Bug#570516: md UUID changed

[Steve McIntyre]

On Mon, Feb 21, 2011 at 03:12:43PM +0100, Martin Krafft wrote:
>also sprach Steve McIntyre <steve at einval.com> [2011.02.21.1259 +0100]:
>> However, now when I use the 2.6.28 system I get similar problems.
>> Previously-working devices are now not working. I'm seeing
>> complaints all over the place that homehost definitions don't
>> match when trying to assemble devices. *I* did not change anything
>> here, so it suggests that something (the kernel RAID layer? mdadm
>> init scripts in the initramfs?) has modified my superblocks in
>> a broken way and stopped my system booting.
>
>As the bug report (and Neil) says, this could be related to
>homehost, but I cannot quite figure out how this would happen.
>
>It is true that mdadm on Debian was using --auto-update-homehost up
>until 3.0-1, because IIRC that was the only way to provide
>a migration path from before-homehost to then.
>
>Why the UUID would now change is complete outside of my knowledge
>and imagination. I have never seen this problem.

I'm surprised too - I've never seen anything like this before either,
and I've been using md for many years on this and other systems.

>> Since then, I've tried to force update of the homehost/uuid
>> settings in the superblocks, but to no avail. Then I saw that
>> I was on the old superblock version that didn't store the homehost
>> itself.
>
>They do store them, but as part of the UUID. However, without
>--auto-update-homehost, I do not see a way in which the UUID should
>be updated.
>
>> I've only managed to get things up and running by actually
>> recreating the RAID1 devices by hand, using the same settings as
>> the previous devices. Scary stuff... :-( Yet if I reboot into the
>> new kernel again, things fall apart again. Even on a newly-created
>> v1.2 device.
>
>Try this instead:
>
>In initramfs, remove the /etc/mdadm/mdadm.conf file and replace it
>with a scan:
>
>  mdadm -Escpartitions > /etc/mdadm/mdadm.conf
>
>then -A (assemble) your arrays, either individually or with -Asayes.
>
>Then boot your system.
>
>Once the system is up, compare the output of
>
>  /usr/share/mdadm/mkconf
>
>with the contents of /etc/mdadm/mdadm.conf and update the file with
>the data from the command output.
>
>When done, run
>
>  update-initramfs -u
>
>and now the system should boot.

Hmmm, OK. I'll try that tonight when I'm in front of the system again.

>This bug has been puzzling me all along, which is why I have not
>been able to fix it. I am sorry it caused you grief.

Ack. If I can help debug in any way, just let me know.

-- 
Steve McIntyre, Cambridge, UK.                                steve at einval.com
"I suspect most samba developers are already technically insane... Of
 course, since many of them are Australians, you can't tell." -- Linus Torvalds