A disk with a Chunk Server repository failed in RAID. After replacing the disk, the server does not boot with the following error on file system check stage:
Superblock could not be read or does not describe a correct ext2 filesystem
All file systems are mounted in read-only mode.
How do I repair the server from this state?
Remount the root partition in rw mode and edit fstab to disable the failed disk from being mounted:
~# mount -o remount,rw / ~# vi /etc/fstab ~# grep /dev/cciss/c0d5p1 /etc/fstab #/dev/cciss/c0d5p1 /pstorage/pcs1-cs5 ext4 defaults,noatime 1 2
Reboot the server and recreate ext4 partition on this disk:
~# fdisk /dev/cciss/c0d5 d -> n -> p -> 1 -> enter -> enter -> w ~# mkfs.ext4 /dev/cciss/c0d5p1
Delete the chunk server from cluster:
~# pstorage -c pcs1 rm-cs 1031 -W
Uncomment the line in /etc/fstab for this disk and reboot the server.
Create a new chunk server on the disk:
~# rm -rf /pstorage/pcs1-cs5/* ~# pstorage -c pcs1 make-cs -r /pstorage/pcs1-cs5 -t 0
NOTE: It is advised to avoid using RAID and export all disks separately for Chunk Servers. Such configuration allows online replacement of failed disks without the need to reboot the server. This is made possible with
pstorage-hotplug utility (available starting from PCS 6.0 update 4)