Add readiness probe or similar to ensure that nodes are "ready" for traffic

[antondalgren]

May have to wait for cloudamqp/lavinmq#1097

[markusgust]

there is a bug using this in 0.1.1 when creating 3 node, as running lavinmqctl status returns exit code 2 on followers. In snowstorm we (I think johanE) solve by running lavinmqctl status || lavinmqctl status | grep -q follower., where the second command will return ok.

from kubectl describe

Events:
  Type     Reason     Age                 From               Message
  ----     ------     ----                ----               -------
  Normal   Scheduled  16m                 default-scheduler  Successfully assigned lavinmq-cluster/lavinmq-1 to minikube
  Normal   Pulled     85s (x4 over 16m)   kubelet            Container image "cloudamqp/lavinmq:2.4.1" already present on machine
  Normal   Created    85s (x4 over 16m)   kubelet            Created container: lavinmq
  Normal   Started    85s (x4 over 16m)   kubelet            Started container lavinmq
  Normal   Killing    85s (x3 over 11m)   kubelet            Container lavinmq failed startup probe, will be restarted
  Warning  Unhealthy  75s (x91 over 16m)  kubelet            Startup probe failed: This node is a follower and does not handle lavinmqctl commands.
Please connect to the leader node by using the --host option.