Troubleshooting Timer and Event Observer Services

When running a composite service using a timer or an event observer, the service may write the following warning to the composite service standard log:

[Warning][External][TIMADLM][13][Event dropped, queue is full. Request not being executed: <internal activity diagram ID>]

The Console provides facilities for viewing the  xUML Service Standard Log.

The warning message occurs if a request is due and the forerunner has not finished. The system tries to start a new parallel thread; however, if this is not possible, because the parallel limit is reached (default = 1), the system puts the event into a queue. Since the queue size is also limited (default = 1), the above warning occurs if there is no more space in the event queue. Most of the time, the reason a service logs the warning is that the execution of the timed activities requires more time than is allowed by the tagged value repeatInterval. If the service logs the warning sporadically, the service will recover. However, if the warnings persist, the following counter measures are recommended:

• Confirm that the reason for the slow execution is not pathological, i.e. a backend system is not working properly. This would result in additional error messages in the log.
• Increase the value of the tagged value repeatInterval if the activities must be executed one after the other.
• Increase the value of the tagged value parallelEvents if the activities may be executed in parallel and the reason for the high execution time is primarily I/O.
• Increase the value of the tagged value queuedEvents if the log message occurs sporadically.

In order to analyze the time behavior, use the transaction logging feature of the Console described in the Integration (Bridge) User's Guide.