The English version of is the official project site. Translated sites are community supported on a best-effort basis.

Getting Started to SmallRye Reactive Messaging with Apache Kafka

This guide demonstrates how your Quarkus application can utilize SmallRye Reactive Messaging to interact with Apache Kafka.



  • 大概15分钟

  • 编辑器

  • 安装JDK 11以上版本并正确配置了 JAVA_HOME

  • Apache Maven 3.8.1+

  • Docker and Docker Compose or Podman, and Docker Compose

  • 如果你愿意的话,还可以选择使用Quarkus CLI

  • 如果你想构建原生可执行程序,可以选择安装Mandrel或者GraalVM,并正确配置(或者使用Docker在容器中进行构建)


In this guide, we are going to develop two applications communicating with Kafka. The first application sends a quote request to Kafka and consumes Kafka messages from the quote topic. The second application receives the quote request and sends a quote back.


The first application, the producer, will let the user request some quotes over an HTTP endpoint. For each quote request a random identifier is generated and returned to the user, to mark the quote request as pending. At the same time, the generated request id is sent over a Kafka topic quote-requests.


The second application, the processor, will read from the quote-requests topic, put a random price to the quote, and send it to a Kafka topic named quotes.

Lastly, the producer will read the quotes and send them to the browser using server-sent events. The user will therefore see the quote price updated from pending to the received price in real-time.


We recommend that you follow the instructions in the next sections and create applications step by step. However, you can go right to the completed example.

Clone the Git repository: git clone, or download an archive.

The solution is located in the kafka-quickstart directory.

Creating the Maven Project

First, we need to create two projects: the producer and the processor.

To create the producer project, in a terminal run:

quarkus create app org.acme:kafka-quickstart-producer \
    --extension=resteasy-reactive-jackson,smallrye-reactive-messaging-kafka \

创建Grade项目,请添加 --gradle 或者 --gradle-kotlin-dsl 参数。

关于如何安装并使用Quarkus CLI的更多信息,请参考Quarkus CLI指南

mvn io.quarkus.platform:quarkus-maven-plugin:2.11.2.Final:create \
    -DprojectGroupId=org.acme \
    -DprojectArtifactId=kafka-quickstart-producer \
    -Dextensions="resteasy-reactive-jackson,smallrye-reactive-messaging-kafka" \

创建Grade项目,请添加 -DbuildTool=gradle 或者 -DbuildTool=gradle-kotlin-dsl 参数。

This command creates the project structure and selects two Quarkus extensions we will be using:

  1. RESTEasy Reactive and its Jackson support (to handle JSON) to serve the HTTP endpoint.

  2. The Kafka connector for Reactive Messaging

To create the processor project, from the same directory, run:

quarkus create app org.acme:kafka-quickstart-processor \
    --extension=smallrye-reactive-messaging-kafka \

创建Grade项目,请添加 --gradle 或者 --gradle-kotlin-dsl 参数。

关于如何安装并使用Quarkus CLI的更多信息,请参考Quarkus CLI指南

mvn io.quarkus.platform:quarkus-maven-plugin:2.11.2.Final:create \
    -DprojectGroupId=org.acme \
    -DprojectArtifactId=kafka-quickstart-processor \
    -Dextensions="smallrye-reactive-messaging-kafka" \

创建Grade项目,请添加 -DbuildTool=gradle 或者 -DbuildTool=gradle-kotlin-dsl 参数。

At that point, you should have the following structure:

├── kafka-quickstart-processor
│  ├──
│  ├── mvnw
│  ├── mvnw.cmd
│  ├── pom.xml
│  └── src
│     └── main
│        ├── docker
│        ├── java
│        └── resources
│           └──
└── kafka-quickstart-producer
   ├── mvnw
   ├── mvnw.cmd
   ├── pom.xml
   └── src
      └── main
         ├── docker
         ├── java
         └── resources

Open the two projects in your favorite IDE.

Dev Services

No need to start a Kafka broker when using the dev mode or for tests. Quarkus starts a broker for you automatically. See Dev Services for Kafka for details.

The Quote object

The Quote class will be used in both producer and processor projects. For the sake of simplicity, we will duplicate the class. In both projects, create the src/main/java/org/acme/kafka/model/ file, with the following content:

package org.acme.kafka.model;

public class Quote {

    public String id;
    public int price;

    * Default constructor required for Jackson serializer
    public Quote() { }

    public Quote(String id, int price) { = id;
        this.price = price;

    public String toString() {
        return "Quote{" +
                "id='" + id + '\'' +
                ", price=" + price +

JSON representation of Quote objects will be used in messages sent to the Kafka topic and also in the server-sent events sent to web browsers.

