Bug#542037: mdadm: Possible bad LSB header in initscript

[martin f krafft]

forcemerge 506442 542037
thanks

also sprach Stephane Glondu <steph at glondu.net> [2009.08.17.1514 +0200]:
> My / is inside an LVM which is itself inside an MD device. I've just
> switched to insserv 1.12.0-10, and RAID arrays are stopped (or
> let's say fail to be stopped) at the beginning of the shutdown
> sequence. 
> 
> The attached patch makes mdadm-raid stop at the right time (... at least
> on my system :).

mdadm cannot stop after umountroot because then /sbin/mdadm won't be
available anymore (and neither will be the init.d script).

The shutdown sequence is a dependency cycle: / might be on LVM which
might be on RAID, which needs / to stop it.

In my experience, this causes a warning printed to the screen but
doesn't actually have any bad effects. In particular, the RAID won't
be resynchronised, but I cannot quite explain to you why that
doesn't need to happen.

The only solution would be to pivot_root into the initrd during
shutdown. There is some charm to this idea, but I don't even want to
think about what this entails and whether the benefits would
outweigh the costs.

-- 
 .''`.   martin f. krafft <madduck at d.o>      Related projects:
: :'  :  proud Debian developer               http://debiansystem.info
`. `'`   http://people.debian.org/~madduck    http://vcs-pkg.org
  `-  Debian - when you have better things to do than fixing systems
-------------- next part --------------
A non-text attachment was scrubbed...
Name: not available
Type: application/pgp-signature
Size: 197 bytes
Desc: Digital signature (see http://martin-krafft.net/gpg/)
URL: <http://lists.alioth.debian.org/pipermail/pkg-mdadm-devel/attachments/20090821/e74ed723/attachment.pgp>