Sending and receiving JSON messages with Spring Boot AMQP and RabbitMQ

Exchange JSON messages with Spring Boot AMQP and RabbitMQ

Table of Contents

Setting up a Spring Boot application using AMQP with RabbitMQ is not really difficult, and there are many guides on the Internet showing you the basic setup. But creating an application making use of @RabbitListener annotations and producing and consuming messages in JSON format is trickier, so I would like to share with you a really simple but more serious approach that those hello-messaging apps.

This article covers:

How to send/publish Java Objects as JSON messages using Spring Boot and RabbitMQ’s RabbitTemplate.
How to read/consume JSON messages to Java Objects using Spring Boot and RabbitMQ’s @RabbitListener annotation.
How to send and receive Java Objects through RabbitMQ using default Java serializer.

Updated: The code examples use Java 10 and Spring Boot 2. More info.

Introduction

As you probably know if you reached this article, AMQP is a specification, a protocol that defines messaging and patterns for communication between systems. RabbitMQ is a software that implements AMQP. You can find a lot of tutorials about these concepts on the Internet, apart from many useful books.

Goal

Within this article we’ll set up a Spring Boot application that uses RabbitMQ to send messages to a topic exchange formatted as JSON, and read them using two different queues:

The first Queue will read the contents as a generic org.springframework.amqp.core.Message class.
The second Queue will transform the JSON message to a specific class created by us.

I’ll also show you an alternative configuration to make this work without formatting the message to JSON, but nowadays this format is a good choice for intercommunication between applications and systems that may differ in programming languages, for instance.

I recommend you to check the implementation as you follow these instructions. You can do so by cloning the GitHub repository. Don’t forget to give it a star if you find it useful!

You can find all the source code on GitHub: Spring Boot JSON AMQP messaging. If you like it, give it a star!

Setting up the project

pom.xml

Creating a simple Spring Boot application is pretty straightforward, in my case I use Spring Boot 2.0 and Maven. The only thing you need to do to make use of AMQP with RabbitMQ in Spring Boot is to include the corresponding library, Spring Rabbit:

<dependency>
   <groupId>org.springframework.amqp</groupId>
   <artifactId>spring-rabbit</artifactId>
</dependency>

<groupId>org.springframework.amqp</groupId>

<artifactId>spring-rabbit</artifactId>

</dependency>

The message

In this case, we’re going to send a really simple content, but enough to have an idea about how the message looks like in different formats. The class is called CustomMessage.

public final class CustomMessage implements Serializable{

    private String text;
    private int priority;
    private boolean secret;

    // Default constructor is needed to deserialize JSON
    public CustomMessage() {
    }

    public CustomMessage(String text, int priority, boolean secret) {
        this.text = text;
        this.priority = priority;
        this.secret = secret;
    }

    public String getText() {
        return text;
    }

    public int getPriority() {
        return priority;
    }

    public boolean isSecret() {
        return secret;
    }

    @Override
    public String toString() {
        return "CustomMessage{" +
                "text='" + text + '\'' +
                ", priority=" + priority +
                ", secret=" + secret +
                '}';
    }
}

public final class CustomMessage implements Serializable{

private String text;

private int priority;

private boolean secret;

// Default constructor is needed to deserialize JSON

public CustomMessage() {

}

public CustomMessage(String text, int priority, boolean secret) {

this.text = text;

this.priority = priority;

this.secret = secret;

}

public String getText() {

return text;

}

public int getPriority() {

return priority;

}

public boolean isSecret() {

return secret;

}

@Override

public String toString() {

return "CustomMessage{" +

"text='" + text + '\'' +

", priority=" + priority +

", secret=" + secret +

'}';

}

Pretty basic POJO. The only things to point out are the default constructor, which we need for JSON deserialization, and the toString() method, that will give us a human-readable representation of the objects.

Sending messages

As mentioned before, this example will use a topic exchange to send CustomMessage objects.

@Service
public class CustomMessageSender {

    private static final Logger log = LoggerFactory.getLogger(CustomMessageSender.class);

    private final RabbitTemplate rabbitTemplate;

    @Autowired
    public CustomMessageSender(final RabbitTemplate rabbitTemplate) {
        this.rabbitTemplate = rabbitTemplate;
    }

    @Scheduled(fixedDelay = 3000L)
    public void sendMessage() {
        final CustomMessage message = new CustomMessage("Hello there!", new Random().nextInt(50), false);
        log.info("Sending message...");
        rabbitTemplate.convertAndSend(MessagingApplication.EXCHANGE_NAME, MessagingApplication.ROUTING_KEY, message);
    }
}

