Re: how to recover filesystem after clobbering array?

["Vasco Névoa" <vasco.nevoa@sapo.pt>]

I've successfully recovered my array data. :)
All it took was, as I expected, to rebuild the right pointers. The data 
was always untouched.
Here it goes, for the sake of completeness.

I initially used "mdadm --create --assume-clean ..." on a level1 array 
of 2 disks that came from another machine. I didn't know any other way 
of starting an array in this situation. While this is an acceptable 
practice in some cases, it is better to use "--assemble" and pass the 
necessary info (like uuid).

Unfortunately I apparently lost all the data because the newly created 
metadata superblock was the default 0.9 version, and the original array 
metadata version was 1.2, and this resulted in a missing partition table 
once the array was run.

Then I retrieved the mdadm.conf file from the original machine, and 
there I found the correct metadata version, uuid, array name, array 
device name. I also double-checked the volume partition and filesystem 
type from that machine's /etc/fstab.

So, I zeroed-out the metadata superblocks with "mdadm --zero-superblock 
...", and then proceeded to restore the original metadata superblock 
with "mdadm --create --assume-clean --metadata=1.2 --uuid=... 
--name=...  --level=1 --raid-devices=2 /dev/sde1 missing", and it worked 
just fine. The array was started and I could see the original data 
partition with fdisk and actually mount it. After this successful test, 
I just added the other disk to the array.

The one information I expected from this list was: "No problem, your 
data is still there. If you can recreate the superblock with the same 
metadata version as it used to be, everything reverts to normal." 
Unfortunately I had no such support.

Cheers,
Vasco.

On 19-07-2011 12:01, Vasco Névoa wrote:
> Thank you very much, Tyler and David.
> Good advice.
>
> The array is level 1, built upon primary partitions of the devices.
> The file system is EXT4.
> The mdadm command that I stupidly used to create instead of starting 
> the array included "--assume-clean" but no "--build".
> I checked via /proc/mdadm that the array came up without syncing 
> anything (or at least it was ultra-fast, less than 5 seconds).
> So I firmly believe the data is all there, on both disks.
> I just need to rebuild the metadata to point to the data again somehow.
> Right?...
>
> Citando "Tyler J. Wagner" <tyler@tolaris.com>:
>
>> On 2011-07-19 11:17, David Brown wrote:
>>> Once you have got image files for each of your disks, make copies of
>>> these image files to another spare disk. Keep careful notes of exactly
>>> what you have done here, and which files are which. And put your
>>> original disks, carefully labelled, on a shelf somewhere.
>>>
>>> Now you are in a position to attempt data recovery on your copied
>>> files. If you do something wrong, you can simply re-copy the image
>>> files and try again. You still have absolutely no guarantees that
>>> you'll get anything back - but at least you can be sure you are not
>>> going to make anything worse.
>>
>> Follow David's advice.
>>
>> What was the filesystem on the array?
>>
>> Now, use testdisk, photorec, and foremost to seek through the raw images
>> and extract files. The good news is, most video formats are detectable
>> by these tools. All can be installed with your package manager.
>>
>> Regards,
>> Tyler
>>
>> -- 
>> "Offending fundamentalists isn't my goal - but if it is an inevitable
>> side-effect of defending human rights, so be it."
>>    -- Johann Hari
>>
>> -- 
>> To unsubscribe from this list: send the line "unsubscribe linux-raid" in
>> the body of a message to majordomo@vger.kernel.org
>> More majordomo info at  http://vger.kernel.org/majordomo-info.html
>>
>
>

--
To unsubscribe from this list: send the line "unsubscribe linux-raid" in
the body of a message to majordomo@vger.kernel.org
More majordomo info at  http://vger.kernel.org/majordomo-info.html