Regular testing

Regularly schedule a testing window where a failure is conducted in a controlled environment. That way you can evaluate a fallover before anything happens in your production cluster.

It should include fallovers of all nodes and full verification of tested protected applications. This is strongly encouraged if you are changing or evolving your cluster environment.