@Service

public class CustomMessageSender {

private static final Logger log = LoggerFactory.getLogger(CustomMessageSender.class);

private final RabbitTemplate rabbitTemplate;

@Autowired

public CustomMessageSender(final RabbitTemplate rabbitTemplate) {

this.rabbitTemplate = rabbitTemplate;

}

@Scheduled(fixedDelay = 3000L)

public void sendMessage() {

final CustomMessage message = new CustomMessage("Hello there!", new Random().nextInt(50), false);

log.info("Sending message...");

rabbitTemplate.convertAndSend(MessagingApplication.EXCHANGE_NAME, MessagingApplication.ROUTING_KEY, message);

}

Some comments about this service:

@Scheduled annotation is used to schedule a method to be executed at a given time, or with a given frequency. In this case, this method will be executed every 3 seconds (until the end of the times or you stop the application, whatever happens earlier). You can check more info here.
RabbitTemplate is used to convert and send a message using RabbitMQ. Is a helper class, as many other Template classes existing in Spring (such as JdbcTemplate , RestTemplate , etc.). Spring Boot creates a predefined one for you, but in this case, we need to tune it a little bit to make it use JSON converter when producing messages.
The exchange and the routing key are AMQP concepts that I recommend you to read about that if you still don’t know what they do. As a summary, we are sending messages through a given channel (exchange), and the routing key allows us to filter for who are those generated. This sample application is not intended for showing advanced exchange / queues configuration so you will see that this is a basic setup.

Receiving messages

Of course, we need to build the other part of the communication too. For this post, I’ve created the receivers in the same application for better logistics (having a single repo), but you shouldn’t have problems to replicate this configuration with two different apps if you want. Actually, you could run this Spring Boot application twice and remove the producer in one instance, the consumers in another one, to make it more like a ‘real case’.

@Service
public class CustomMessageListener {

    private static final Logger log = LoggerFactory.getLogger(CustomMessageListener.class);

    @RabbitListener(queues = MessagingApplication.QUEUE_GENERIC_NAME)
    public void receiveMessage(final Message message) {
        log.info("Received message as generic: {}", message.toString());
    }

    @RabbitListener(queues = MessagingApplication.QUEUE_SPECIFIC_NAME)
    public void receiveMessage(final CustomMessage customMessage) {
        log.info("Received message as specific class: {}", customMessage.toString());
    }
}

@Service

public class CustomMessageListener {

private static final Logger log = LoggerFactory.getLogger(CustomMessageListener.class);

@RabbitListener(queues = MessagingApplication.QUEUE_GENERIC_NAME)

public void receiveMessage(final Message message) {

log.info("Received message as generic: {}", message.toString());

}

@RabbitListener(queues = MessagingApplication.QUEUE_SPECIFIC_NAME)

public void receiveMessage(final CustomMessage customMessage) {

log.info("Received message as specific class: {}", customMessage.toString());

}

Here you can see our goal: the idea is to show at the same time how the same message, being sent by the CustomMessageSender class, can be received through two different queues. First queue (GENERIC) is receiving the message as a generic Message class, so there is no implicit conversion. Spoiler alert: when we log the message we should see something like a JSON string in the payload since it has not been converted.

On the other hand, the queue SPECIFIC is expecting a CustomMessage class (the method argument). That will trigger a logic inside Spring to find a converter from JSON to that specific class. This part should be easy to configure, but sometimes is the part in which we the developers struggle most. In this case is easy to guess that the text logged should be the CustomMessage’s implementation of toString().

Configuration

This is the most interesting part since it’s what is making everything work together. First, you will find some annotations in the class:

@SpringBootApplication
@EnableRabbit
@EnableScheduling
public class MessagingApplication implements RabbitListenerConfigurer {

@SpringBootApplication

@EnableRabbit

@EnableScheduling

public class MessagingApplication implements RabbitListenerConfigurer {

Apart from the one that makes Spring Boot works, @EnableRabbit is making sure that Rabbit Listeners and other Rabbit configurations are detected and processed. @EnableScheduling needs to be added to support the @Scheduled method for sending messages that we covered before in CustomMessageSender class.

