2.6.3

© 2016 - 2020 by Pivotal Software, Inc.

Copies of this document may be made for your own use and for distribution to others, provided that you do not charge any fee for such copies and further provided that each copy contains this Copyright Notice, whether distributed in print or electronically.

1. Preface

The Spring for Apache Kafka project applies core Spring concepts to the development of Kafka-based messaging solutions. We provide a “template” as a high-level abstraction for sending messages. We also provide support for Message-driven POJOs.

2. What’s new?

2.1. What’s New in 2.6 Since 2.5

This section covers the changes made from version 2.5 to version 2.6. For changes in earlier version, see Change History.

2.1.1. Kafka Client Version

This version requires the 2.6.0 kafka-clients.

2.1.2. Listener Container Changes

The default EOSMode is now BETA. See Exactly Once Semantics for more information.

Various error handlers (that extend FailedRecordProcessor) and the DefaultAfterRollbackProcessor now reset the BackOff if recovery fails. In addition, you can now select the BackOff to use based on the failed record and/or exception. See Seek To Current Container Error Handlers, Recovering Batch Error Handler, Publishing Dead-letter Records and After-rollback Processor for more information.

You can now configure an adviceChain in the container properties. See Listener Container Properties for more information.

When the container is configured to publish ListenerContainerIdleEvent s, it now publishes a ListenerContainerNoLongerIdleEvent when a record is received after publishing an idle event. See Application Events and Detecting Idle and Non-Responsive Consumers for more information.

2.1.3. @KafkaListener Changes

When using manual partition assignment, you can now specify a wildcard for determining which partitions should be reset to the initial offset. In addition, if the listener implements ConsumerSeekAware, onPartitionsAssigned() is called after the manual assignment. (Also added in version 2.5.5). See Explicit Partition Assignment for more information.

Convenience methods have been added to AbstractConsumerSeekAware to make seeking easier. See Seeking to a Specific Offset for more information.

2.1.4. ErrorHandler Changes

Subclasses of FailedRecordProcessor (e.g. SeekToCurrentErrorHandler, DefaultAfterRollbackProcessor, RecoveringBatchErrorHandler) can now be configured to reset the retry state if the exception is a different type to that which occurred previously with this record. See Seek To Current Container Error Handlers, After-rollback Processor, Recovering Batch Error Handler for more information.

2.1.5. Producer Factory Changes

You can now set a maximum age for producers after which they will be closed and recreated. See Transactions for more information.

3. Introduction

This first part of the reference documentation is a high-level overview of Spring for Apache Kafka and the underlying concepts and some code snippets that can help you get up and running as quickly as possible.

3.1. Quick Tour for the Impatient

This is the five-minute tour to get started with Spring Kafka.

Prerequisites: You must install and run Apache Kafka. Then you must grab the spring-kafka JAR and all of its dependencies. The easiest way to do that is to declare a dependency in your build tool. The following example shows how to do so with Maven:

<dependency>
  <groupId>org.springframework.kafka</groupId>
  <artifactId>spring-kafka</artifactId>
  <version>2.6.3</version>
</dependency>

The following example shows how to do so with Gradle:

compile 'org.springframework.kafka:spring-kafka:2.6.3'
When using Spring Boot, omit the version and Boot will automatically bring in the correct version that is compatible with your Boot version:
<dependency>
  <groupId>org.springframework.kafka</groupId>
  <artifactId>spring-kafka</artifactId>
</dependency>

The following example shows how to do so with Gradle:

compile 'org.springframework.kafka:spring-kafka'

3.1.1. Compatibility

This quick tour works with the following versions:

  • Apache Kafka Clients 2.4.1

  • Spring Framework 5.3.x

  • Minimum Java version: 8

3.1.2. A Very, Very Quick Example

As the following example shows, you can use plain Java to send and receive a message:

@Test
public void testAutoCommit() throws Exception {
    logger.info("Start auto");
    ContainerProperties containerProps = new ContainerProperties("topic1", "topic2");
    final CountDownLatch latch = new CountDownLatch(4);
    containerProps.setMessageListener(new MessageListener<Integer, String>() {

        @Override
        public void onMessage(ConsumerRecord<Integer, String> message) {
            logger.info("received: " + message);
            latch.countDown();
        }

    });
    KafkaMessageListenerContainer<Integer, String> container = createContainer(containerProps);
    container.setBeanName("testAuto");
    container.start();
    Thread.sleep(1000); // wait a bit for the container to start
    KafkaTemplate<Integer, String> template = createTemplate();
    template.setDefaultTopic("topic1");
    template.sendDefault(0, "foo");
    template.sendDefault(2, "bar");
    template.sendDefault(0, "baz");
    template.sendDefault(2, "qux");
    template.flush();
    assertTrue(latch.await(60, TimeUnit.SECONDS));
    container.stop();
    logger.info("Stop auto");

}
private KafkaMessageListenerContainer<Integer, String> createContainer(
                        ContainerProperties containerProps) {
    Map<String, Object> props = consumerProps();
    DefaultKafkaConsumerFactory<Integer, String> cf =
                            new DefaultKafkaConsumerFactory<Integer, String>(props);
    KafkaMessageListenerContainer<Integer, String> container =
                            new KafkaMessageListenerContainer<>(cf, containerProps);
    return container;
}

private KafkaTemplate<Integer, String> createTemplate() {
    Map<String, Object> senderProps = senderProps();
    ProducerFactory<Integer, String> pf =
              new DefaultKafkaProducerFactory<Integer, String>(senderProps);
    KafkaTemplate<Integer, String> template = new KafkaTemplate<>(pf);
    return template;
}

private Map<String, Object> consumerProps() {
    Map<String, Object> props = new HashMap<>();
    props.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
    props.put(ConsumerConfig.GROUP_ID_CONFIG, group);
    props.put(ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG, true);
    props.put(ConsumerConfig.AUTO_COMMIT_INTERVAL_MS_CONFIG, "100");
    props.put(ConsumerConfig.SESSION_TIMEOUT_MS_CONFIG, "15000");
    props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, IntegerDeserializer.class);
    props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
    return props;
}

private Map<String, Object> senderProps() {
    Map<String, Object> props = new HashMap<>();
    props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
    props.put(ProducerConfig.RETRIES_CONFIG, 0);
    props.put(ProducerConfig.BATCH_SIZE_CONFIG, 16384);
    props.put(ProducerConfig.LINGER_MS_CONFIG, 1);
    props.put(ProducerConfig.BUFFER_MEMORY_CONFIG, 33554432);
    props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, IntegerSerializer.class);
    props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, StringSerializer.class);
    return props;
}

3.1.3. With Java Configuration

You can do the same work as appears in the previous example with Spring configuration in Java. The following example shows how to do so:

@Autowired
private Listener listener;

@Autowired
private KafkaTemplate<Integer, String> template;

@Test
public void testSimple() throws Exception {
    template.send("annotated1", 0, "foo");
    template.flush();
    assertTrue(this.listener.latch1.await(10, TimeUnit.SECONDS));
}

@Configuration
@EnableKafka
public class Config {

    @Bean
    ConcurrentKafkaListenerContainerFactory<Integer, String>
                        kafkaListenerContainerFactory() {
        ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
                                new ConcurrentKafkaListenerContainerFactory<>();
        factory.setConsumerFactory(consumerFactory());
        return factory;
    }

    @Bean
    public ConsumerFactory<Integer, String> consumerFactory() {
        return new DefaultKafkaConsumerFactory<>(consumerConfigs());
    }

    @Bean
    public Map<String, Object> consumerConfigs() {
        Map<String, Object> props = new HashMap<>();
        props.put(ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG, embeddedKafka.getBrokersAsString());
        ...
        return props;
    }

    @Bean
    public Listener listener() {
        return new Listener();
    }

    @Bean
    public ProducerFactory<Integer, String> producerFactory() {
        return new DefaultKafkaProducerFactory<>(producerConfigs());
    }

    @Bean
    public Map<String, Object> producerConfigs() {
        Map<String, Object> props = new HashMap<>();
        props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, embeddedKafka.getBrokersAsString());
        ...
        return props;
    }

    @Bean
    public KafkaTemplate<Integer, String> kafkaTemplate() {
        return new KafkaTemplate<Integer, String>(producerFactory());
    }

}
public class Listener {

    private final CountDownLatch latch1 = new CountDownLatch(1);

    @KafkaListener(id = "foo", topics = "annotated1")
    public void listen1(String foo) {
        this.latch1.countDown();
    }

}

3.1.4. Even Quicker, with Spring Boot

Spring Boot can make things even simpler. The following Spring Boot application sends three messages to a topic, receives them, and stops:

@SpringBootApplication
public class Application implements CommandLineRunner {

    public static Logger logger = LoggerFactory.getLogger(Application.class);

    public static void main(String[] args) {
        SpringApplication.run(Application.class, args).close();
    }

    @Autowired
    private KafkaTemplate<String, String> template;

    private final CountDownLatch latch = new CountDownLatch(3);

    @Override
    public void run(String... args) throws Exception {
        this.template.send("myTopic", "foo1");
        this.template.send("myTopic", "foo2");
        this.template.send("myTopic", "foo3");
        latch.await(60, TimeUnit.SECONDS);
        logger.info("All received");
    }

    @KafkaListener(topics = "myTopic")
    public void listen(ConsumerRecord<?, ?> cr) throws Exception {
        logger.info(cr.toString());
        latch.countDown();
    }

}

Boot takes care of most of the configuration. When we use a local broker, the only properties we need are the following:

Example 1. application.properties
spring.kafka.consumer.group-id=foo
spring.kafka.consumer.auto-offset-reset=earliest

We need the first property because we are using group management to assign topic partitions to consumers, so we need a group. The second property ensures the new consumer group gets the messages we sent, because the container might start after the sends have completed.

