This topic contains sections marked as revised for this release
This topic shows how your message flows can benefit from user exits.
Before you start you must read User exits.
In the diagram below, the MQInput node is used as an example, but the function applies to all input nodes, including user-defined nodes.
Event 1. The message is dequeued from the input source (read into the flow). However, built-in nodes and user-defined nodes differ slightly in the way in which user exits are invoked. For built-in input nodes, the user exit is invoked as soon as possible after the data has been read from the external source. For user-defined input nodes, the user exit is called just before the node propagates the message.
Event 2. The message is propagated to the node for processing.
Event 3. The node processing is completed.
Event 4. The user exit is invoked after the transaction has completed, so the user exit processing is not part of that transaction. The user exit is invoked even if no transactional processing was completed by the flow. Where the message flow property Commit Count is greater than 1, many-to-one ratios exist between event 1 and event 4. This ratio also exists for some scenarios specific to the particular input node; for example, when an MQInputNode is configured with the Commit by Message Group property selected.
You can register multiple user exits, and, if they are registered, they are invoked in a defined order (see mqsichangeflowuserexits command). Any changes made to the message assembly by a user exit are visible to subsequent user exits.
The user exits can be registered on a dynamic basis, without needing to redeploy the configuration.