karalabe
commented
9 years ago Iris tries to be a message transport only, on top of which you can implement more specific logic to suit your needs. It provides various messaging primitives, but it does not provide any persistency. In essence, if a node crashes, it takes with itself all locally queued messages.
What happens in the event you flood a specific queue with messages (1M) and it takes a while to distribute but entire nodes crash without having any left?
How does Iris handle situations like this?
Thank you!