Background

Because the Red Hat cluster schema is changed when used with Open Sharedroot at the clusternode elements (com_info) it has been so that the cluster configuration was changed from the management tools so that the com_info elements have been removed.

Solution

As of Red Hat Enterprise Linux 5.3 this behavior has changed and those tools can be used again. The only thing that can be ignored is that system-config-cluster mangles the validation but will not throw away those com_info elements.

If luci does not show a "Add virtual machine" you need to install the libvirt rpms.

There might be a way around those validation at some point in time and described.

Note

Don't forget to build a new initrd after each cluster configuration change. Note that before any node in the cluster can be rebootet successfully the cluster configuration has to be applied to the initrd.