[mdlug] Recovering damaged EXT2

[Aaron Kulkis]

David McMillan wrote:
> Jay Nugent wrote:
>> Greetings,
>>
>> On Tue, 15 Sep 2009, David McMillan wrote:
>>
>>   
>>> 	So, one of the two big USB drives I've been using as a poor man's 
>>> RAID-1 (using periodic rsync) died, and I bought a replacement and 
>>> immediately began a fresh rsync to copy the remaining drive to the new 
>>> one.  No big deal... except that, thanks to a dangling cord and careless 
>>> footwork (my own), the remaining drive of the original pair fell off the 
>>> table and hit the floor.  It *seems* okay -- no weird noises, parted 
>>> recognizes its filesystem (ext2) and size, but....
>>> 	Now, when I try to mount it, I get a "bad superblock" error.  If I try 
>>> to run fsck or e2fsck, I get "the superblock is corrupt" and a 
>>> suggestion to run e2fsck with the "e2fsck -b 8192 <device".  But if I 
>>> try that, I get a "device busy" error message, even though the partition 
>>> is not mounted and no other process is accessing that drive.
>>>
>>> 	So... am I screwed?  Or are there tools that might allow me to recover 
>>> what was on this drive?
>>>     
>>    Just in case the platter is still readable (or partially readable), you 
>> *CAN* have the filesystem use one of the backup Superblocks since the 
>> master Superblock indicates that it is corrupted.  So try this...
>>
>>    As root...
>>       mke2fs -n /dev/hda1   <---- or whatever you device is
>>
>>    The -n flag causes mke2fs to NOT actually create a filesystem, but
>> display what it would do if it were to create a filesystem.
>>
>>   
>     Well, this part worked...
>>    This will tell you what number to plug into the 'e2fsck -b <8192> 
>> device' command you mentioned above.  You will now have a list of the 
>> locations of ALL the copies of the Superblock that should reside on that 
>> partition.   Once you plug one of those into the e2fsck command it will 
>> 'correct' the corrpted Superblock (if it indeed can still write to the 
>> platter).
>>   
>     But this just produced more "device busy" errors.  I worked my way 
> through every superblock that mke2fs reported, but no dice.  Looks like 
> I'm SOL.

At this point, you need to look at obtaining another drive
of the same make and model number, and then transferring the
platters to that drive chassis.

If that doesn't work, then you need to hire a forensic data
recovery service.