4. Reference

This part of the reference documentation details the various components that comprise Spring for Apache Kafka. The main chapter covers the core classes to develop a Kafka application with Spring.

4.1. Using Spring for Apache Kafka

This section offers detailed explanations of the various concerns that impact using Spring for Apache Kafka. For a quick but less detailed introduction, see Quick Tour for the Impatient.

4.1.1. Connecting to Kafka

Starting with version 2.5, each of these extends KafkaResourceFactory. This allows changing the bootstrap servers at runtime by adding a Supplier<String> to their configuration: setBootstrapServersSupplier(() → …​). This will be called for all new connections to get the list of servers. Consumers and Producers are generally long-lived. To close existing Producers, call reset() on the DefaultKafkaProducerFactory. To close existing Consumers, call stop() (and then start()) on the KafkaListenerEndpointRegistry and/or stop() and start() on any other listener container beans.

For convenience, the framework also provides an ABSwitchCluster which supports two sets of bootstrap servers; one of which is active at any time. Configure the ABSwitchCluster and add it to the producer and consumer factories, and the KafkaAdmin, by calling setBootstrapServersSupplier(). When you want to switch, call primary() or secondary() and call reset() on the producer factory to establish new connection(s); for consumers, stop() and start() all listener containers. When using @KafkaListener s, stop() and start() the KafkaListenerEndpointRegistry bean.

See the Javadocs for more information.

Factory Listeners

Starting with version 2.5, the DefaultKafkaProducerFactory and DefaultKafkaConsumerFactory can be configured with a Listener to receive notifications whenever a producer or consumer is created or closed.

Producer Factory Listener
interface Listener<K, V> {

    default void producerAdded(String id, Producer<K, V> producer) {
    }

    default void producerRemoved(String id, Producer<K, V> producer) {
    }

}
Consumer Factory Listener
interface Listener<K, V> {

    default void consumerAdded(String id, Consumer<K, V> consumer) {
    }

    default void consumerRemoved(String id, Consumer<K, V> consumer) {
    }

}

In each case, the id is created by appending the client-id property (obtained from the metrics() after creation) to the factory beanName property, separated by ..

These listeners can be used, for example, to create and bind a Micrometer KafkaClientMetrics instance when a new client is created (and close it when the client is closed).

The framework provides listeners that do exactly that; see Micrometer Native Metrics.

4.1.2. Configuring Topics

If you define a KafkaAdmin bean in your application context, it can automatically add topics to the broker. To do so, you can add a NewTopic @Bean for each topic to the application context. Version 2.3 introduced a new class TopicBuilder to make creation of such beans more convenient. The following example shows how to do so:

@Bean
public KafkaAdmin admin() {
    Map<String, Object> configs = new HashMap<>();
    configs.put(AdminClientConfig.BOOTSTRAP_SERVERS_CONFIG, ...);
    return new KafkaAdmin(configs);
}

@Bean
public NewTopic topic1() {
    return TopicBuilder.name("thing1")
            .partitions(10)
            .replicas(3)
            .compact()
            .build();
}

@Bean
public NewTopic topic2() {
    return TopicBuilder.name("thing2")
            .partitions(10)
            .replicas(3)
            .config(TopicConfig.COMPRESSION_TYPE_CONFIG, "zstd")
            .build();
}

@Bean
public NewTopic topic3() {
    return TopicBuilder.name("thing3")
            .assignReplicas(0, Arrays.asList(0, 1))
            .assignReplicas(1, Arrays.asList(1, 2))
            .assignReplicas(2, Arrays.asList(2, 0))
            .config(TopicConfig.COMPRESSION_TYPE_CONFIG, "zstd")
            .build();
}

Starting with version 2.6, you can omit .partitions() and/or replicas() and the broker defaults will be applied to those properties. The broker version must be at least 2.4.0 to support this feature - see KIP-464.

@Bean
public NewTopic topic4() {
    return TopicBuilder.name("defaultBoth")
            .build();
}

@Bean
public NewTopic topic5() {
    return TopicBuilder.name("defaultPart")
            .replicas(1)
            .build();
}

@Bean
public NewTopic topic6() {
    return TopicBuilder.name("defaultRepl")
            .partitions(3)
            .build();
}
When using Spring Boot, a KafkaAdmin bean is automatically registered so you only need the NewTopic @Bean s.

By default, if the broker is not available, a message is logged, but the context continues to load. You can programmatically invoke the admin’s initialize() method to try again later. If you wish this condition to be considered fatal, set the admin’s fatalIfBrokerNotAvailable property to true. The context then fails to initialize.

If the broker supports it (1.0.0 or higher), the admin increases the number of partitions if it is found that an existing topic has fewer partitions than the NewTopic.numPartitions.

For more advanced features, you can use the AdminClient directly. The following example shows how to do so:

@Autowired
private KafkaAdmin admin;

...

    AdminClient client = AdminClient.create(admin.getConfigurationProperties());
    ...
    client.close();

4.1.3. Sending Messages

This section covers how to send messages.

Using KafkaTemplate

This section covers how to use KafkaTemplate to send messages.

Overview

The KafkaTemplate wraps a producer and provides convenience methods to send data to Kafka topics. The following listing shows the relevant methods from KafkaTemplate:

ListenableFuture<SendResult<K, V>> sendDefault(V data);

ListenableFuture<SendResult<K, V>> sendDefault(K key, V data);

ListenableFuture<SendResult<K, V>> sendDefault(Integer partition, K key, V data);

ListenableFuture<SendResult<K, V>> sendDefault(Integer partition, Long timestamp, K key, V data);

ListenableFuture<SendResult<K, V>> send(String topic, V data);

ListenableFuture<SendResult<K, V>> send(String topic, K key, V data);

ListenableFuture<SendResult<K, V>> send(String topic, Integer partition, K key, V data);

ListenableFuture<SendResult<K, V>> send(String topic, Integer partition, Long timestamp, K key, V data);

ListenableFuture<SendResult<K, V>> send(ProducerRecord<K, V> record);

ListenableFuture<SendResult<K, V>> send(Message<?> message);

Map<MetricName, ? extends Metric> metrics();

List<PartitionInfo> partitionsFor(String topic);

<T> T execute(ProducerCallback<K, V, T> callback);

// Flush the producer.

void flush();

interface ProducerCallback<K, V, T> {

    T doInKafka(Producer<K, V> producer);

}

See the Javadoc for more detail.

The sendDefault API requires that a default topic has been provided to the template.

The API takes in a timestamp as a parameter and stores this timestamp in the record. How the user-provided timestamp is stored depends on the timestamp type configured on the Kafka topic. If the topic is configured to use CREATE_TIME, the user specified timestamp is recorded (or generated if not specified). If the topic is configured to use LOG_APPEND_TIME, the user-specified timestamp is ignored and the broker adds in the local broker time.

The metrics and partitionsFor methods delegate to the same methods on the underlying Producer. The execute method provides direct access to the underlying Producer.

To use the template, you can configure a producer factory and provide it in the template’s constructor. The following example shows how to do so:

@Bean
public ProducerFactory<Integer, String> producerFactory() {
    return new DefaultKafkaProducerFactory<>(producerConfigs());
}

@Bean
public Map<String, Object> producerConfigs() {
    Map<String, Object> props = new HashMap<>();
    props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
    props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, StringSerializer.class);
    props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, StringSerializer.class);
    // See https://kafka.apache.org/documentation/#producerconfigs for more properties
    return props;
}

@Bean
public KafkaTemplate<Integer, String> kafkaTemplate() {
    return new KafkaTemplate<Integer, String>(producerFactory());
}

Starting with version 2.5, you can now override the factory’s ProducerConfig properties to create templates with different producer configurations from the same factory.

