Comment 13 for bug 1124384

I believe the problem here is that -- in the case where a JobClass has no running instances -- a reload is causing the original JobClass to be destroyed. That eventually leads to Event->blockers to reach zero (as a result of the JobClasses EventOperators being freed) causing the event to be destroyed. That behaviour is correct inasmuch as events are being destroyed when they should. What is not correct is that the reload results in all this happening *before* the newly-loaded JobClass gets a chance to increment Event->blockers to avoid the event being freed.