PDA

View Full Version : [ubuntu] mdadm booting: md stopped


kriebel
July 24th, 2008, 05:02 AM
Guru's

I happen to have a typical problem with mdadm. I made a dump of my original (Ubuntu 8.04 server) system with dump. The original system had the following RAID1 configuration:

md0 /dev/sda1
/dev/sdb1

and

md1 /dev/sda2
/dev/sdb2

I destroyed the original system and booted with sysrescd. I recreated the devices with fdisk en created the RAID array including a mkfs.ext3. Cylinders of the partitions are the same on both disks. The arrays sync without problems (as far as I can see). This all looks good when I do a cat /proc/mdstat. After that I restored the original system with restore on /dev/md0. After the restore I can mount /dev/md0 and a fsck detects no problems. Reinstalled grub without errors. Also changes the UUID number in /etc/mdadm/mdadm.conf on the restored disk with mdadm --detail --scan >> /mnt/custom/etc/mdadm/mdadm.conf.

When booting the system the problem occurs. Grub starts and the system starts to boot. Disk are detected (Attached scsi etc) but right after that I get 4 messages saying: md: md0 stopped and md: md1 stopped.

Then I get Busybox which tells me:

Check root= bootarg cat /proc/cmdline or missing modules, devices: cat /proc/modules ls /dev

ALERT! /dev/md0 does not exist

md0 and md1 do exist in /dev and cat /proc/cmdline gives root=/dev/md0 ro

I'm probably missing a step in the procedure but I can't figure out which one. Any experts around here who have an idea? Would be very much appreciated ](*,)





p.s. As you can see I'm testing if mdadm is suitable for me in my production environmnts and a disaster recovery is an important issue. Im testing this in a virtualmachine inside Virtualbox but I don't expect it to be an issue in this situation