Bug#689221: installation-reports: QNAP TS-409U does not reboot after installation

[Luca Capello]

Hi there!

On Tue, 16 Oct 2012 00:05:59 +0200, Luca Capello wrote:
> On Mon, 08 Oct 2012 23:12:33 +0200, Luca Capello wrote:
>> At the next reboot mdadm does not segfault anymore, so at least this
>> problem is fixed in 3.2.5-3~bpo60+1, it could be related to:
>>
>>   <http://bugs.debian.org/621786>
>
> I am quite sure I got it by this one, I do not know why I have not
> completely read it the first time, it would have saved me quite a lot of
> tests...
>
> Cc:ing all the persons involved in that bug: Arnaud, I would try
> mdadm_3.2.5-3~bpo60+1 (from backports, and a more recent kernel just to
> be sure).  If we experience the same bug that should be solve it.
[...]
> I will prepare a d-i network-console based on 20110106+squeeze4+b1 plus
> mdadm_3.2.5-3~bpo60+1 and try again the full installation next Sunday
> 21st: if everything goes well, I will close this bug :-)

Summary: installing mdadm_3.2.5-3~bpo60+1 *before* rebooting allow a
normal boot into the new installed system, so I guess this bug is fixed.

The long story is that it was even easier, because the backports udeb
was already available, so I used it with the following procedure:

1) flash d-i
2) ssh installer at IP
3) start 'basic mode'
4) at 'Select a language' screen -> Go back
5) execute a shell
=====
~ # wget http://ftp.de.debian.org/debian-backports/pool/main/m/mdadm/mdadm-udeb_3.2.5-3~bpo60+1_armel.udeb
~ # udpkg -i mdadm-udeb_3.2.5-3~bpo60\+1_armel.udeb 
~ # exit
=====
6) continue installation (no more segfaults)
7) manual partitioning (the same layout as the first try, thus without
   recreating the RAID-5, but reusing it)
8) installation continued, but at "Configure mdadm" and "Flash
   kernel..." there were segfaults (mdadm from within the target, thus
   the squeeze version, not from backports)
9) installation finished -> Go back
10) execute a shell
=====
~ # mount /sys/ /target/sys/ -o bind
~ # mount /proc/ /target/proc/ -o bind
~ # chroot /target /bin/bash
root at jem:/# mv /etc/mdadm/mdadm.conf{,.squeeze}
[missing ARRAY entry]
root at jem:/# echo "deb http://backports.debian.org/debian-backports squeeze-backports main" >>/etc/apt/sources.list
root at jem:/# apt-get update
root at jem:/# apt-get -t squeeze-backports install mdadm
Preparing to replace mdadm 3.1.4-1+8efb9d1+squeeze1 (using .../mdadm_3.2.5-3~bpo60+1_armel.deb) ...
[...]
Generating mdadm.conf... done.
update-initramfs: deferring update (trigger activated)
Generating udev events for MD arrays...done.
Starting MD monitoring service: mdadm --monitor.
Processing triggers for initramfs-tools ...
update-initramfs: Generating /boot/initrd.img-2.6.32-5-orion5x
Generating kernel u-boot image... done.
Flashing kernel... done.
Flashing initramfs... done.
root at jem:/#
=====
11) now finish the installation, reboot and enjoy, finally!

Please note that I previously reported a kernel panic at the first
reboot after having installed mdadm from backports.  Well, it happened
again but on a random basis, which seems to correlated with other
segfaults I experienced:
=====
root at jem:/etc# apt-get install etckeeper
[...]
Selecting previously deselected package etckeeper.
Unpacking etckeeper (from .../etckeeper_0.48_all.deb) ...
Selecting previously deselected package rsync.
Unpacking rsync (from .../rsync_3.0.7-2_armel.deb) ...
Processing triggers for man-db ...
locale: Cannot set LC_ALL to default locale: No such file or directory
Setting up libcurl3-gnutls (7.21.0-2.1+squeeze2) ...
Segmentation fault
dpkg: error processing libcurl3-gnutls (--configure):
 subprocess installed post-installation script returned error exit status 139
[...]
root at jem:/etc# git status
error: corrupt loose object '1ef26ebf2691cf6b09a55a2a10a02ff7634f3b24'
fatal: object 1ef26ebf2691cf6b09a55a2a10a02ff7634f3b24 is corrupted
root at jem:/etc# git log
commit 361404b10074730c1710edc5bef5c06a332872e9
Author: root <root at jem.pca.it>
Date:   Mon Oct 22 19:08:44 2012 +0200

    Initial commit
root at jem:/etc# apt-get install screen
[...]
Setting up screen (4.0.3-14+b1) ...
fatal: object 49111c287d678ddd1148f8c9832625036e0aef33 is corrupted
[master E: Problem executing scripts DPkg::Post-Invoke 'if [ -x /usr/sbin/etckeeper ]; then etckeeper post-install; fi'
E: Sub-process returned an error code
root at jem:/etc# apt-get --reinstall install screen
[...]
Setting up screen (4.0.3-14+b1) ...
[master 01881cc] committing changes in /etc after apt run
 Committer: root <root at jem.pca.it>
[...]
 1 files changed, 1 insertions(+), 1 deletions(-)
root at jem:/etc#
=====

The various segfaults could be related to the RAID-5 being repaired,
given that RAIDs are created in a degrade state:
=====
root at jem:/etc# cat /proc/mdstat
Personalities : [raid6] [raid5] [raid4]
md0 : active raid5 sda1[0] sdd1[3] sdc1[2] sdb1[1]
      5860538880 blocks super 1.2 level 5, 512k chunk, algorithm 2 [4/4] [UUUU]
      [>....................]  resync =  3.2% (63772868/1953512960) finish=2486.0min speed=12668K/sec

unused devices: <none>
root at jem:/etc# 
=====

Thx, bye,
Gismo / Luca
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 835 bytes
Desc: not available
URL: <http://lists.alioth.debian.org/pipermail/pkg-mdadm-devel/attachments/20121022/b97a1bc9/attachment.pgp>