I have an aggregator right before an outbound adapter (in order to batch requests).
I would like to ensure that, when the outbound adapter is failing, the aggregator stops releasing messages, and, in turn, the backpressure propagates up the integration flow, until finally the outbound adapter comes back online.
This seems to me to be a desirable pattern within EIP, and, while I'm pretty sure it involves queue backed channels, I'm unsure how this is implemented for a couple of reasons.
A) If the failure start of the adapter is connoted by exceptions (and in fact I'm using @retryable on the outbound adapter), what is the 'right' way to requeue messages?
B) I'm assuming the aggregator's reaper thread will respect downstream queue 'backpressure.' That said, how does one tell the aggregator to stop consuming messages from upstream in that situation?
