CentOS - Sep 2011 - CentOS 6: corosync and pacemaker won't stop (patch)

Hi,

I cannot 'halt' my CentOS 6 servers while running corosync+pacemaker.
I believe the runlevels used to stop corosync and pacemaker are not in the
correct order and create the infinite "Waiting for corosync services to
unload..." loop thing.

This is my first time with this cluster technology but apparently pacemaker
has to be stopped /before/ corosync.
Applying the following patch to pacemaker init script fix this issue.

# diff -Nu /etc/init.d/pacemaker{.orig,}
--- /etc/init.d/pacemaker.orig  2011-09-27 14:08:32.984117731 +0200
+++ /etc/init.d/pacemaker       2011-09-27 14:08:56.125143439 +0200
@@ -6,7 +6,7 @@
 #
 # License: Revised BSD

-# chkconfig: - 90 90
+# chkconfig: - 90 75
 # description: Pacemaker Cluster Manager
 # processname: pacemaker
 #

Florian

Florian Crouzat wrote on?2011-09-29:
> Hi,
>
> I cannot 'halt' my CentOS 6 servers while running
corosync+pacemaker. I
> believe the runlevels used to stop corosync and pacemaker are not in the
> correct order and create the infinite "Waiting for corosync services
to
> unload..." loop thing.
>
> This is my first time with this cluster technology but apparently
> pacemaker has to be stopped /before/ corosync. Applying the following
> patch to pacemaker init script fix this issue.
>
> # diff -Nu /etc/init.d/pacemaker{.orig,}
> --- /etc/init.d/pacemaker.orig  2011-09-27 14:08:32.984117731 +0200
> +++ /etc/init.d/pacemaker       2011-09-27 14:08:56.125143439 +0200
> @@ -6,7 +6,7 @@
>  #
>  # License: Revised BSD
> -# chkconfig: - 90 90
> +# chkconfig: - 90 75
>  # description: Pacemaker Cluster Manager
>  # processname: pacemaker
>  #
Btw, it has already been fixedupstream.
https://github.com/ClusterLabs/pacemaker/blob/master/mcp/pacemaker.in#L9


Greetings,
Florian