Checklist
1 If possible, install a test environment first with clustering enabled.
2 Make sure the high-availability port and clustering port numbers that you want to use are available.
3 Decide on a partition name and use it on both servers in the cluster.
4 Set up the load balancer to use an HTTP heartbeat for contacting JBoss nodes.
5 Ensure that JBoss nodes:
• Are on the same partition.
• Connect to the same JMS database.
• Can communicate with the chosen protocol. If one node cannot see the other, try switching to TCP.
6 In Highway:
• Edit the JMS endpoints to connect to the JMS queue in a standard way.
• Specify the server and the port for the endpoints to use.
• Specify that JMS clustering is enabled.
7 For the listeners:
• Set the URL for the primary and secondary JBoss nodes in Registry Edit.
• Set up brokers.
8 Test different startup sequences to ensure clustering is working correctly for all scenarios. Start SERVER1, then SERVER2. Shut down JBoss on SERVER1. Ensure that queues migrate to SERVER2. Check if the listeners now all consume messages from SERVER2. Start up SERVER2 first, then SERVER1. Shut down JBoss on SERVER2. Check that all the listeners switch to SERVER1.
Note: Consider the services startup sequence when clustering is in use. When SERVER1 is started, JBoss is typically configured to start automatically as a service. SERVER2 should not be started until JBoss on SERVER1 has started up fully. This prevents any confusion between the two nodes as to which server is the primary. Similarly, any script that shuts down SERVER1 and SERVER2 should ensure that there is no crossover between SERVER1 and SERVER2 when starting up or shutting down.