Topology Updates Cause Client Connection Failures when Broker Cluster is Behind a Load Balancer

When configuring a client with a static URI (pointing to a load balancer appliance) in the following environment:

- Spring Boot Artemis Core Client 
- A-MQ Broker 7.0.3 clustered configuration
- Load Balancing Appliance between the client and the broker cluster
- Client ConnectionFactory configured with a static broker URL with the IP address of the load balancer 
- Load Balancer configured to forward traffic on port 61616 to the live brokers in the cluster
- Firewall rules allow traffic only between the load balancer and the broker cluster

The client initially connects successfully to the broker through the load balancer. However, we see with trace logging enabled that the client subsequently receives a topology update from the broker containing the actual hostname of the broker. The client subsequently attempts to connect to the broker at the updated hostname and fails, due to firewall rules.

There does not appear to be a way to ignore these updates in the core client. Setting useHA=false on the URI parameters does not defeat the updates or resolve the issue, though we verified that the property is set in the ServerLocatorImpl.

Masking the updates at the broker side by configuring the broker to send the load balancer hostname or IP address would break clustering.

We need a way to defeat topology updates in environments where host / IP information is not allowed to cross network boundaries.