@Bean
public KafkaTemplate<String, String> stringTemplate(ProducerFactory<String, String> pf) {
    return new KafkaTemplate<>(pf);
}

@Bean
public KafkaTemplate<String, byte[]> bytesTemplate(ProducerFactory<String, byte[]> pf) {
    return new KafkaTemplate<>(pf,
            Collections.singletonMap(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, ByteArraySerializer.class));
}

Note that a bean of type ProducerFactory<?, ?> (such as the one auto-configured by Spring Boot) can be referenced with different narrowed generic types.

You can also configure the template by using standard <bean/> definitions.

Then, to use the template, you can invoke one of its methods.

When you use the methods with a Message<?> parameter, the topic, partition, and key information is provided in a message header that includes the following items:

  • KafkaHeaders.TOPIC

  • KafkaHeaders.PARTITION_ID

  • KafkaHeaders.MESSAGE_KEY

  • KafkaHeaders.TIMESTAMP

The message payload is the data.

Optionally, you can configure the KafkaTemplate with a ProducerListener to get an asynchronous callback with the results of the send (success or failure) instead of waiting for the Future to complete. The following listing shows the definition of the ProducerListener interface:

public interface ProducerListener<K, V> {

    void onSuccess(ProducerRecord<K, V> producerRecord, RecordMetadata recordMetadata);

    void onError(ProducerRecord<K, V> producerRecord, RecordMetadata recordMetadata,
            Exception exception);

}

By default, the template is configured with a LoggingProducerListener, which logs errors and does nothing when the send is successful.

For convenience, default method implementations are provided in case you want to implement only one of the methods.

Notice that the send methods return a ListenableFuture<SendResult>. You can register a callback with the listener to receive the result of the send asynchronously. The following example shows how to do so:

ListenableFuture<SendResult<Integer, String>> future = template.send("myTopic", "something");
future.addCallback(new ListenableFutureCallback<SendResult<Integer, String>>() {

    @Override
    public void onSuccess(SendResult<Integer, String> result) {
        ...
    }

    @Override
    public void onFailure(Throwable ex) {
        ...
    }

});

SendResult has two properties, a ProducerRecord and RecordMetadata. See the Kafka API documentation for information about those objects.

The Throwable in onFailure can be cast to a KafkaProducerException; its failedProducerRecord property contains the failed record.

Starting with version 2.5, you can use a KafkaSendCallback instead of a ListenableFutureCallback, making it easier to extract the failed ProducerRecord, avoiding the need to cast the Throwable:

ListenableFuture<SendResult<Integer, String>> future = template.send("topic", 1, "thing");
future.addCallback(new KafkaSendCallback<Integer, String>() {

    @Override
    public void onSuccess(SendResult<Integer, String> result) {
        ...
    }

    @Override
    public void onFailure(KafkaProducerException ex) {
        ProducerRecord<Integer, String> failed = ex.getFailedProducerRecord();
        ...
    }

});

You can also use a pair of lambdas:

ListenableFuture<SendResult<Integer, String>> future = template.send("topic", 1, "thing");
future.addCallback(result -> {
        ...
    }, (KafkaFailureCallback<Integer, String>) ex -> {
            ProducerRecord<Integer, String> failed = ex.getFailedProducerRecord();
            ...
    });

If you wish to block the sending thread to await the result, you can invoke the future’s get() method; using the method with a timeout is recommended. You may wish to invoke flush() before waiting or, for convenience, the template has a constructor with an autoFlush parameter that causes the template to flush() on each send. Flushing is only needed if you have set the linger.ms producer property and want to immediately send a partial batch.

Examples

This section shows examples of sending messages to Kafka:

Example 2. Non Blocking (Async)
public void sendToKafka(final MyOutputData data) {
    final ProducerRecord<String, String> record = createRecord(data);

    ListenableFuture<SendResult<Integer, String>> future = template.send(record);
    future.addCallback(new KafkaSendCallback<SendResult<Integer, String>>() {

        @Override
        public void onSuccess(SendResult<Integer, String> result) {
            handleSuccess(data);
        }

        @Override
        public void onFailure(KafkaProducerException ex) {
            handleFailure(data, record, ex);
        }

    });
}
Blocking (Sync)
public void sendToKafka(final MyOutputData data) {
    final ProducerRecord<String, String> record = createRecord(data);

    try {
        template.send(record).get(10, TimeUnit.SECONDS);
        handleSuccess(data);
    }
    catch (ExecutionException e) {
        handleFailure(data, record, e.getCause());
    }
    catch (TimeoutException | InterruptedException e) {
        handleFailure(data, record, e);
    }
}

Note that the cause of the ExecutionException is KafkaProducerException with the failedProducerRecord property.

Using RoutingKafkaTemplate

Starting with version 2.5, you can use a RoutingKafkaTemplate to select the producer at runtime, based on the destination topic name.

The routing template does not support transactions, execute, flush, or metrics operations because the topic is not known for those operations.

The template requires a map of java.util.regex.Pattern to ProducerFactory<Object, Object> instances. This map should be ordered (e.g. a LinkedHashMap) because it is traversed in order; you should add more specific patterns at the beginning.

The following simple Spring Boot application provides an example of how to use the same template to send to different topics, each using a different value serializer.

@SpringBootApplication
public class Application {

    public static void main(String[] args) {
        SpringApplication.run(Application.class, args);
    }

    @Bean
    public RoutingKafkaTemplate routingTemplate(GenericApplicationContext context,
            ProducerFactory<Object, Object> pf) {

        // Clone the PF with a different Serializer, register with Spring for shutdown
        Map<String, Object> configs = new HashMap<>(pf.getConfigurationProperties());
        configs.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, ByteArraySerializer.class);
        DefaultKafkaProducerFactory<Object, Object> bytesPF = new DefaultKafkaProducerFactory<>(configs);
        context.registerBean(DefaultKafkaProducerFactory.class, "bytesPF", bytesPF);

        Map<Pattern, ProducerFactory<Object, Object>> map = new LinkedHashMap<>();
        map.put(Pattern.compile("two"), bytesPF);
        map.put(Pattern.compile(".+"), pf); // Default PF with StringSerializer
        return new RoutingKafkaTemplate(map);
    }

    @Bean
    public ApplicationRunner runner(RoutingKafkaTemplate routingTemplate) {
        return args -> {
            routingTemplate.send("one", "thing1");
            routingTemplate.send("two", "thing2".getBytes());
        };
    }

}

The corresponding @KafkaListener s for this example are shown in Annotation Properties.

For another technique to achieve similar results, but with the additional capability of sending different types to the same topic, see Delegating Serializer and Deserializer.

Using DefaultKafkaProducerFactory

As seen in Using KafkaTemplate, a ProducerFactory is used to create the producer.

When not using Transactions, by default, the DefaultKafkaProducerFactory creates a singleton producer used by all clients, as recommended in the KafkaProducer javadocs. However, if you call flush() on the template, this can cause delays for other threads using the same producer. Starting with version 2.3, the DefaultKafkaProducerFactory has a new property producerPerThread. When set to true, the factory will create (and cache) a separate producer for each thread, to avoid this issue.

When producerPerThread is true, user code must call closeThreadBoundProducer() on the factory when the producer is no longer needed. This will physically close the producer and remove it from the ThreadLocal. Calling reset() or destroy() will not clean up these producers.

When creating a DefaultKafkaProducerFactory, key and/or value Serializer classes can be picked up from configuration by calling the constructor that only takes in a Map of properties (see example in Using KafkaTemplate), or Serializer instances may be passed to the DefaultKafkaProducerFactory constructor (in which case all Producer s share the same instances). Alternatively you can provide Supplier<Serializer> s (starting with version 2.3) that will be used to obtain separate Serializer instances for each Producer:

@Bean
public ProducerFactory<Integer, CustomValue> producerFactory() {
    return new DefaultKafkaProducerFactory<>(producerConfigs(), null, () -> new CustomValueSerializer());
}

@Bean
public KafkaTemplate<Integer, CustomValue> kafkaTemplate() {
    return new KafkaTemplate<Integer, CustomValue>(producerFactory());
}
Using ReplyingKafkaTemplate

Version 2.1.3 introduced a subclass of KafkaTemplate to provide request/reply semantics. The class is named ReplyingKafkaTemplate and has two additional methods; the following shows the method signatures:

RequestReplyFuture<K, V, R> sendAndReceive(ProducerRecord<K, V> record);

RequestReplyFuture<K, V, R> sendAndReceive(ProducerRecord<K, V> record,
    Duration replyTimeout);

The result is a ListenableFuture that is asynchronously populated with the result (or an exception, for a timeout). The result also has a sendFuture property, which is the result of calling KafkaTemplate.send(). You can use this future to determine the result of the send operation.

