Cause: When a broker first joins a cluster, it queries the cluster for all known fail-over addresses known to the cluster. No further attempts are made by the broker to keep the list of fail-over addresses up to date.
Consequence: Should the membership of the cluster change, new fail-over addresses are introduced. However, the broker will not be aware of these new fail-over addresses. Should the broker need to fail-over, the fail-over process may not work properly and the broker will not be able to remain in the cluster.
Fix: The broker has been modified to constantly monitor for changes in the cluster's fail-over address list.
Result: Should the set of valid fail-over addresses change after the broker has initially connected, the broker will be aware of these changes and should the broker need to perform fail-over it will succeed.