When an event is submitted to the EventPublisher to be published asynchronously, because the system only allows a finite number of pending events, it's possible queuing the event might fail.

When that happens the system logs stack traces for all of the active event threads, to facilitate diagnosing what listener(s) are holding event threads. However, because events can be raised from event handlers, it's possible this can be triggered from an event thread. When that happens, the logging only serves to exacerbate the problem because it's holding one of the event threads until all of the stack traces can be logged.