Then you’ll find some constants and, next to them, the Exchange / Queues configuration:

@Bean
public TopicExchange appExchange() {
    return new TopicExchange(EXCHANGE_NAME);
}

@Bean
public Queue appQueueGeneric() {
    return new Queue(QUEUE_GENERIC_NAME);
}

@Bean
public Queue appQueueSpecific() {
    return new Queue(QUEUE_SPECIFIC_NAME);
}

@Bean
public Binding declareBindingGeneric() {
    return BindingBuilder.bind(appQueueGeneric()).to(appExchange()).with(ROUTING_KEY);
}

@Bean
public Binding declareBindingSpecific() {
    return BindingBuilder.bind(appQueueSpecific()).to(appExchange()).with(ROUTING_KEY);
}

@Bean

public TopicExchange appExchange() {

return new TopicExchange(EXCHANGE_NAME);

}

@Bean

public Queue appQueueGeneric() {

return new Queue(QUEUE_GENERIC_NAME);

}

@Bean

public Queue appQueueSpecific() {

return new Queue(QUEUE_SPECIFIC_NAME);

}

@Bean

public Binding declareBindingGeneric() {

return BindingBuilder.bind(appQueueGeneric()).to(appExchange()).with(ROUTING_KEY);

}

@Bean

public Binding declareBindingSpecific() {

return BindingBuilder.bind(appQueueSpecific()).to(appExchange()).with(ROUTING_KEY);

}

These are the topic exchange, both queues and the bindings between them (both queues are bound to the same exchange with the same routing key). Just a sample configuration.

Down in the same class, we can find the configuration related to producing messages:

@Bean
public RabbitTemplate rabbitTemplate(final ConnectionFactory connectionFactory) {
    final RabbitTemplate rabbitTemplate = new RabbitTemplate(connectionFactory);
    rabbitTemplate.setMessageConverter(producerJackson2MessageConverter());
    return rabbitTemplate;
}

@Bean
public Jackson2JsonMessageConverter producerJackson2MessageConverter() {
    return new Jackson2JsonMessageConverter();
}

@Bean

public RabbitTemplate rabbitTemplate(final ConnectionFactory connectionFactory) {

final RabbitTemplate rabbitTemplate = new RabbitTemplate(connectionFactory);

rabbitTemplate.setMessageConverter(producerJackson2MessageConverter());

return rabbitTemplate;

}

@Bean

public Jackson2JsonMessageConverter producerJackson2MessageConverter() {

return new Jackson2JsonMessageConverter();

}

The only thing we’re doing here is overriding the default RabbitTemplate created by Spring Boot with another one that will use a message converter of type Jackson2JsonMessageConverter. Remember: this configuration only affects to messages being sent by RabbitTemplate.

And then we need to configure how our application should interpret the received messages. You need to implement a method from the interface RabbitListenerConfigurer, responsible for handling the RabbitListener annotations. As you can see below, we need to configure the MessageHandlerMethodFactory to be not plainly the default one, but using a JSON message converter. Take into account that this converter is not the same as configured for producing messages since we’re using different classes.

@Bean
public MappingJackson2MessageConverter consumerJackson2MessageConverter() {
   return new MappingJackson2MessageConverter();
}

@Bean
public DefaultMessageHandlerMethodFactory messageHandlerMethodFactory() {
   DefaultMessageHandlerMethodFactory factory = new DefaultMessageHandlerMethodFactory();
   factory.setMessageConverter(consumerJackson2MessageConverter());
   return factory;
}

@Override
public void configureRabbitListeners(final RabbitListenerEndpointRegistrar registrar) {
   registrar.setMessageHandlerMethodFactory(messageHandlerMethodFactory());
}

@Bean

public MappingJackson2MessageConverter consumerJackson2MessageConverter() {

return new MappingJackson2MessageConverter();

}

@Bean

public DefaultMessageHandlerMethodFactory messageHandlerMethodFactory() {

DefaultMessageHandlerMethodFactory factory = new DefaultMessageHandlerMethodFactory();

factory.setMessageConverter(consumerJackson2MessageConverter());

return factory;

}

@Override

public void configureRabbitListeners(final RabbitListenerEndpointRegistrar registrar) {

registrar.setMessageHandlerMethodFactory(messageHandlerMethodFactory());

}

Running the application

First, you need to make sure that the RabbitMQ server is running. In the repository, there is a docker-compose.yml file that you can use if you have Docker installed. Just run:

docker-compose up -d

1	docker-compose up -d

Also, if you want to use a configuration which is different from the default one, make sure you include those changes within the application.properties file (port number, authentication, etc.). You don’t need to do this if you use the default settings as they are set in this sample codebase.

