From: Konstantin Kletschke on
Hello,

I tried to do a Kernel update on an (mainly) lenny box after it had an
uptime of over 400 days. This means quite old Box with old kernel meets
bleeding edge modern stuff. Additionally I moved from lilo to grub2 (if
that is from concern).

Well the update itself did run very well, grub2 loads new kernel and
ramdisk and iniiates to start userspace. There pvscan detects no
physical volumes anymore all of a sudden. Booting the old kernel again
runs well and the bos comes up and pvscan works and the logical volumes
are mounted and working in the end.

The old kernel is a debian packge:

2.6.17-2-686 #1 SMP Thu Aug 31 12:53:18 UTC 2006 i686 GNU/Linux

The new kernel is also debian package:

linux-image-2.6.32-5-686

Additional the versions of the following packages may be relevant:

lvm2: 2.02.39-7
libdevmapper1.02.1: 2:1.02.27-4
udev: 0.098-2

The box is a headless production system in a server hosting farm so I
sadly was not able to investigate the situation proper, pvscan simply
did not found any volumes.

What can cause this error, what of my packages is too old to trigger
such behaviour? Is there a save way upgradig from this situation to a
modern kernel?

Kind Regards, Konsti


--
To UNSUBSCRIBE, email to debian-user-REQUEST(a)lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster(a)lists.debian.org
Archive: http://lists.debian.org/87lj9h6n63.fsf(a)dell-01.ku-gbr.de
From: Michael Tsang on
On Monday 12 July 2010 17:53:56 Konstantin Kletschke wrote:
> Well the update itself did run very well, grub2 loads new kernel and
> ramdisk and iniiates to start userspace. There pvscan detects no
> physical volumes anymore all of a sudden. Booting the old kernel again
> runs well and the bos comes up and pvscan works and the logical volumes
> are mounted and working in the end.
>
> The old kernel is a debian packge:
>
> 2.6.17-2-686 #1 SMP Thu Aug 31 12:53:18 UTC 2006 i686 GNU/Linux
>
> The new kernel is also debian package:
>
> linux-image-2.6.32-5-686
>
> Additional the versions of the following packages may be relevant:
>
> lvm2: 2.02.39-7
> libdevmapper1.02.1: 2:1.02.27-4
> udev: 0.098-2
>
> Kind Regards, Konsti
Are you using the kernel to autodetect the LVM volumes (using type 0xfd). It
does not handle the volumes well, about a week ago, it *corrupted* my array
because it couldn't distinguish the superblocks of /dev/sdb and /dev/sdb1.
Finally, I used userspace tools (mdadm) to detect my array by specifying the
arrays in /etc/mdadm/mdadm.conf

Also, did you specify any of /dev/hd? in /etc/mdadm/mdadm.conf? The new kernel
treats IDE devices as SATA devices so you will use /dev/sd? to specify the
hard disks.
--
Please avoid sending me Word or PowerPoint attachments.
See http://www.gnu.org/philosophy/no-word-attachments.html
From: konsti on

Sorry, this is not a reply but in initial message to the list.

> Are you using the kernel to autodetect the LVM volumes (using type
0xfd). It

Well I use only the kernel modules (dm_mod) and let userspace do its
stuff,
mainly debian's /etc/init.d/lvm does the job well. It relies on pvscan
detecting the volumes.

There never was mdadm or md on this box, only one single disk /dev/sdb
containing no partition table but the physical volume.

> Also, did you specify any of /dev/hd? in /etc/mdadm/mdadm.conf? The new
kernel

As said, no mdadm tools on this machine. I actually have a
/etc/mdadm/mdadm.conf
(its contens is "DEVICE partitions" only), but it is not owned by any
package.

> treats IDE devices as SATA devices so you will use /dev/sd? to specify
the
> hard disks.

I thought of that already, but the disks are SATA so the were named sda
and sdb any
time.

Konsti


PS.: Pleas CC me, since I am not on this List.


--
To UNSUBSCRIBE, email to debian-user-REQUEST(a)lists.debian.org
with a subject of "unsubscribe". Trouble? Contact listmaster(a)lists.debian.org
Archive: http://lists.debian.org/a3a9b47b38dbc2c1af3123ce082c16ab(a)localhost