Adapter Listener Support in a Cluster
Beginning with Integration Server 9.5, an adapter listener can be active either on multiple nodes or on a single node in an Integration Server cluster.
If the adapter back end supports the non-duplication of messages to which multiple clients are subscribed, the listener is multi-node. It is active on all nodes in the Integration Server cluster. In this case, each node retrieves and processes a different message.
If the adapter back end does not support the non-duplication of messages to which multiple clients are subscribed, the listener is single-node. It is active on a single node in the cluster - the active or primary node - thus avoiding duplicate messages. If the active node goes down, another node in the cluster becomes active, providing failover support. The active node stores incoming events from the back end in the distributed (shared) cache. In this way, message processing continues when the active node goes down.
Important:
If message order is important when the adapter processes events from the back end, use a combination of a single-threaded and single-node listener, or a single-threaded and multi-node listener that is enabled on only one node in the cluster. For information about how multi-threaded listeners work, see
Single-Threaded and Multi-Threaded
Listeners.
You can enable, disable, or suspend a listener on a single node or on all nodes in a cluster by changing the listener state on one of the nodes. For information about changing listener states in a cluster, see
Enabling Listeners in a Cluster and
Disabling Listeners in a
Cluster.