Bug#517449: Found in 2.6.30

[Berni Elbourn <berni@elbournb.fsnet.co.uk>]

Grim news there is the same behaviour using 2.6.30...

ski:~# cat /proc/version
Linux version 2.6.30-bpo.1-amd64 (Debian 2.6.30-5~bpo50+1) 
(nobse@debian.org) (gcc version 4.3.2 (Debian 4.3.2-1.1) ) #1 SMP Wed 
Aug 5 11:07:47 UTC 2009

And logs:

Aug  6 22:54:39 ski kernel: [ 2160.704087] INFO: task md2_resync:4371 
blocked for more than 120 seconds.
Aug  6 22:54:39 ski kernel: [ 2160.704270] INFO: task partimage:6788 
blocked for more than 120 seconds.
Aug  6 22:54:39 ski kernel: [ 2160.704418] INFO: task lvs:6911 blocked 
for more than 120 seconds.
Aug  6 22:56:39 ski kernel: [ 2280.704152] INFO: task partimage:6788 
blocked for more than 120 seconds.
Aug  6 22:56:39 ski kernel: [ 2280.704329] INFO: task lvs:6911 blocked 
for more than 120 seconds.

Looking back over things (for my systems) this bug is exposed when a 
raid 1 resync is in progress and partimage is taking a backup of one of 
the logical volume snap shots on the same physical disk.

Using dd instead of partimage in the backup script does not expose the 
bug - even when using the 2.6.26-17 kernel. I have lots more testing to 
do...more soon.

Berni

PS:

lvm2: 2.02.39-7
Part image: 0.6.7-1