I'm trying to track down some memory issues. I think it might have to do with too many large tuples being in flight, but that brought up the question of where those tuples sit while waiting to be processed.
Assuming I have a bolt that emits tuples but the next bolt isn't ready to accept it yet, is the tuple queued in the emitting process or in the process of the bolt that will consume it for processing? I guess another way to ask the question is: When is the tuple transferred to the next bolt? When that bolt asks for it or when it is first emitted? Thanks! Jason
