Link to home
Start Free TrialLog in
Avatar of mstrelan
mstrelanFlag for Australia

asked on

Ext3 partition won't mount

Hi experts,

I have a server running Ubuntu 10.04 server, which is virtualizing another Ubuntu 10.04 server with qemu / kvm. For some reason somebody yesterday decided to press the power button on the physical server. The virtual server never came back up.

I opened up virt-manager and had a look at my virtual machine. It had just a blinking cursor. I had previously thought maybe it was fsck'ing but it didn't appear to be. I pressed a few buttons on the keyboard, trying to get some input, and all of a sudden it showed a message saying something like "skipping /backup" which is the partition that has all of my files on it. Ubuntu loaded but /dev/sdb was not mounted at /backup. I have not been able to mount it, it gave an error that I can't remember but it may have been about a superblock.

Anyway I'm analysing the partition with testdisk at the moment, but I don't really know what to do from here. Any suggestions? Am I screwed? I have some of the data backed up on an external drive, so that's a bit of a relief. The partition is just a .qcow2 file as well, so maybe there is something I can do from the physical server?

Avatar of torimar
Flag of Germany image

Link to home
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Avatar of mstrelan


"To search for them, run TestDisk and in the Advanced menu, select the partition and choose Superblock."

No partition available
How did you proceed? Did you boot the VM off a bootable CD?
I recommend Parted Magic for the purpose:

Then follow this guide in order to check whether the partition table may be restored:
The actual data image is on a separate qcow2 image than the operating system, so assumably it is safe to do any fsck / testdisk etc type of stuff without a live disc because the data partition is not mounted.

I have tried to "analyse" in test disk but it did not find any partitions. I am going to try and mount the image directly in the host OS.
UPDATE - Good news. On the host OS I was able to mount the disk image and access all my files! Yay!

qemu-nbd -c /dev/nbd0 data.qcow2
partprobe /dev/nbd0
mount /dev/nbd0p1 /mnt/data/
ls -l /mnt/data

Open in new window

Now I just need to know how to get it to mount in the guest OS. I will try the e2fsck with alternate superblock, but I want to take a backup of the disk image first.
That is really good news indeed.

For a backup, I assume it will suffice to copy the qcow2 file to a safe backup location, the continue to work on the original.
Link to home
This solution is only available to members.
To access this solution, you must be a member of Experts Exchange.
Start Free Trial
Thanks torimar, the format / refill method will be my plan b if e2fsck doesn't work. I'm off to buy a bigger hard drive to back up the data. Will be back later to post my results and award points.
Ok I wasn't able to fix the disk with testdisk or anything, so the best solution is to mount the image in the host os and scp all the files to a new partition. Thanks torimar for the help