Change cluster policy to availability_of_members

Because the cluster isn't managed by Pacemaker, using
majority_of_members is misleading.
1 file changed