Spring Cloud Stream is a Spring module based of another module, Spring Cloud Function.
It allows creating event-driven applications by abstracting the binder type.
You can switch Kafka, RabbitMQ, AWS Kinesis and others, but you will not need to change anything in the code.
In this project I show you the Spring Cloud Stream with Kafka.
In the project there is an example of consumer and two examples of produces (both in imperative way and reactive way).
In the SensorEventFunctions class, I show you Spring Cloud Stream in reactive mode,
in the SensorEventImperativeFunctions, I show you the same flow in the imperative mode.
There is a docker-compose.yml file in the root of project to run Zookeeper and Kafka containers.
With Spring Boot 3.1, we have the docker compose support out of box (adding the spring-boot-docker-compose dependency).
So, you can just run the application with your IDE or with command: ./mvnw spring-boot:run, and Spring will execute
the docker compose in docker-compose.yml file and then the application.
There is an automatic producer that send the date every 5 seconds (the sensorEventAnotherProducer producer).
The project works both imperative and reactive mode, by the spring profiles (reactive and imperative).
The default mode is reactive, but you can replace the profile in application.yml.
For the imperative mode, there is an automatic DQL and retry management of DLQ is provided by Spring.
The DLQ is handled by the enable-dlq and dlq-name properties. When the consumer throws an exception, the message
will be processed again for two more times (the default of maxAttempts is 3), and after the message will be sent in DLQ.
For the reactive mode, you need to do a custom handling for DLQ and retries (see DlqEventUtil class).