opensuse 11.1: after growing a raid1 by one disk, no md1 on boot

opensuse 11.1: after growing a raid1 by one disk, no md1 on boot

[Bernd Rieke]

Hi all,

we made the following steps:

install an opensuse 11.1 as RAID1 on two scsi-disks:

/dev/md0 .... swap (/dev/sda1 /dev/sdb1)
/dev/md1 .... root (/dev/sda2 /dev/sdb2)

Just the install with no modifications, nothing else!

Everything works fine. We rebooted the system several times, it came up
as expected. Then we noticed that we forgot the third disk within the
arrays as we meant to have for more redundancy. So we made a

mdadm --grow -n 3 /dev/md0
mdadm --grow -n 3 /dev/md1
mdadm -a /dev/md0 /dev/sdc1
mdadm -a /dev/md1 /dev/sdc2

The third disk was synced, everything fine again. But on the next reboot
the system stopped with 'waiting for /dev/md1 to appear'.

When the system is installed initially with 3 disks there are no
problems. To produce the problem it needs only the --grow -n 3. After
this the root-partition /dev/md1 seems to be corrupted or not known to 
the kernel for any reasons.

When we try to assemble with the rescue system it tells us that there is
no superblock on /dev/sdx2. And when we try to mount one of the disks
directly it tells us that the filesystem linux-raid-xxx (forgot the xxx,
it was something like 'type', i think) is not a valid filesystem-type.

What goes wrong?

Kernel: 2.6.27.7-9-pae #1 SMP 2008-12-04 18:10:04 +0100

Greetings Bernd Rieke

Re: opensuse 11.1: after growing a raid1 by one disk, no md1 on boot

[Bill Davidsen]

Bernd Rieke wrote:
> Hi all,
>
> we made the following steps:
>
> install an opensuse 11.1 as RAID1 on two scsi-disks:
>
> /dev/md0 .... swap (/dev/sda1 /dev/sdb1)
> /dev/md1 .... root (/dev/sda2 /dev/sdb2)
>
> Just the install with no modifications, nothing else!
>
> Everything works fine. We rebooted the system several times, it came up
> as expected. Then we noticed that we forgot the third disk within the
> arrays as we meant to have for more redundancy. So we made a
>
> mdadm --grow -n 3 /dev/md0
> mdadm --grow -n 3 /dev/md1
> mdadm -a /dev/md0 /dev/sdc1
> mdadm -a /dev/md1 /dev/sdc2
>
> The third disk was synced, everything fine again. But on the next reboot
> the system stopped with 'waiting for /dev/md1 to appear'.
>
> When the system is installed initially with 3 disks there are no
> problems. To produce the problem it needs only the --grow -n 3. After
> this the root-partition /dev/md1 seems to be corrupted or not known to 
> the kernel for any reasons.
>
> When we try to assemble with the rescue system it tells us that there is
> no superblock on /dev/sdx2. And when we try to mount one of the disks
> directly it tells us that the filesystem linux-raid-xxx (forgot the xxx,
> it was something like 'type', i think) is not a valid filesystem-type.
>
> What goes wrong?
>
> Kernel: 2.6.27.7-9-pae #1 SMP 2008-12-04 18:10:04 +0100

I'm guessing that this would be cured by creating a new initrd file, 
because I but the mdadm.conf file used device names instead of 
PARTITIONS and a UUID. Mind you, guessing, but look at /etc/mdadm.conf 
and see if in fact this is the case after install.

If you can get up on a recovery CD, check the mdadm.conf, fix if needed, 
run mkinitrd and reboot, assuming SuSE gave you the right tools.

If you are a wizard you can boot from a recovery CD, unpack the initrd 
file somewhere, patch the mdadm.conf, recompress, save the initrd file, 
and everything will be fine. I have done this before, so if you find 
that this is your problem and you understand the outline of steps I 
mentioned, go to it. I'm not about to try details, if I forget one you lose.

-- 
Bill Davidsen <davidsen@tmr.com>
  "Woe unto the statesman who makes war without a reason that will still
  be valid when the war is over..." Otto von Bismark