openSuse 11.1 kernel 2.6.27.48 does not find lvm vol groups during boot

English Install/Boot/Login
1

Hi,

after a security update to kernel 2.6.27.48 (from 2.6.27.45) my system does not boot anymore. “Cannot find volume group xxx”. I’m using disks with md raid 1 and a root/swap partition on lvm. /boot is a md device.

System boot fails, when the kernel is not able to find the volgrp anymore.

Booting a rescue system, I can still mount the lvm devices.

How can I repair my system? How to rollback to the old kernel, which has been
deleted (rpms) on disk? Or any other suggestions? Special boot param? Something seems to be broken in the new installed kernel …

Many thanks in advance,

Stefan

2

Hi there,

i really dont know exactly, but shouldnt the old kernel be available at grub?

If not, or if you just have a fast boot, you could try to boot via an rescue System like the one from the opensuse cd or an knoppix and edit the /boot/grub/menu.lst the way you need it. Hopefully the old kernel is just available in /boot/ so you could just edit the entry, or set the timout to an higher value.
If everything fails, you could try to use an kernel from another system (same arch would be useful :slight_smile: and copy it to /boot, maybe it works

Greetz Joerg

3

Hi Joerg,

unfortunately the security update deleted the old kernel. Suse might ot might not know, why they are doing it this way.

I don’t have another system available with the old kernel. In that case I would have to copy all the needed stuff
from /boot and also the modules in /lib. Then - after a successful - boot, installing the rpms with yast to get a clean
system.

Is there a way to install the very old kernel from the 11.1 rescue DVD leaving the rest as it is?

Regards

Stefan

4

Hi,

Maybe you could just boot into the rescue,
mount the disk,
chroot into it
and install via zypper an other kernel.

Hopefully it just works :slight_smile:

Or you could just upgrade to 11.2 or .3 via dvd, so it should set a new kernel too

Greetz

5

Hi there,

I managed it to get the older kernel running now with access to my
lvm root volume using the chroot trick installing the old kernel.
I had to copy /dev/* into the chrooted environment to get the kernel
base rpm installed building its initrd.

But then I lost DNS resolving capabilities. Some libraries might have been
updated as well during the security update which has corrupted my system.

I still cannot find a logfile, which shows me the rpms being changed.
Any ideas why DNS fails now or where to lookup the changes made
to the system before all the pain started?

Stefan

6

Just in case it happens again,
the normal way of the chroot

  1. mount -t proc none /mntpoint/proc
  2. mount -o bind /dev /mntpoint/dev
  3. cp /etc/resolv.conf /mntpoint/etc/resolv.conf
    (make sure that you have an internet conenction before step 3 :slight_smile:
  4. chroot /mntpoint/ /bin/bash
7

Maybe there is now lvm support in your new initial ramdisk. I had same problem after upgrading to 11.3. Try ‘mkinitrd -f lvm2’ and then reboot.