In case of topology restart/failure, how in-flight messages in topology are replayed

[vaibhavtupe]

In case of Kafka-spout, there is a zookeeper, who has offset of last read message, so in case of topology restart or failures, state or offset can be read from zookeeper, and spout can start reading from there.

So here in rabbit-spout, if topology is restarted, how in-flight messages can be replyed or requeued ?

[burov4j]

Hi!

Sorry for the long time answer.
Unacked RabbitMQ messages returns back to RabbitMQ if topology restart/failure.
You may set "x-dead-letter-exchange" and "x-dead-letter-routing-key" headers to your RabbitMQ queue and all in-flight messages will sent to it.
Another way you may set property KEY_REQUEUE_ON_FAIL to "true" as described here: https://github.com/burov4j/storm-rabbitmq.
In the case your in-flight messages will return to the original queue.