Quarkus has built-in capabilities to deal with JSON Kafka messages. In a following section, we will create serializer/deserializer classes for Jackson.

Sending quote request

Inside the producer project, create the src/main/java/org/acme/kafka/producer/ file and add the following content:

package org.acme.kafka.producer;

import java.util.UUID;


import org.acme.kafka.model.Quote;
import org.eclipse.microprofile.reactive.messaging.Channel;
import org.eclipse.microprofile.reactive.messaging.Emitter;

public class QuotesResource {

    Emitter<String> quoteRequestEmitter; (1)

     * Endpoint to generate a new quote request id and send it to "quote-requests" Kafka topic using the emitter.
    public String createRequest() {
        UUID uuid = UUID.randomUUID();
        quoteRequestEmitter.send(uuid.toString()); (2)
        return uuid.toString(); (3)
1 Inject a Reactive Messaging Emitter to send messages to the quote-requests channel.
2 On a post request, generate a random UUID and send it to the Kafka topic using the emitter.
3 Return the same UUID to the client.

The quote-requests channel is going to be managed as a Kafka topic, as that’s the only connector on the classpath. If not indicated otherwise, like in this example, Quarkus uses the channel name as topic name. So, in this example, the application writes into the quote-requests topic. Quarkus also configures the serializer automatically, because it finds that the Emitter produces String values.

When you have multiple connectors, you would need to indicate which connector you want to use in the application configuration.

Processing quote requests

Now let’s consume the quote request and give out a price. Inside the processor project, create the src/main/java/org/acme/kafka/processor/ file and add the following content:

package org.acme.kafka.processor;

import java.util.Random;

import javax.enterprise.context.ApplicationScoped;

import org.acme.kafka.model.Quote;
import org.eclipse.microprofile.reactive.messaging.Incoming;
import org.eclipse.microprofile.reactive.messaging.Outgoing;

import io.smallrye.reactive.messaging.annotations.Blocking;

 * A bean consuming data from the "quote-requests" Kafka topic (mapped to "requests" channel) and giving out a random quote.
 * The result is pushed to the "quotes" Kafka topic.
public class QuotesProcessor {

    private Random random = new Random();

    @Incoming("requests") (1)
    @Outgoing("quotes")   (2)
    @Blocking             (3)
    public Quote process(String quoteRequest) throws InterruptedException {
        // simulate some hard working task
        return new Quote(quoteRequest, random.nextInt(100));
1 Indicates that the method consumes the items from the requests channel.
2 Indicates that the objects returned by the method are sent to the quotes channel.
3 Indicates that the processing is blocking and cannot be run on the caller thread.

For every Kafka record from the quote-requests topic, Reactive Messaging calls the process method, and sends the returned Quote object to the quotes channel. In this case, we need to configure the channel in the file, to configures the requests and quotes channels:


# Configure the incoming `quote-requests` Kafka topic

Note that in this case we have one incoming and one outgoing connector configuration, each one distinctly named. The configuration keys are structured as follows:


The channel-name segment must match the value set in the @Incoming and @Outgoing annotation:

  • quote-requests → Kafka topic from which we read the quote requests

  • quotes → Kafka topic in which we write the quotes

More details about this configuration is available on the Producer configuration and Consumer configuration section from the Kafka documentation. These properties are configured with the prefix kafka. An exhaustive list of configuration properties is available in Kafka Reference Guide - Configuration. instructs the application to start reading the topics from the first offset, when there is no committed offset for the consumer group. In other words, it will also process messages sent before we start the processor application.

There is no need to set serializers or deserializers. Quarkus detects them, and if none are found, generates them using JSON serialization.

Receiving quotes

Back to our producer project. Let’s modify the QuotesResource to consume quotes from Kafka and send them back to the client via Server-Sent Events:

import io.smallrye.mutiny.Multi;


Multi<Quote> quotes; (1)

 * Endpoint retrieving the "quotes" Kafka topic and sending the items to a server sent event.
@Produces(MediaType.SERVER_SENT_EVENTS) (2)
public Multi<Quote> stream() {
    return quotes; (3)
1 Injects the quotes channel using the @Channel qualifier
2 Indicates that the content is sent using Server Sent Events
3 Returns the stream (Reactive Stream)

No need to configure anything, as Quarkus will automatically associate the quotes channel to the quotes Kafka topic. It will also generate a deserializer for the Quote class.

