Ubuntu Data Rescue

March 01, 2020

Ubuntu data rescue

example 1: System does not boot anymore after an OS upgrade (kernel panic). Data is on a LVM.
example 2: System does not boot anymore (HDD errors). Data is on an ext3 partition.

Example 1

System does not boot anymore after an OS upgrade (kernel panic). Data is on a LVM.

Automatic:

Booted Ubuntu 16.04 LTS with a LAN connection active. The LVM was setup properly and recognized perfectly. This was not the case without network access.

In this case it's easy to backup the data and proceed with a re-installation.

Select do something else
Setup /, swap and /boot partitions
Select the place where bootloader will be stored
Take care that no checkbox for format partition is set (otherwise data will be wiped!)

The setup warns against deletion of folders during install (backup!) but should leave /home alone.

After installation recreate the users in the same order as on the original system and/or take care to chown the files appropriately.

Manual steps:

Install LVM2 tools (now you need network access!)

apt install lvm2

Check if partitions are recognized

fdisk -lu

Once installed, run pvscan to scan all disks for physical volume. this to make sure your LVM harddisk is detected by Ubuntu

pvscan

After that run vgscan to scan disks for volume groups

vgscan

Activate all volume groups available.

vgchange -a y

Run lvscan to scan all disks for logical volume. You can see partitions inside the hard disk now active.

lvscan

Mount the partition to any directory you want, usually to /mnt You can access the partition in the /mnt directory and can backup your data

After booting in the new system check that the data is now all ok and call

apt update && apt upgrade

Maybe several times until no error comes (some packages might be incompatible external ones or remain on the system and not get cleaned up correctly).

Example 2

References:

Example 3

Non LVM volume, a simple ext4 partition with disk errors. SMART values look ok but disk access is very slow on some sectors.

Use ddrescue to create an image file out of the partition in two steps. Step 1 creates an image file and a log file to record problematic sectors:

ddrescue -n /dev/sdb disk-image-file.img rescue.log

After this first run there might be problematic sectors which can be retried using re-reads, following command tries to reread max 3 times:

ddrescue -d -r 3 /dev/sdb disk-image-file.img rescue.log

If that fails you can try again but retrimmed, so it tries to reread full sectors:

ddrescue -d --retrim -r 3 /dev/sdb disk-image-file.img rescue.log

How to mount and use the disk image ((use the offset option if you made a full image and want a given partition)):

mount -o loop,ro disk-image-file.img mountpoint
# find out the offset of the partition you want to mount using fdisk
mount -o loop,ro,offset=x disk-image-file.img mountpoint

Repair the partition eventually using the appropriate checker for the partition format:

fsck.ext3 disk-image-file.img
fsck.ext4 disk-image-file.img
...

References: