Evidence of cluster nodes is saved in database in the table Node, see Data model.

If there is no infomation about node in database then it will be created when the application server starts.

Every node can be in three states:

• RUN - Node handles new and existing (saved messages in ESB) messages
• HANDLES_EXISTING_MESSAGES - Node handles only existing messages. New messages/requests are rejected. Used is Stopping mode.
• STOPPED - Node not handles existing messages and new messages

Every processing message (in states IN_QUEUE and PROCESSING) has informace which on which node is being processed. When message finishes current processing then reference to the node will be deleted.

There is configuration property ohf.cluster.actualNodeInstance.code that defines unique code (name) of current running cluster node.