Pacemaker, as a cluster resource manager, has some defaults that we are interested in changing. As Pacemaker is so powerful, it makes several assumptions about the composition of cluster resources and nodes it controls, one of which is that there are several nodes, and not just two.
This works well for large cooperative networks of web servers or independent services that can operate in a transient manner. However, we have two nodes that are very much dependent on shared storage that can only be used by one node at a time. So, in this recipe, we are going to perform three tasks:
- Disable STONITH because we don't currently have STONITH-enabled hardware
- Disable the cluster quorum because two systems cannot produce a meaningful vote
- Enable resource stickiness to prevent disruptive automated node swaps
