This is related to https://issues.jboss.org/browse/WFLY-8492.
Currently, the messaging-activemq resources that defines the call-failover-timeout attribute uses a default value of -1.
That means that during failover, if an Artemis client sends a Packet in a non-blocking call, they will wait forever to get a response. It leaves no opportunity for the client to "give up" (e.g. if the application is asked to shut down).
Artemis changed their default values to 30s (in ARTEMIS-255) and we should do the same for the messaging-activemq attribute.