Universal Messaging Clustering Requirements
In Universal Messaging, a cluster is defined as multiple server instances in Universal Messaging installations on the same or different machines that are configured for clustering.
This section is intended for use with the rest of this guide. You must perform all tasks in this guide for your products, in the order they are written, with the additional tasks or exceptions noted in this section.
When you install the new
Universal Messaging servers instances, create a cluster that matches your old cluster. You can install the new
Universal Messaging server instances in parallel.
When you install the latest fixes on the new
Universal Messaging servers instances, you can install the fixes in parallel.
When you shut down the old products, shut down all cluster nodes.
When you run the
Universal Messaging migration utility to migrate server instances from old
Universal Messaging installations to new
Universal Messaging installations, you can migrate the instances in parallel.
3. Clear channels and queues by deleting the data directories from all but one of the new server instances.
4. If the new cluster nodes are on the same host machines as the old cluster nodes, start all new cluster nodes.
If the new cluster nodes are on different machines than the old cluster nodes, do the following:
a. Go to the new_Software AG_directory/UniversalMessaging/server/instance/bin directory for one cluster node and create a file named remote_realms_bootstrap.conf. Add a cross-host mapping property to the file for every node that is part of the cluster in the format instance.address=new_host_name_or_IP (for example, umserver1.address=myUM1.com or umserver2.address=192.168.0.1). Copy the file to the same location for all the other nodes.
b. Start all cluster nodes. The remote_realms_bootstrap.conf file in every installation will be renamed remote_realms_bootstrap_old.conf. When a source Universal Messaging server instance is bound to a specific network interface (absolute IP address), the interface of the migrated instance address will bind to 0.0.0.0 (all known interfaces on the port).
5. Open Enterprise Manager to make sure the cluster is running correctly.