AMQP connection failover doesn't work when connecting to AMQ Broker via OpenShift routes

Issue

• I am running an AMQ Broker cluster with 2 instances on OpenShift (deployed using the operator). The AMQPS ports (XXXX) of these instances are exposed to the public internet with OpenShift routes. I have also created a Red Hat Fuse application that publishes messages to a queue in this cluster. The connection to the cluster is accomplished using a failover connection string:

failover:(amqps://host1:443,amqps://host2:443)?failover.nested.transport.trustAll=true

• After having published a few messages I delete the pod of the first broker to restart it. I expect that the messages will then be published to the second broker (which is still running), but this doesn't happen. Instead, Fuse keeps trying to connect to the first broker until it has been restarted.