Description

fixes: #618

Implementation notes

• current docs state All roles should have a direct ListenerClass, such as cluster-internal or external-unstable
  • ephemeral listener volumes are used by calling add_listener_volume_by_listener_class (unlike HDFS, where the NameNodes have a persistent listener volume)
  • I'm not sure this is correct: should they all be stable ListenerClasses using persistent volumes and not managed by the operator?
• unlike Kafka (where the bootstrap listener is constructed by the operator), but like HDFS (where the listener-operator creates the listenters based on the PVCs), the listener references have to be fetched from k8s: this is done using a similiar pattern to HDFS (see mod.rs: functions rolegroup_ref, rolegroup_ref_and_replicas, pod_refs, listener_refs)
• a config map named *-ui-endpoints (e.g. test-hbase-ui-endpoints) is created, containing an endpoint for one replica in each role e.g.

  • apiVersion: v1
    data:
      hbase.master.ui: 172.19.0.5:32563
      hbase.regionserver.ui: 172.19.0.3:31687
      hbase.restserver.ui: test-hbase-restserver-default-0-listener.kuttl-test-glad-zebra.svc.cluster.local:8085
    kind: ConfigMap
    

• these endpoints are checked in the test, which assigns different listener classes to each role, so all get checked

Definition of Done Checklist

• Not all of these items are applicable to all PRs, the author should update this template to only leave the boxes in that are relevant
• Please make sure all these things are done and tick the boxes