Software RAID Recovered

FizzerJE · 09-06-2012, 10:04 AM

Hi, I'll try to keep it brief..

Today had a kernel panic. Dirty degraded RAID array where the OS is installed.

System is on a software raid 5 partition with encryption and LVM2.

OS install was CentOS 6.2.

Booting into LiveCD 6.3 I have managed to access the two main Logical Volumes I need that holds the data most valued to me (home,scratch).

Code:

mdadm --create /dev/md3 --chunk=64 --assume-clean --metadata=0.9 --level=raid5 --raid-devices=3 /dev/sda3 /dev/sdg3 /dev/sdh3
vgscan
vgchange -ay
<VG_System now available 7 logical partitions>
cryptsetup luksOpen /dev/mapper/VG_System-<scratch/home> crypt-<scratch/home>

cryptsetup luksOpen does not work for var and root etc.

I believe this is due to when I was trying to fix the array, I add a disk and it started to re-build with the incorrect chunk size.

The superblocks then read as a newer Device with the incorrect size of 512K.

Luckly re-creating the raid in the correct device order and chunk size has allowed me to open the two partitions I need.

I now need to reinstall the OS.

Running the installer (CentOS LiveCD 6.3) the complete VG_System partitions from the LVM are available to use.

BUT

both 'scratch' and 'home' are showing as unrecognised file system types....

Advice on getting the OS back on??

I was tempted to use the LVM as is.
Formatting / var etc. and install BUT leaving 'home' and 'scratch' alone.

Thought on this??

ReaperX7 · 09-06-2012, 12:49 PM

If you still have the disk parity intact try removing the new volume you attempted to regenerate, and regenerate it again. You MAY be required to track down a 6.2 disk because things may have changed between 6.2's Live CD and 6.3's.

Just a good question but do you have workable backups of the data?