Message serialization in Kafka

In this example we used Jackson to serialize/deserialize Kafka messages. For more options on message serialization, see Kafka Reference Guide - Serialization.

We strongly suggest adopting a contract-first approach using a schema registry. To learn more about how to use Apache Kafka with the schema registry and Avro, follow the Using Apache Kafka with Schema Registry and Avro guide.

The HTML page

Final touch, the HTML page requesting quotes and displaying the prices obtained over SSE.

Inside the producer project, create the src/main/resources/META-INF/resources/quotes.html file with the following content:

<!DOCTYPE html>
<html lang="en">
    <meta charset="UTF-8">

    <link rel="stylesheet" type="text/css"
    <link rel="stylesheet" type="text/css"
<div class="container">
    <div class="card">
        <div class="card-body">
            <h2 class="card-title">Quotes</h2>
            <button class="btn btn-info" id="request-quote">Request Quote</button>
            <div class="quotes"></div>
<script src=""></script>
    $("#request-quote").click((event) => {
        fetch("/quotes/request", {method: "POST"})
        .then(res => res.text())
        .then(qid => {
            var row = $(`<h4 class='col-md-12' id='${qid}'>Quote # <i>${qid}</i> | <strong>Pending</strong></h4>`);

    var source = new EventSource("/quotes");
    source.onmessage = (event) => {
      var json = JSON.parse(;
      $(`#${}`).html((index, html) => {
        return html.replace("Pending", `\$\xA0${json.price}`);

Nothing spectacular here. When the user clicks the button, HTTP request is made to request a quote, and a pending quote is added to the list. On each quote received over SSE, the corresponding item in the list is updated.

Get it running

You just need to run both applications. In one terminal, run:

mvn -f kafka-quickstart-producer quarkus:dev

In another terminal, run:

mvn -f kafka-quickstart-processor quarkus:dev

Quarkus starts a Kafka broker automatically, configures the application and shares the Kafka broker instance between different applications. See Dev Services for Kafka for more details.

Open http://localhost:8080/quotes.html in your browser and request some quotes by clicking the button.

Running in JVM or Native mode

When not running in dev or test mode, you will need to start your Kafka broker. You can follow the instructions from the Apache Kafka website or create a docker-compose.yaml file with the following content:

version: '3.5'


    command: [
      "sh", "-c",
      "bin/ config/"
      - "2181:2181"
      LOG_DIR: /tmp/logs
      - kafka-quickstart-network

    command: [
      "sh", "-c",
      "bin/ config/ --override listeners=$${KAFKA_LISTENERS} --override advertised.listeners=$${KAFKA_ADVERTISED_LISTENERS} --override zookeeper.connect=$${KAFKA_ZOOKEEPER_CONNECT}"
      - zookeeper
      - "9092:9092"
      LOG_DIR: "/tmp/logs"
      KAFKA_ZOOKEEPER_CONNECT: zookeeper:2181
      - kafka-quickstart-network

    image: quarkus-quickstarts/kafka-quickstart-producer:1.0-${QUARKUS_MODE:-jvm}
      context: producer
      dockerfile: src/main/docker/Dockerfile.${QUARKUS_MODE:-jvm}
      - kafka
      - "8080:8080"
      - kafka-quickstart-network

    image: quarkus-quickstarts/kafka-quickstart-processor:1.0-${QUARKUS_MODE:-jvm}
      context: processor
      dockerfile: src/main/docker/Dockerfile.${QUARKUS_MODE:-jvm}
      - kafka
      - kafka-quickstart-network

    name: kafkaquickstart

Make sure you first build both applications in JVM mode with:

mvn -f kafka-quickstart-producer package
mvn -f kafka-quickstart-processor package

Once packaged, run docker-compose up.

This is a development cluster, do not use in production.

You can also build and run our applications as native executables. First, compile both applications as native:

mvn -f kafka-quickstart-producer package -Dnative -Dquarkus.native.container-build=true
mvn -f kafka-quickstart-processor package -Dnative -Dquarkus.native.container-build=true

Run the system with:

export QUARKUS_MODE=native
docker-compose up --build


This guide has shown how you can interact with Kafka using Quarkus. It utilizes SmallRye Reactive Messaging to build data streaming applications.

For the exhaustive list of features and configuration options, check the Reference guide for Apache Kafka Extension.

In this guide we explore Smallrye Reactive Messaging framework to interact with Apache Kafka. Quarkus extension for Kafka also allows using Kafka clients directly.