Cassandra controller brings up node pools one at a time

[munnerz]

Because we rely on the Scale Action to set the initial node count after the CreateCluster action has created the Cassandra statefulset, the controller creates one node pool at a time, waiting for all replicas to become ready before creating the next.

In some situations, this could potentially lead to a cluster never coming up.

We should either perform these in parallel, or instead just set the replica count to nodePool.replicas in the CreateCluster action (which is what the Elasticsearch controller does).

/area cassandra

[wallrj]

In some situations, this could potentially lead to a cluster never coming up.

Can you give examples of these situations?