Then you can run the application class from your IDE or by executing mvn spring-boot:run from the command line. When you do that, you should see these lines in the console:

INFO 13100 --- [pool-4-thread-1] c.t.rabbitmqconfig.CustomMessageSender   : Sending message...
INFO 13100 --- [cTaskExecutor-1] c.t.r.CustomMessageListener              : Received message as specific class: CustomMessage{text='Hello there!', priority=40, secret=false}
INFO 13100 --- [cTaskExecutor-1] c.t.r.CustomMessageListener              : Received message as generic: (Body:'{"text":"Hello there!","priority":40,"secret":false}' MessageProperties [headers={__TypeId__=com.thepracticaldeveloper.rabbitmqconfig.CustomMessage}, timestamp=null, messageId=null, userId=null, receivedUserId=null, appId=null, clusterId=null, type=null, correlationId=null, correlationIdString=null, replyTo=null, contentType=application/json, contentEncoding=UTF-8, contentLength=0, deliveryMode=null, receivedDeliveryMode=PERSISTENT, expiration=null, priority=0, redelivered=false, receivedExchange=appExchange, receivedRoutingKey=messages.key, receivedDelay=null, deliveryTag=3, messageCount=0, consumerTag=amq.ctag-cZstJRR8omlFYGJDJ0zvcA, consumerQueue=appGenericQueue])

INFO 13100 --- [pool-4-thread-1] c.t.rabbitmqconfig.CustomMessageSender : Sending message...

INFO 13100 --- [cTaskExecutor-1] c.t.r.CustomMessageListener : Received message as specific class: CustomMessage{text='Hello there!', priority=40, secret=false}

INFO 13100 --- [cTaskExecutor-1] c.t.r.CustomMessageListener : Received message as generic: (Body:'{"text":"Hello there!","priority":40,"secret":false}' MessageProperties [headers={__TypeId__=com.thepracticaldeveloper.rabbitmqconfig.CustomMessage}, timestamp=null, messageId=null, userId=null, receivedUserId=null, appId=null, clusterId=null, type=null, correlationId=null, correlationIdString=null, replyTo=null, contentType=application/json, contentEncoding=UTF-8, contentLength=0, deliveryMode=null, receivedDeliveryMode=PERSISTENT, expiration=null, priority=0, redelivered=false, receivedExchange=appExchange, receivedRoutingKey=messages.key, receivedDelay=null, deliveryTag=3, messageCount=0, consumerTag=amq.ctag-cZstJRR8omlFYGJDJ0zvcA, consumerQueue=appGenericQueue])

Working as expected! Our listener for CustomMessage is showing us our implementation of toString() method. The last line though, the method listening for the Message class, is showing us the raw contents, which turns out to be –of course– JSON format.

Changing from JSON format to default

I invite you to comment the lines that set up the producer and consumer message converters (they are indicated in the source code). Then, remove the implements clause from the configuration class. What you will see then is the default configuration for Spring Boot, which is using a Java-specific format to encode messages.

INFO 6660 --- [pool-4-thread-1] c.t.rabbitmqconfig.CustomMessageSender   : Sending message...
INFO 6660 --- [cTaskExecutor-1] c.t.r.CustomMessageListener              : Received message as generic: (Body:'[B@56f321e7(byte[143])' MessageProperties [headers={}, timestamp=null, messageId=null, userId=null, receivedUserId=null, appId=null, clusterId=null, type=null, correlationId=null, correlationIdString=null, replyTo=null, contentType=application/x-java-serialized-object, contentEncoding=null, contentLength=0, deliveryMode=null, receivedDeliveryMode=PERSISTENT, expiration=null, priority=0, redelivered=false, receivedExchange=appExchange, receivedRoutingKey=messages.key, receivedDelay=null, deliveryTag=3, messageCount=0, consumerTag=amq.ctag-ovFVyrVodTF4fWW7_PELaQ, consumerQueue=appGenericQueue])
INFO 6660 --- [cTaskExecutor-1] c.t.r.CustomMessageListener              : Received message as specific class: CustomMessage{text='Hello there!', priority=45, secret=false}

INFO 6660 --- [pool-4-thread-1] c.t.rabbitmqconfig.CustomMessageSender : Sending message...

INFO 6660 --- [cTaskExecutor-1] c.t.r.CustomMessageListener : Received message as generic: (Body:'[[email protected](byte[143])' MessageProperties [headers={}, timestamp=null, messageId=null, userId=null, receivedUserId=null, appId=null, clusterId=null, type=null, correlationId=null, correlationIdString=null, replyTo=null, contentType=application/x-java-serialized-object, contentEncoding=null, contentLength=0, deliveryMode=null, receivedDeliveryMode=PERSISTENT, expiration=null, priority=0, redelivered=false, receivedExchange=appExchange, receivedRoutingKey=messages.key, receivedDelay=null, deliveryTag=3, messageCount=0, consumerTag=amq.ctag-ovFVyrVodTF4fWW7_PELaQ, consumerQueue=appGenericQueue])

INFO 6660 --- [cTaskExecutor-1] c.t.r.CustomMessageListener : Received message as specific class: CustomMessage{text='Hello there!', priority=45, secret=false}

As you can see in the headers of the message, now the format is contentType=application/x-java-serialized-object. This is just another approach, which could turn out to be not the best one if you are using, for instance, different kind of applications (non-Java) for receiving messages.

