at /dev/md0. Unfortunately, it says it is corrupt and
> cannot be mounted. I tried using e2fsck / fsck but it says the
> superblock is corrupt. Trying to use copies of the superblock were
> unsuccessful.
>
> I am confident the data is still there and want to get to it. Is there
> a way to take one of the drives and convert it from being in the RAID1
> set to just a standard ext2 partition in a non-destructive way? I
> figured this should be possible with the second drive just to be sure
> not to destroy both copies of the data.

Try a read-only loopback mount of the partition (either). However, I 
think you're missing something else, although I don't have a clue what. 
Unless the O/S started writing bad data or the hardware got sick, you 
should be able to recover. In any case this allows you to do something 
non-destructive and use offset= depending on the superblock location.

-- 
bill davidsen <davidsen@tmr.com>
  CTO TMR Associates, Inc
  Doing interesting things with small computers since 1979