If the first method is used, or the replyTimeout argument is null, the template’s defaultReplyTimeout property is used (5 seconds by default).

The following Spring Boot application shows an example of how to use the feature:

@SpringBootApplication
public class KRequestingApplication {

    public static void main(String[] args) {
        SpringApplication.run(KRequestingApplication.class, args).close();
    }

    @Bean
    public ApplicationRunner runner(ReplyingKafkaTemplate<String, String, String> template) {
        return args -> {
            ProducerRecord<String, String> record = new ProducerRecord<>("kRequests", "foo");
            RequestReplyFuture<String, String, String> replyFuture = template.sendAndReceive(record);
            SendResult<String, String> sendResult = replyFuture.getSendFuture().get(10, TimeUnit.SECONDS);
            System.out.println("Sent ok: " + sendResult.getRecordMetadata());
            ConsumerRecord<String, String> consumerRecord = replyFuture.get(10, TimeUnit.SECONDS);
            System.out.println("Return value: " + consumerRecord.value());
        };
    }

    @Bean
    public ReplyingKafkaTemplate<String, String, String> replyingTemplate(
            ProducerFactory<String, String> pf,
            ConcurrentMessageListenerContainer<String, String> repliesContainer) {

        return new ReplyingKafkaTemplate<>(pf, repliesContainer);
    }

    @Bean
    public ConcurrentMessageListenerContainer<String, String> repliesContainer(
            ConcurrentKafkaListenerContainerFactory<String, String> containerFactory) {

        ConcurrentMessageListenerContainer<String, String> repliesContainer =
                containerFactory.createContainer("replies");
        repliesContainer.getContainerProperties().setGroupId("repliesGroup");
        repliesContainer.setAutoStartup(false);
        return repliesContainer;
    }

    @Bean
    public NewTopic kRequests() {
        return TopicBuilder.name("kRequests")
            .partitions(10)
            .replicas(2)
            .build();
    }

    @Bean
    public NewTopic kReplies() {
        return TopicBuilder.name("kReplies")
            .partitions(10)
            .replicas(2)
            .build();
    }

}

Note that we can use Boot’s auto-configured container factory to create the reply container.

If a non-trivial deserializer is being used for replies, consider using an ErrorHandlingDeserializer that delegates to your configured deserializer. When so configured, the RequestReplyFuture will be completed exceptionally and you can catch the ExecutionException, with the DeserializationException in its cause property.

The template sets a header (named KafkaHeaders.CORRELATION_ID by default), which must be echoed back by the server side.

In this case, the following @KafkaListener application responds:

@SpringBootApplication
public class KReplyingApplication {

    public static void main(String[] args) {
        SpringApplication.run(KReplyingApplication.class, args);
    }

    @KafkaListener(id="server", topics = "kRequests")
    @SendTo // use default replyTo expression
    public String listen(String in) {
        System.out.println("Server received: " + in);
        return in.toUpperCase();
    }

    @Bean
    public NewTopic kRequests() {
        return TopicBuilder.name("kRequests")
            .partitions(10)
            .replicas(2)
            .build();
    }

    @Bean // not required if Jackson is on the classpath
    public MessagingMessageConverter simpleMapperConverter() {
        MessagingMessageConverter messagingMessageConverter = new MessagingMessageConverter();
        messagingMessageConverter.setHeaderMapper(new SimpleKafkaHeaderMapper());
        return messagingMessageConverter;
    }

}

The @KafkaListener infrastructure echoes the correlation ID and determines the reply topic.

See Forwarding Listener Results using @SendTo for more information about sending replies. The template uses the default header KafKaHeaders.REPLY_TOPIC to indicate the topic to which the reply goes.

Starting with version 2.2, the template tries to detect the reply topic or partition from the configured reply container. If the container is configured to listen to a single topic or a single TopicPartitionOffset, it is used to set the reply headers. If the container is configured otherwise, the user must set up the reply headers. In this case, an INFO log message is written during initialization. The following example uses KafkaHeaders.REPLY_TOPIC:

record.headers().add(new RecordHeader(KafkaHeaders.REPLY_TOPIC, "kReplies".getBytes()));

When you configure with a single reply TopicPartitionOffset, you can use the same reply topic for multiple templates, as long as each instance listens on a different partition. When configuring with a single reply topic, each instance must use a different group.id. In this case, all instances receive each reply, but only the instance that sent the request finds the correlation ID. This may be useful for auto-scaling, but with the overhead of additional network traffic and the small cost of discarding each unwanted reply. When you use this setting, we recommend that you set the template’s sharedReplyTopic to true, which reduces the logging level of unexpected replies to DEBUG instead of the default ERROR.

The following is an example of configuring the reply container to use the same shared reply topic:

@Bean
public ConcurrentMessageListenerContainer<String, String> replyContainer(
        ConcurrentKafkaListenerContainerFactory<String, String> containerFactory) {

    ConcurrentMessageListenerContainer<String, String> container = containerFactory.createContainer("topic2");
    container.getContainerProperties().setGroupId(UUID.randomUUID().toString()); // unique
    Properties props = new Properties();
    props.setProperty(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, "latest"); // so the new group doesn't get old replies
    container.getContainerProperties().setKafkaConsumerProperties(props);
    return container;
}
If you have multiple client instances and you do not configure them as discussed in the preceding paragraph, each instance needs a dedicated reply topic. An alternative is to set the KafkaHeaders.REPLY_PARTITION and use a dedicated partition for each instance. The Header contains a four-byte int (big-endian). The server must use this header to route the reply to the correct partition (@KafkaListener does this). In this case, though, the reply container must not use Kafka’s group management feature and must be configured to listen on a fixed partition (by using a TopicPartitionOffset in its ContainerProperties constructor).
The DefaultKafkaHeaderMapper requires Jackson to be on the classpath (for the @KafkaListener). If it is not available, the message converter has no header mapper, so you must configure a MessagingMessageConverter with a SimpleKafkaHeaderMapper, as shown earlier.

By default, 3 headers are used:

  • KafkaHeaders.CORRELATION_ID - used to correlate the reply to a request

  • KafkaHeaders.REPLY_TOPIC - used to tell the server where to reply

  • KafkaHeaders.REPLY_PARTITION - (optional) used to tell the server which partition to reply to

These header names are used by the @KafkaListener infrastructure to route the reply.

Starting with version 2.3, you can customize the header names - the template has 3 properties correlationHeaderName, replyTopicHeaderName, and replyPartitionHeaderName. This is useful if your server is not a Spring application (or does not use the @KafkaListener).

Reply Type Message<?>

When the @KafkaListener returns a Message<?>, with versions before 2.5, it was necessary to populate the reply topic and correlation id headers. In this example, we use the reply topic header from the request:

@KafkaListener(id = "requestor", topics = "request")
@SendTo
public Message<?> messageReturn(String in) {
    return MessageBuilder.withPayload(in.toUpperCase())
            .setHeader(KafkaHeaders.TOPIC, replyTo)
            .setHeader(KafkaHeaders.MESSAGE_KEY, 42)
            .setHeader(KafkaHeaders.CORRELATION_ID, correlation)
            .build();
}

This also shows how to set a key on the reply record.

Starting with version 2.5, the framework will detect if these headers are missing and populate them with the topic - either the topic determined from the @SendTo value or the incoming KafkaHeaders.REPLY_TOPIC header (if present). It will also echo the incoming KafkaHeaders.CORRELATION_ID and KafkaHeaders.REPLY_PARTITION, if present.

@KafkaListener(id = "requestor", topics = "request")
@SendTo  // default REPLY_TOPIC header
public Message<?> messageReturn(String in) {
    return MessageBuilder.withPayload(in.toUpperCase())
            .setHeader(KafkaHeaders.MESSAGE_KEY, 42)
            .build();
}
Aggregating Multiple Replies

The template in Using ReplyingKafkaTemplate is strictly for a single request/reply scenario. For cases where multiple receivers of a single message return a reply, you can use the AggregatingReplyingKafkaTemplate. This is an implementation of the client-side of the Scatter-Gather Enterprise Integration Pattern.

Like the ReplyingKafkaTemplate, the AggregatingReplyingKafkaTemplate constructor takes a producer factory and a listener container to receive the replies; it has a third parameter BiPredicate<List<ConsumerRecord<K, R>>, Boolean> releaseStrategy which is consulted each time a reply is received; when the predicate returns true, the collection of ConsumerRecord s is used to complete the Future returned by the sendAndReceive method.

