Slowing down processing on business errors

@sebarys wrote:

Hello,

I’m trying implement stream where:

• messages are read from init topic
• each message is processed (business logic part)
• based on processing result optionally publish message
• commit original message

In code it looks like below:

messagesSource
				.mapAsync(parallelismLevel, this::processMessage) //
				.log("after-message-processing", loggingAdapter)
				.via(Producer.flexiFlow(producerSettings))
				.map(ProducerMessage.Results::passThrough)
				.toMat(createCommittingSink(100L), Keep.both())
				.mapMaterializedValue(Consumer::createDrainingControl) //
				.run(materializer);

As processing message require communication with external services I would like slow down processing in case when upstream service couldn’t handle traffic.

Assuming that in this::processMessage I’m calling service A via HTTP and it can return e.g. 429 Too Many Requests that will result in PleaseSlowDownException is there a way to say downstream (in this case Kafka consumer source) to slow down?

Thanks for help in advance!

Posts: 3

Participants: 2

Read full topic