My home server hard drive started to have problems (bad sectors, noisy, etc..)
I decided it was time to reinstall CentOS before it was too late
I had some important data on the old drive, I would recover the data by just pluging it when the new system would be installed on the new drive.
So, I installed CentOS 4.4 and partitioned the new drive pretty much like the old one..
This is my partitioning scheme :
/boot /dev/hda1 / /dev/hda2 /backup /dev/hda3 /tmp /dev/hda5 swap /dev/hda6
After the installation, the system boots fine, so I decide to stop the machine, plug the old drive as primary slave and retrieve the data from it.
At reboot, I get some errors (mounted as read-only, already mounted, blah blah) then the system hangs with this message : “kernel panic - not syncing : Attempted to kill init”
Actually, the system tries to mount the partitions by calling their label, and not the device like they used to do in the past :
/etc/fstab by default looks like this :
LABEL=/ / ext3 defaults 1 1
The issue here is the system was finding two partitions named with the same label !
I rebooted the machine in runlevel 1 and edited /etc/fstab like this :
/dev/hda2 / ext3 defaults 1 1
I also had to edit /etc/grub.conf and change the “root=” argument from “root=LABEL=/” to “root=/dev/hda2”
Now, the system calls the device and boots fine, I can manually mount the old partitions and start retrieving my data.
I’ll switch back to the label thing when the data have been transfered.
If anyone has a hint on why they prefer calling by the drive labels over devices, please drop me a comment..