There is an additional property returnPartialOnTimeout (default false). When this is set to true, instead of completing the future with a KafkaReplyTimeoutException, a partial result completes the future normally (as long as at least one reply record has been received).

Starting with version 2.3.5, the predicate is also called after a timeout (if returnPartialOnTimeout is true). The first argument is the current list of records; the second is true if this call is due to a timeout. The predicate can modify the list of records.

AggregatingReplyingKafkaTemplate<Integer, String, String> template =
        new AggregatingReplyingKafkaTemplate<>(producerFactory, container,
                        coll -> coll.size() == releaseSize);
...
RequestReplyFuture<Integer, String, Collection<ConsumerRecord<Integer, String>>> future =
        template.sendAndReceive(record);
future.getSendFuture().get(10, TimeUnit.SECONDS); // send ok
ConsumerRecord<Integer, Collection<ConsumerRecord<Integer, String>>> consumerRecord =
        future.get(30, TimeUnit.SECONDS);

Notice that the return type is a ConsumerRecord with a value that is a collection of ConsumerRecord s. The "outer" ConsumerRecord is not a "real" record, it is synthesized by the template, as a holder for the actual reply records received for the request. When a normal release occurs (release strategy returns true), the topic is set to aggregatedResults; if returnPartialOnTimeout is true, and timeout occurs (and at least one reply record has been received), the topic is set to partialResultsAfterTimeout. The template provides constant static variables for these "topic" names:

/**
 * Pseudo topic name for the "outer" {@link ConsumerRecords} that has the aggregated
 * results in its value after a normal release by the release strategy.
 */
public static final String AGGREGATED_RESULTS_TOPIC = "aggregatedResults";

/**
 * Pseudo topic name for the "outer" {@link ConsumerRecords} that has the aggregated
 * results in its value after a timeout.
 */
public static final String PARTIAL_RESULTS_AFTER_TIMEOUT_TOPIC = "partialResultsAfterTimeout";

The real ConsumerRecord s in the Collection contain the actual topic(s) from which the replies are received.

The listener container for the replies MUST be configured with AckMode.MANUAL or AckMode.MANUAL_IMMEDIATE; the consumer property enable.auto.commit must be false (the default since version 2.3). To avoid any possibility of losing messages, the template only commits offsets when there are zero requests outstanding, i.e. when the last outstanding request is released by the release strategy. After a rebalance, it is possible for duplicate reply deliveries; these will be ignored for any in-flight requests; you may see error log messages when duplicate replies are received for already released replies.
If you use an ErrorHandlingDeserializer with this aggregating template, the framework will not automatically detect DeserializationException s. Instead, the record (with a null value) will be returned intact, with the deserialization exception(s) in headers. It is recommended that applications call the utility method ReplyingKafkaTemplate.checkDeserialization() method to determine if a deserialization exception occurred. See its javadocs for more information.

4.1.4. Receiving Messages

You can receive messages by configuring a MessageListenerContainer and providing a message listener or by using the @KafkaListener annotation.

Message Listeners

When you use a message listener container, you must provide a listener to receive data. There are currently eight supported interfaces for message listeners. The following listing shows these interfaces:

public interface MessageListener<K, V> { (1)

    void onMessage(ConsumerRecord<K, V> data);

}

public interface AcknowledgingMessageListener<K, V> { (2)

    void onMessage(ConsumerRecord<K, V> data, Acknowledgment acknowledgment);

}

public interface ConsumerAwareMessageListener<K, V> extends MessageListener<K, V> { (3)

    void onMessage(ConsumerRecord<K, V> data, Consumer<?, ?> consumer);

}

public interface AcknowledgingConsumerAwareMessageListener<K, V> extends MessageListener<K, V> { (4)

    void onMessage(ConsumerRecord<K, V> data, Acknowledgment acknowledgment, Consumer<?, ?> consumer);

}

public interface BatchMessageListener<K, V> { (5)

    void onMessage(List<ConsumerRecord<K, V>> data);

}

public interface BatchAcknowledgingMessageListener<K, V> { (6)

    void onMessage(List<ConsumerRecord<K, V>> data, Acknowledgment acknowledgment);

}

public interface BatchConsumerAwareMessageListener<K, V> extends BatchMessageListener<K, V> { (7)

    void onMessage(List<ConsumerRecord<K, V>> data, Consumer<?, ?> consumer);

}

public interface BatchAcknowledgingConsumerAwareMessageListener<K, V> extends BatchMessageListener<K, V> { (8)

    void onMessage(List<ConsumerRecord<K, V>> data, Acknowledgment acknowledgment, Consumer<?, ?> consumer);

}
1 Use this interface for processing individual ConsumerRecord instances received from the Kafka consumer poll() operation when using auto-commit or one of the container-managed commit methods.
2 Use this interface for processing individual ConsumerRecord instances received from the Kafka consumer poll() operation when using one of the manual commit methods.
3 Use this interface for processing individual ConsumerRecord instances received from the Kafka consumer poll() operation when using auto-commit or one of the container-managed commit methods. Access to the Consumer object is provided.
4 Use this interface for processing individual ConsumerRecord instances received from the Kafka consumer poll() operation when using one of the manual commit methods. Access to the Consumer object is provided.
5 Use this interface for processing all ConsumerRecord instances received from the Kafka consumer poll() operation when using auto-commit or one of the container-managed commit methods. AckMode.RECORD is not supported when you use this interface, since the listener is given the complete batch.
6 Use this interface for processing all ConsumerRecord instances received from the Kafka consumer poll() operation when using one of the manual commit methods.
7 Use this interface for processing all ConsumerRecord instances received from the Kafka consumer poll() operation when using auto-commit or one of the container-managed commit methods. AckMode.RECORD is not supported when you use this interface, since the listener is given the complete batch. Access to the Consumer object is provided.
8 Use this interface for processing all ConsumerRecord instances received from the Kafka consumer poll() operation when using one of the manual commit methods. Access to the Consumer object is provided.
The Consumer object is not thread-safe. You must only invoke its methods on the thread that calls the listener.
Message Listener Containers

Two MessageListenerContainer implementations are provided:

  • KafkaMessageListenerContainer

  • ConcurrentMessageListenerContainer

The KafkaMessageListenerContainer receives all message from all topics or partitions on a single thread. The ConcurrentMessageListenerContainer delegates to one or more KafkaMessageListenerContainer instances to provide multi-threaded consumption.

Starting with version 2.2.7, you can add a RecordInterceptor to the listener container; it will be invoked before calling the listener allowing inspection or modification of the record. If the interceptor returns null, the listener is not called. The interceptor is not invoked when the listener is a batch listener.

Starting with version 2.3, the CompositeRecordInterceptor can be used to invoke multiple interceptors.

By default, when using transactions, the interceptor is invoked after the transaction has started. Starting with version 2.3.4, you can set the listener container’s interceptBeforeTx property to invoke the interceptor before the transaction has started instead.

No interceptor is provided for batch listeners because Kafka already provides a ConsumerInterceptor.

Starting with versions 2.3.8, 2.4.6, the ConcurrentMessageListenerContainer now supports Static Membership when the concurrency is greater than one. The group.instance.id is suffixed with -n with n starting at 1. This, together with an increased session.timeout.ms, can be used to reduce rebalance events, for example, when application instances are restarted.

Using KafkaMessageListenerContainer

The following constructor is available:

public KafkaMessageListenerContainer(ConsumerFactory<K, V> consumerFactory,
                    ContainerProperties containerProperties)

It receives a ConsumerFactory and information about topics and partitions, as well as other configuration, in a ContainerProperties object. ContainerProperties has the following constructors:

public ContainerProperties(TopicPartitionOffset... topicPartitions)

public ContainerProperties(String... topics)

public ContainerProperties(Pattern topicPattern)

The first constructor takes an array of TopicPartitionOffset arguments to explicitly instruct the container about which partitions to use (using the consumer assign() method) and with an optional initial offset. A positive value is an absolute offset by default. A negative value is relative to the current last offset within a partition by default. A constructor for TopicPartitionOffset that takes an additional boolean argument is provided. If this is true, the initial offsets (positive or negative) are relative to the current position for this consumer. The offsets are applied when the container is started. The second takes an array of topics, and Kafka allocates the partitions based on the group.id property — distributing partitions across the group. The third uses a regex Pattern to select the topics.

