[SOLVED] LVM issue

T3RM1NVT0R · 05-04-2015, 02:34 PM

I have an idea but not sure about the success rate. Instead of running e2fsck to reconstruct the superblock using backup superblock why can't we try mounting LV with backup superblock?

Basically this is what you have to do, perform pvcreate, vgcfgrestore and vgchange as you did yesterday. To find the superblock run dumpe2fs:

Code:

dumpe2fs /dev/mapper/VolGroup00-LogVol02 | less

scroll down where it shows backup superblock.

Now try mounting it as follows:

Code:

mount -t ext3 -sb <backup-superblock> /dev/mapper/VolGroup00-LogVol02 /newdisk

I thought it is worth giving a try before you run e2fsck on the disk.

Note: You will have multiple backup superblock try couple of them and see if you get any result

rknichols · 05-04-2015, 02:53 PM

Unfortunately, dumpe2fs will fail because the primary super block is missing, but testdisk will report the backup super block locations if you run the "Advanced" (Filesystem utils) action and then select "Superblock". That number will be in filesystem blocksize units, so it needs to be multiplied by 4 (typically) to convert to the 1K block size units needed by the mount command.

I just gave that a try, but could not mount the filesystem if the root directory was overwritten.

T3RM1NVT0R · 05-04-2015, 03:28 PM

Yes you are right. What about mke2fs -n, that will give an idea where backup superblocks are. This is something which has to be tried with caution, if command mke2fs run without -n switch it will end up formatting the partition.

rknichols · 05-04-2015, 03:59 PM

I even tried mkfs.ext3 with the "-S" and "-b 4096" options to rewrite the super blocks and group descriptors, and I still didn't get a filesystem that I could mount. It makes sense. The kernel can't do much with a filesystem that lacks its root directory. The only way I got anything recoverable was by letting fsck.ext3 "do its thing".

vpp · 05-04-2015, 04:28 PM

Thanks a lot for all these hints! I will try them shortly.

vpp · 05-04-2015, 04:37 PM

I started by doing a deep search with Testdisk. Still at the beginning with following results.
ext4 surprised a little.

vpp · 05-04-2015, 05:12 PM

I also went through all backup super blocks recommended by mke2fs -n with e2fsck -n -b, but all checks reported just
e2fsck: Bad magic number in super-block while trying to open /dev/mapper/VolGroup00-LogVol02

The superblock could not be read or does not describe a correct ext2
filesystem. If the device is valid and it really contains an ext2
filesystem (and not swap or ufs or something else), then the superblock
is corrupt, and you might try running e2fsck with an alternate superblock:
e2fsck -b 8193 <device>

rknichols · 05-04-2015, 05:14 PM

That looks like the result you get if you choose the "Analyze" (Analyse current partition structure ...) function rather than the "Advanced" (Filesystem Utils) function. Filesystem Utils should be essentially instantaneous in detecting the ext3 filesystem.

rknichols · 05-04-2015, 05:17 PM

Quote:

Originally Posted by vpp

I also went through all backup super blocks recommended by mke2fs -n with e2fsck -n -b, but all checks reported just
e2fsck: Bad magic number in super-block while trying to open /dev/mapper/VolGroup00-LogVol02

Did you multiply those numbers by 4? mke2fs reports in fs block size units (typically 4K). The mount command wants 1K block size.

[EDIT] Oops, sorry. You were talking about e2fsck, not mount. Really, e2fsck should find the backup super blocks on its own.

vpp · 05-04-2015, 05:19 PM

Did I need to multiply the reported backup locations also for e2fsck command?

vpp · 05-04-2015, 05:21 PM

Oops sorry, I didn't read your message till the end...

vpp · 05-04-2015, 05:27 PM

I am ready to give the final command for this problematic hard disk. After getting whatever out of it, I will then initialize it for traditional use.

rknichols · 05-04-2015, 05:31 PM

That's about where I am, too. Something other than the simple removal of the LVM structure went on. What's left doesn't seem recoverable. Sorry things didn't work out better.

vpp · 05-04-2015, 05:33 PM

Anyway, thanks to all of you, you are invaluable!

T3RM1NVT0R · 05-04-2015, 06:03 PM

You're welcome.

Thanks to you too for all the testing and rknichols for fantastic discussion!