Using CICS system trace entry points
CICS system tracing is probably the most powerful technique for finding out whether a transaction started. You might need to direct the trace output to the auxiliary trace destination, depending on how certain you can be about the time the task is expected to start. Even a large internal trace table might wrap and overlay the data you want to see if you are not too sure about when the task should start.
Before you begin
Procedure
- Select special tracing for just your task, and disable tracing for all other tasks by setting the master system trace flag off.
- Set up special tracing for the level one trace points for the components that are likely to be used during the invocation of the task. The components you choose depend on how the task is initiated: by direct request from a terminal or by automatic transaction initialization. The components should always include loader domain (LD), program manager (PG), transaction manager (XM), and dispatcher domain (DS). Make sure that special tracing is disabled for all other components, to minimize the amount of trace data that is collected and the tracing overhead.
- Now turn tracing on, and attempt to start your task.
- When you are sure that the time has passed when the output should have appeared, stop tracing, and format the trace data set.
Results
- Loader domain, when it loaded your program, if the program was not already in main storage.
- Transaction manager, when it attached your task to the dispatcher.
- Dispatcher domain, when your task got its first dispatch. You might also see subsequent entries showing your task being suspended, and then resumed.
- Program manager, for any program management functions associated with your task.
If trace entries for any of these processes are missing, that should help you to find where the failure occurred.