To assign a MessageListener to a container, you can use the ContainerProps.setMessageListener method when creating the Container. The following example shows how to do so:

ContainerProperties containerProps = new ContainerProperties("topic1", "topic2");
containerProps.setMessageListener(new MessageListener<Integer, String>() {
    ...
});
DefaultKafkaConsumerFactory<Integer, String> cf =
                        new DefaultKafkaConsumerFactory<>(consumerProps());
KafkaMessageListenerContainer<Integer, String> container =
                        new KafkaMessageListenerContainer<>(cf, containerProps);
return container;

Note that when creating a DefaultKafkaConsumerFactory, using the constructor that just takes in the properties as above means that key and value Deserializer classes are picked up from configuration. Alternatively, Deserializer instances may be passed to the DefaultKafkaConsumerFactory constructor for key and/or value, in which case all Consumers share the same instances. Another option is to provide Supplier<Deserializer> s (starting with version 2.3) that will be used to obtain separate Deserializer instances for each Consumer:

DefaultKafkaConsumerFactory<Integer, CustomValue> cf =
                        new DefaultKafkaConsumerFactory<>(consumerProps(), null, () -> new CustomValueDeserializer());
KafkaMessageListenerContainer<Integer, String> container =
                        new KafkaMessageListenerContainer<>(cf, containerProps);
return container;

Refer to the Javadoc for ContainerProperties for more information about the various properties that you can set.

Since version 2.1.1, a new property called logContainerConfig is available. When true and INFO logging is enabled each listener container writes a log message summarizing its configuration properties.

By default, logging of topic offset commits is performed at the DEBUG logging level. Starting with version 2.1.2, a property in ContainerProperties called commitLogLevel lets you specify the log level for these messages. For example, to change the log level to INFO, you can use containerProperties.setCommitLogLevel(LogIfLevelEnabled.Level.INFO);.

Starting with version 2.2, a new container property called missingTopicsFatal has been added (default: false since 2.3.4). This prevents the container from starting if any of the configured topics are not present on the broker. It does not apply if the container is configured to listen to a topic pattern (regex). Previously, the container threads looped within the consumer.poll() method waiting for the topic to appear while logging many messages. Aside from the logs, there was no indication that there was a problem.

As of version 2.3.5, a new container property called authorizationExceptionRetryInterval has been introduced. This causes the container to retry fetching messages after getting any AuthorizationException from KafkaConsumer. This can happen when, for example, the configured user is denied access to read certain topic. Defining authorizationExceptionRetryInterval should help the application to recover as soon as proper permissions are granted.

By default, no interval is configured - authorization errors are considered fatal, which causes the container to stop.
Using ConcurrentMessageListenerContainer

The single constructor is similar to the KafkaListenerContainer constructor. The following listing shows the constructor’s signature:

public ConcurrentMessageListenerContainer(ConsumerFactory<K, V> consumerFactory,
                            ContainerProperties containerProperties)

It also has a concurrency property. For example, container.setConcurrency(3) creates three KafkaMessageListenerContainer instances.

For the first constructor, Kafka distributes the partitions across the consumers using its group management capabilities.

When listening to multiple topics, the default partition distribution may not be what you expect. For example, if you have three topics with five partitions each and you want to use concurrency=15, you see only five active consumers, each assigned one partition from each topic, with the other 10 consumers being idle. This is because the default Kafka PartitionAssignor is the RangeAssignor (see its Javadoc). For this scenario, you may want to consider using the RoundRobinAssignor instead, which distributes the partitions across all of the consumers. Then, each consumer is assigned one topic or partition. To change the PartitionAssignor, you can set the partition.assignment.strategy consumer property (ConsumerConfigs.PARTITION_ASSIGNMENT_STRATEGY_CONFIG) in the properties provided to the DefaultKafkaConsumerFactory.

When using Spring Boot, you can assign set the strategy as follows:

spring.kafka.consumer.properties.partition.assignment.strategy=\
org.apache.kafka.clients.consumer.RoundRobinAssignor

When the container properties are configured with TopicPartitionOffset s, the ConcurrentMessageListenerContainer distributes the TopicPartitionOffset instances across the delegate KafkaMessageListenerContainer instances.

If, say, six TopicPartitionOffset instances are provided and the concurrency is 3; each container gets two partitions. For five TopicPartitionOffset instances, two containers get two partitions, and the third gets one. If the concurrency is greater than the number of TopicPartitions, the concurrency is adjusted down such that each container gets one partition.

The client.id property (if set) is appended with -n where n is the consumer instance that corresponds to the concurrency. This is required to provide unique names for MBeans when JMX is enabled.

Starting with version 1.3, the MessageListenerContainer provides access to the metrics of the underlying KafkaConsumer. In the case of ConcurrentMessageListenerContainer, the metrics() method returns the metrics for all the target KafkaMessageListenerContainer instances. The metrics are grouped into the Map<MetricName, ? extends Metric> by the client-id provided for the underlying KafkaConsumer.

Starting with version 2.3, the ContainerProperties provides an idleBetweenPolls option to let the main loop in the listener container to sleep between KafkaConsumer.poll() calls. An actual sleep interval is selected as the minimum from the provided option and difference between the max.poll.interval.ms consumer config and the current records batch processing time.

Committing Offsets

Several options are provided for committing offsets. If the enable.auto.commit consumer property is true, Kafka auto-commits the offsets according to its configuration. If it is false, the containers support several AckMode settings (described in the next list). The default AckMode is BATCH. Starting with version 2.3, the framework sets enable.auto.commit to false unless explicitly set in the configuration. Previously, the Kafka default (true) was used if the property was not set.

The consumer poll() method returns one or more ConsumerRecords. The MessageListener is called for each record. The following lists describes the action taken by the container for each AckMode (when transactions are not being used):

  • RECORD: Commit the offset when the listener returns after processing the record.

  • BATCH: Commit the offset when all the records returned by the poll() have been processed.

  • TIME: Commit the offset when all the records returned by the poll() have been processed, as long as the ackTime since the last commit has been exceeded.

  • COUNT: Commit the offset when all the records returned by the poll() have been processed, as long as ackCount records have been received since the last commit.

  • COUNT_TIME: Similar to TIME and COUNT, but the commit is performed if either condition is true.

  • MANUAL: The message listener is responsible to acknowledge() the Acknowledgment. After that, the same semantics as BATCH are applied.

  • MANUAL_IMMEDIATE: Commit the offset immediately when the Acknowledgment.acknowledge() method is called by the listener.

When using transactions, the offset(s) are sent to the transaction and the semantics are equivalent to RECORD or BATCH, depending on the listener type (record or batch).

MANUAL, and MANUAL_IMMEDIATE require the listener to be an AcknowledgingMessageListener or a BatchAcknowledgingMessageListener. See Message Listeners.

Depending on the syncCommits container property, the commitSync() or commitAsync() method on the consumer is used. syncCommits is true by default; also see setSyncCommitTimeout. See setCommitCallback to get the results of asynchronous commits; the default callback is the LoggingCommitCallback which logs errors (and successes at debug level).

Because the listener container has it’s own mechanism for committing offsets, it prefers the Kafka ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG to be false. Starting with version 2.3, it unconditionally sets it to false unless specifically set in the consumer factory or the container’s consumer property overrides.

The Acknowledgment has the following method:

public interface Acknowledgment {

    void acknowledge();

}

This method gives the listener control over when offsets are committed.

Starting with version 2.3, the Acknowledgment interface has two additional methods nack(long sleep) and nack(int index, long sleep). The first one is used with a record listener, the second with a batch listener. Calling the wrong method for your listener type will throw an IllegalStateException.

nack() can only be called on the consumer thread that invokes your listener.

With a record listener, when nack() is called, any pending offsets are committed, the remaing records from the last poll are discarded, and seeks are performed on their partitions so that the failed record and unprocessed records are redelivered on the next poll(). The consumer thread can be paused before redelivery, by setting the sleep argument. This is similar functionality to throwing an exception when the container is configured with a SeekToCurrentErrorHandler.

When using a batch listener, you can specify the index within the batch where the failure occurred. When nack() is called, offsets will be committed for records before the index and seeks are performed on the partitions for the failed and discarded records so that they will be redelivered on the next poll(). This is an improvement over the SeekToCurrentBatchErrorHandler, which can only seek the entire batch for redelivery.

When using partition assignment via group management, it is important to ensure the sleep argument (plus the time spent processing records from the previous poll) is less than the consumer max.poll.interval.ms property.
Listener Container Auto Startup

The listener containers implement SmartLifecycle, and autoStartup is true by default. The containers are started in a late phase (Integer.MAX-VALUE - 100). Other components that implement SmartLifecycle, to handle data from listeners, should be started in an earlier phase. The - 100 leaves room for later phases to enable components to be auto-started after the containers.

