Description of problem:

drbd-rgmananger rpm package depends on Red Hat Cluster Suite version 2 but works on RHCS 3. To get it working, I built the drbd package using the following commands:

# cd drbd-8.3.8.1/
# ./configure
# make rpm RPMOPT="--with rgmanager"

To install, I had to use the "--nodeps" option to ignore the dependency on RHCS version 2.x:

# rpm -Uvh --nodeps ~/rpmbuild/RPMS/x86_64/drbd-rgmanager-8.3.8.1-1.el6.x86_64.rpm

Unfortunately, the resource agent did not work out of the box. Perhaps I should have added a delay in the cluster.conf, but I added "sleep 1" in the beginning of the resource agent /usr/share/cluster/drbd.sh

Failover works!!!

The point is that I was tempted to use gfs2 over dual primary drbd precisely because of the drbd-rgmanager package installation obstacles and the fact that I could not find anybody else who had tried the resource agent on RHCS 3. Users should not be tempted to use dual primary drbd unless strictly necessary, since they (as I did) will struggle with frustrating split brain issues.
 

Version-Release number of selected component (if applicable):

 - drbd 8.3.8.1 source
 - RHEL 6.0 and 6.1
 - Red Hat Cluster Suite 3.0.12-22.el6