HiveMQ Kubernetes Operator repeatedly shutdown/restarts pods every 2 hours

Question

What could be the reason for HiveMQ Kubernetes Operator to repeatedly shut down/restart a node after exactly 2 hours?

Answer

The port named “mqtt” must have a listener in HiveMQ broker with the same port number. In case when the “mqtt” port is set to 1883 but not configured in the listenerConfiguration, it will prevent the liveness probe from functioning and the node will subsequently fail.

The HiveMQ Kubernetes Operator does automatically configure the liveness probe with the port named “mqtt”. If the “mqtt” port is 1883, it will check the 1883 TCP socket. If port 1883 is not configured in the listenerConfiguration, then this port will not be open on the pods and the liveness probe will fail.

To fix this, either make sure that the port named “mqtt” is configured in the listenerConfiguration or pick a port number that is already configured in the listenerConfiguration and rename it to "mqtt".

The following in the values.yaml will expose the port named "mqtt" with port number 1883:

 hivemq: 
  ports:
  - name: "mqtt"
    port: 1883
    expose: true

The matching listenerConfiguration in the config.xml:

<hivemq>
    <listeners>
        <tcp-listener>
            <port>1883</port>
            <bind-address>0.0.0.0</bind-address>
        </tcp-listener>
    </listeners>
</hivemq>