@KafkaListener Annotation

The @KafkaListener annotation is used to designate a bean method as a listener for a listener container. The bean is wrapped in a MessagingMessageListenerAdapter configured with various features, such as converters to convert the data, if necessary, to match the method parameters.

You can configure most attributes on the annotation with SpEL by using #{…​} or property placeholders (${…​}). See the Javadoc for more information.

Record Listeners

The @KafkaListener annotation provides a mechanism for simple POJO listeners. The following example shows how to use it:

public class Listener {

    @KafkaListener(id = "foo", topics = "myTopic", clientIdPrefix = "myClientId")
    public void listen(String data) {
        ...
    }

}

This mechanism requires an @EnableKafka annotation on one of your @Configuration classes and a listener container factory, which is used to configure the underlying ConcurrentMessageListenerContainer. By default, a bean with name kafkaListenerContainerFactory is expected. The following example shows how to use ConcurrentMessageListenerContainer:

@Configuration
@EnableKafka
public class KafkaConfig {

    @Bean
    KafkaListenerContainerFactory<ConcurrentMessageListenerContainer<Integer, String>>
                        kafkaListenerContainerFactory() {
        ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
                                new ConcurrentKafkaListenerContainerFactory<>();
        factory.setConsumerFactory(consumerFactory());
        factory.setConcurrency(3);
        factory.getContainerProperties().setPollTimeout(3000);
        return factory;
    }

    @Bean
    public ConsumerFactory<Integer, String> consumerFactory() {
        return new DefaultKafkaConsumerFactory<>(consumerConfigs());
    }

    @Bean
    public Map<String, Object> consumerConfigs() {
        Map<String, Object> props = new HashMap<>();
        props.put(ProducerConfig.BOOTSTRAP_SERVERS_CONFIG, embeddedKafka.getBrokersAsString());
        ...
        return props;
    }
}

Notice that, to set container properties, you must use the getContainerProperties() method on the factory. It is used as a template for the actual properties injected into the container.

Starting with version 2.1.1, you can now set the client.id property for consumers created by the annotation. The clientIdPrefix is suffixed with -n, where n is an integer representing the container number when using concurrency.

Starting with version 2.2, you can now override the container factory’s concurrency and autoStartup properties by using properties on the annotation itself. The properties can be simple values, property placeholders, or SpEL expressions. The following example shows how to do so:

@KafkaListener(id = "myListener", topics = "myTopic",
        autoStartup = "${listen.auto.start:true}", concurrency = "${listen.concurrency:3}")
public void listen(String data) {
    ...
}
Explicit Partition Assignment

You can also configure POJO listeners with explicit topics and partitions (and, optionally, their initial offsets). The following example shows how to do so:

@KafkaListener(id = "thing2", topicPartitions =
        { @TopicPartition(topic = "topic1", partitions = { "0", "1" }),
          @TopicPartition(topic = "topic2", partitions = "0",
             partitionOffsets = @PartitionOffset(partition = "1", initialOffset = "100"))
        })
public void listen(ConsumerRecord<?, ?> record) {
    ...
}

You can specify each partition in the partitions or partitionOffsets attribute but not both.

As with most annotation properties, you can use SpEL expressions; for an example of how to generate a large list of partitions, see Manually Assigning All Partitions.

Starting with version 2.5.5, you can apply an initial offset to all assigned partitions:

@KafkaListener(id = "thing3", topicPartitions =
        { @TopicPartition(topic = "topic1", partitions = { "0", "1" },
             partitionOffsets = @PartitionOffset(partition = "*", initialOffset = "0"))
        })
public void listen(ConsumerRecord<?, ?> record) {
    ...
}

The * wildcard represents all partitions in the partitions attribute. There must only be one @PartitionOffset with the wildcard in each @TopicPartition.

In addition, when the listener implements ConsumerSeekAware, onPartitionsAssigned is now called, even when using manual assignment. This allows, for example, any arbitrary seek operations at that time.

Manual Acknowledgment

When using manual AckMode, you can also provide the listener with the Acknowledgment. The following example also shows how to use a different container factory.

@KafkaListener(id = "cat", topics = "myTopic",
          containerFactory = "kafkaManualAckListenerContainerFactory")
public void listen(String data, Acknowledgment ack) {
    ...
    ack.acknowledge();
}
Consumer Record Metadata

Finally, metadata about the record is available from message headers. You can use the following header names to retrieve the headers of the message:

  • KafkaHeaders.OFFSET

  • KafkaHeaders.RECEIVED_MESSAGE_KEY

  • KafkaHeaders.RECEIVED_TOPIC

  • KafkaHeaders.RECEIVED_PARTITION_ID

  • KafkaHeaders.RECEIVED_TIMESTAMP

  • KafkaHeaders.TIMESTAMP_TYPE

Starting with version 2.5 the RECEIVED_MESSAGE_KEY is not present if the incoming record has a null key; previously the header was populated with a null value. This change is to make the framework consistent with spring-messaging conventions where null valued headers are not present.

The following example shows how to use the headers:

@KafkaListener(id = "qux", topicPattern = "myTopic1")
public void listen(@Payload String foo,
        @Header(name = KafkaHeaders.RECEIVED_MESSAGE_KEY, required = false) Integer key,
        @Header(KafkaHeaders.RECEIVED_PARTITION_ID) int partition,
        @Header(KafkaHeaders.RECEIVED_TOPIC) String topic,
        @Header(KafkaHeaders.RECEIVED_TIMESTAMP) long ts
        ) {
    ...
}

Starting with version 2.5, instead of using discrete headers, you can receive record metadata in a ConsumerRecordMetadata parameter.

@KafkaListener(...)
public void listen(String str, ConsumerRecordMetadata meta) {
    ...
}

This contains all the data from the ConsumerRecord except the key and value.

Batch listeners

Starting with version 1.1, you can configure @KafkaListener methods to receive the entire batch of consumer records received from the consumer poll. To configure the listener container factory to create batch listeners, you can set the batchListener property. The following example shows how to do so:

@Bean
public KafkaListenerContainerFactory<?, ?> batchFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
            new ConcurrentKafkaListenerContainerFactory<>();
    factory.setConsumerFactory(consumerFactory());
    factory.setBatchListener(true);  // <<<<<<<<<<<<<<<<<<<<<<<<<
    return factory;
}

The following example shows how to receive a list of payloads:

@KafkaListener(id = "list", topics = "myTopic", containerFactory = "batchFactory")
public void listen(List<String> list) {
    ...
}

The topic, partition, offset, and so on are available in headers that parallel the payloads. The following example shows how to use the headers:

@KafkaListener(id = "list", topics = "myTopic", containerFactory = "batchFactory")
public void listen(List<String> list,
        @Header(KafkaHeaders.RECEIVED_MESSAGE_KEY) List<Integer> keys,
        @Header(KafkaHeaders.RECEIVED_PARTITION_ID) List<Integer> partitions,
        @Header(KafkaHeaders.RECEIVED_TOPIC) List<String> topics,
        @Header(KafkaHeaders.OFFSET) List<Long> offsets) {
    ...
}

Alternatively, you can receive a List of Message<?> objects with each offset and other details in each message, but it must be the only parameter (aside from optional Acknowledgment, when using manual commits, and/or Consumer<?, ?> parameters) defined on the method. The following example shows how to do so:

@KafkaListener(id = "listMsg", topics = "myTopic", containerFactory = "batchFactory")
public void listen14(List<Message<?>> list) {
    ...
}

@KafkaListener(id = "listMsgAck", topics = "myTopic", containerFactory = "batchFactory")
public void listen15(List<Message<?>> list, Acknowledgment ack) {
    ...
}

@KafkaListener(id = "listMsgAckConsumer", topics = "myTopic", containerFactory = "batchFactory")
public void listen16(List<Message<?>> list, Acknowledgment ack, Consumer<?, ?> consumer) {
    ...
}

No conversion is performed on the payloads in this case.

If the BatchMessagingMessageConverter is configured with a RecordMessageConverter, you can also add a generic type to the Message parameter and the payloads are converted. See Payload Conversion with Batch Listeners for more information.