More info

I hope you find this guide useful. If you are interested in using RabbitMQ in a Microservice Architecture to support an Event-Driven Architecture, and some other related topics like Service Discovery, Routing, Async vs. Sync communication, etc., you can get my book Learn Microservices with Spring Boot (also available on Apress), which will guide you step by step to get real experience.

Get the book on the Apress Store or Amazon

How to deploy a Spring Boot WAR to Wildfly, JBoss ... Deploying a Spring Boot WAR to Wildfly, JBoss or Tomcat Updated: Code examples use Wildfly 13, Java 10 and Spring Boot 2. More info about code examp...

Full Reactive Stack: Conclusions Full Reactive Stack: Conclusions and Comparison between WebFlux and MVC In this final section, we run the application and see how everything works fo...

21 Comments

Really good article. Thanks
Peter
4 months ago
Nice article. Can we use this code for sending and receiving text messages?
bakhti
5 months ago
1. Author
  Thanks. Yes you can, but I’d stick to a JSON structure anyways. You can always wrap those text messages in a minimal JSON structure so you can extend it in future if you need it.
  Moisés
  5 months ago
Hi, I don’t want to use the @scheduled (fixedDelay=3000L) to send the messages every 3 seconds, I want to send the message exactly once. Can you suggest the changes I would need to make?
Mayday27
5 months ago
1. Author
  It depends on the trigger you want to use to send that message. If you want to start the app, send a message and wait for it to be delivered to shut it down, you could use a CountDownLatch object. You can also expose a REST API and then trigger that message whenever you call it… Feel free to further detail what you want to do on the GitHub repository, maybe I or someone else can help.
  Moisés
  5 months ago
How can I get dynamic queue names here?
Pratik
5 months ago
1. Author
  I don’t know if I got exactly what you meant here. I used annotations in this guide to set up queues and exchanges but you can always create queues programmatically with Java, and that way you’d get dynamic queues.
  Moisés
  5 months ago
Very interesting article. I have done it before but with Spring Cloud Stream (using Spring Boot and RabbitMQ). Do you know what are the advantages of one approach over the other?
jedrus07
6 months ago
1. Author
  Thanks! Spring Cloud Stream is an abstraction, implementing the pub-sub pattern in a generic way so you can pick the binder later without (in theory) changing much of your code: say RabbitMQ or Kafka. It might be interesting for example for people who are familiar with Spring Integration since it has native integration with it. As an abstraction though, it has also the inconvenient that if you want to use specific features, you have to write vendor-specific code.
  Moisés
  6 months ago
Thanks Moisés, very helpful
Anonymous
9 months ago
Thanks, but whenever I start Spring application, it logs that application started twice. Do you have any idea why this is happening?
SanketKD
9 months ago
1. Author
  Could you report an issue in the Github repo attaching the stack trace? I can have a look at it.
  Moisés
  9 months ago
  1. I have asked this question here: https://stackoverflow.com/q/47344882/8955751.
    You can find details there.
    I am not getting any stacktrace, since application is running correctly.
    Only my concern is, why spring application is logging that it started twice.
    SanketKD
    9 months ago
Thanks, can u do Event Sourcing with RabbitMq
Anonymous
1 year ago
1. Author
  Technically you can, but you still need the event store so it’s better to go for a framework that can provide you all you need, you’ll get a better integration. Have a look at https://www.confluent.io/blog/event-sourcing-cqrs-stream-processing-apache-kafka-whats-connection/, about how to do Event Sourcing with Kafka.
  Moisés
  1 year ago
Thank you ver much cheers!!!!
aditya prakash Bv
1 year ago
Nice post! It helps us a lot with migration of our rabbits to spring boot
Peter
1 year ago
Thank you very much for the article ! It was really helpful.
Maneesh Mohanan
1 year ago
Hi MOISÉS, You saved my day! Wonderful post…
Naveen Kumar K
1 year ago
Thanks for this article, I’m relatively new to spring and I found it very useful. Good luck with your book 🙂
paris
1 year ago
1. Author
  Thanks for your feedback! I’ll try to cover this and some other topics on my book so feel free to show your interest on Leanpub if you want to get a good discount!
  Moisés
  1 year ago