You can also receive a list of ConsumerRecord<?, ?> objects, but it must be the only parameter (aside from optional Acknowledgment, when using manual commits and Consumer<?, ?> parameters) defined on the method. The following example shows how to do so:

@KafkaListener(id = "listCRs", topics = "myTopic", containerFactory = "batchFactory")
public void listen(List<ConsumerRecord<Integer, String>> list) {
    ...
}

@KafkaListener(id = "listCRsAck", topics = "myTopic", containerFactory = "batchFactory")
public void listen(List<ConsumerRecord<Integer, String>> list, Acknowledgment ack) {
    ...
}

Starting with version 2.2, the listener can receive the complete ConsumerRecords<?, ?> object returned by the poll() method, letting the listener access additional methods, such as partitions() (which returns the TopicPartition instances in the list) and records(TopicPartition) (which gets selective records). Again, this must be the only parameter (aside from optional Acknowledgment, when using manual commits or Consumer<?, ?> parameters) on the method. The following example shows how to do so:

@KafkaListener(id = "pollResults", topics = "myTopic", containerFactory = "batchFactory")
public void pollResults(ConsumerRecords<?, ?> records) {
    ...
}
If the container factory has a RecordFilterStrategy configured, it is ignored for ConsumerRecords<?, ?> listeners, with a WARN log message emitted. Records can only be filtered with a batch listener if the <List<?>> form of listener is used.
Annotation Properties

Starting with version 2.0, the id property (if present) is used as the Kafka consumer group.id property, overriding the configured property in the consumer factory, if present. You can also set groupId explicitly or set idIsGroup to false to restore the previous behavior of using the consumer factory group.id.

You can use property placeholders or SpEL expressions within most annotation properties, as the following example shows:

@KafkaListener(topics = "${some.property}")

@KafkaListener(topics = "#{someBean.someProperty}",
    groupId = "#{someBean.someProperty}.group")

Starting with version 2.1.2, the SpEL expressions support a special token: __listener. It is a pseudo bean name that represents the current bean instance within which this annotation exists.

Consider the following example:

@Bean
public Listener listener1() {
    return new Listener("topic1");
}

@Bean
public Listener listener2() {
    return new Listener("topic2");
}

Given the beans in the previous example, we can then use the following:

public class Listener {

    private final String topic;

    public Listener(String topic) {
        this.topic = topic;
    }

    @KafkaListener(topics = "#{__listener.topic}",
        groupId = "#{__listener.topic}.group")
    public void listen(...) {
        ...
    }

    public String getTopic() {
        return this.topic;
    }

}

If, in the unlikely event that you have an actual bean called __listener, you can change the expression token byusing the beanRef attribute. The following example shows how to do so:

@KafkaListener(beanRef = "__x", topics = "#{__x.topic}",
    groupId = "#{__x.topic}.group")

Starting with version 2.2.4, you can specify Kafka consumer properties directly on the annotation, these will override any properties with the same name configured in the consumer factory. You cannot specify the group.id and client.id properties this way; they will be ignored; use the groupId and clientIdPrefix annotation properties for those.

The properties are specified as individual strings with the normal Java Properties file format: foo:bar, foo=bar, or foo bar.

@KafkaListener(topics = "myTopic", groupId = "group", properties = {
    "max.poll.interval.ms:60000",
    ConsumerConfig.MAX_POLL_RECORDS_CONFIG + "=100"
})

The following is an example of the corresponding listeners for the example in Using RoutingKafkaTemplate.

@KafkaListener(id = "one", topics = "one")
public void listen1(String in) {
    System.out.println("1: " + in);
}

@KafkaListener(id = "two", topics = "two",
        properties = "value.deserializer:org.apache.kafka.common.serialization.ByteArrayDeserializer")
public void listen2(byte[] in) {
    System.out.println("2: " + new String(in));
}
Obtaining the Consumer group.id

When running the same listener code in multiple containers, it may be useful to be able to determine which container (identified by its group.id consumer property) that a record came from.

You can call KafkaUtils.getConsumerGroupId() on the listener thread to do this. Alternatively, you can access the group id in a method parameter.

@KafkaListener(id = "bar", topicPattern = "${topicTwo:annotated2}", exposeGroupId = "${always:true}")
public void listener(@Payload String foo,
        @Header(KafkaHeaders.GROUP_ID) String groupId) {
...
}
This is available in record listeners and batch listeners that receive a List<?> of records. It is not available in a batch listener that receives a ConsumerRecords<?, ?> argument. Use the KafkaUtils mechanism in that case.
Container Thread Naming

Listener containers currently use two task executors, one to invoke the consumer and another that is used to invoke the listener when the kafka consumer property enable.auto.commit is false. You can provide custom executors by setting the consumerExecutor and listenerExecutor properties of the container’s ContainerProperties. When using pooled executors, be sure that enough threads are available to handle the concurrency across all the containers in which they are used. When using the ConcurrentMessageListenerContainer, a thread from each is used for each consumer (concurrency).

If you do not provide a consumer executor, a SimpleAsyncTaskExecutor is used. This executor creates threads with names similar to <beanName>-C-1 (consumer thread). For the ConcurrentMessageListenerContainer, the <beanName> part of the thread name becomes <beanName>-m, where m represents the consumer instance. n increments each time the container is started. So, with a bean name of container, threads in this container will be named container-0-C-1, container-1-C-1 etc., after the container is started the first time; container-0-C-2, container-1-C-2 etc., after a stop and subsequent start.

@KafkaListener as a Meta Annotation

Starting with version 2.2, you can now use @KafkaListener as a meta annotation. The following example shows how to do so:

@Target(ElementType.METHOD)
@Retention(RetentionPolicy.RUNTIME)
@KafkaListener
public @interface MyThreeConsumersListener {

    @AliasFor(annotation = KafkaListener.class, attribute = "id")
    String id();

    @AliasFor(annotation = KafkaListener.class, attribute = "topics")
    String[] topics();

    @AliasFor(annotation = KafkaListener.class, attribute = "concurrency")
    String concurrency() default "3";

}

You must alias at least one of topics, topicPattern, or topicPartitions (and, usually, id or groupId unless you have specified a group.id in the consumer factory configuration). The following example shows how to do so:

@MyThreeConsumersListener(id = "my.group", topics = "my.topic")
public void listen1(String in) {
    ...
}
@KafkaListener on a Class

When you use @KafkaListener at the class-level, you must specify @KafkaHandler at the method level. When messages are delivered, the converted message payload type is used to determine which method to call. The following example shows how to do so:

@KafkaListener(id = "multi", topics = "myTopic")
static class MultiListenerBean {

    @KafkaHandler
    public void listen(String foo) {
        ...
    }

    @KafkaHandler
    public void listen(Integer bar) {
        ...
    }

    @KafkaHandler(isDefault = true)
    public void listenDefault(Object object) {
        ...
    }

}

Starting with version 2.1.3, you can designate a @KafkaHandler method as the default method that is invoked if there is no match on other methods. At most, one method can be so designated. When using @KafkaHandler methods, the payload must have already been converted to the domain object (so the match can be performed). Use a custom deserializer, the JsonDeserializer, or the JsonMessageConverter with its TypePrecedence set to TYPE_ID. See Serialization, Deserialization, and Message Conversion for more information.

Due to some limitations in the way Spring resolves method arguments, a default @KafkaHandler cannot receive discrete headers; it must use the ConsumerRecordMetadata as discussed in Consumer Record Metadata.

For example:

@KafkaHandler(isDefault = true)
public void listenDefault(Object object, @Header(KafkaHeaders.RECEIVED_TOPIC) String topic) {
    ...
}

This won’t work if the object is a String; the topic parameter will also get a reference to object.

If you need metadata about the record in a default method, use this

@KafkaListener Lifecycle Management

The listener containers created for @KafkaListener annotations are not beans in the application context. Instead, they are registered with an infrastructure bean of type KafkaListenerEndpointRegistry. This bean is automatically declared by the framework and manages the containers' lifecycles; it will auto-start any containers that have autoStartup set to true. All containers created by all container factories must be in the same phase. See Listener Container Auto Startup for more information. You can manage the lifecycle programmatically by using the registry. Starting or stopping the registry will start or stop all the registered containers. Alternatively, you can get a reference to an individual container by using its id attribute. You can set autoStartup on the annotation, which overrides the default setting configured into the container factory. You can get a reference to the bean from the application context, such as auto-wiring, to manage its registered containers. The following examples show how to do so:

@KafkaListener(id = "myContainer", topics = "myTopic", autoStartup = "false")
public void listen(...) { ... }
@Autowired
private KafkaListenerEndpointRegistry registry;

...

    this.registry.getListenerContainer("myContainer").start();

...

The registry only maintains the life cycle of containers it manages; containers declared as beans are not managed by the registry and can be obtained from the application context. A collection of managed containers can be obtained by calling the registry’s getListenerContainers() method. Version 2.2.5 added a convenience method getAllListenerContainers(), which returns a collection of all containers, including those managed by the registry and those declared as beans. The collection returned will include any prototype beans that have been initialized, but it will not initialize any lazy bean declarations.

@KafkaListener @Payload Validation

Starting with version 2.2, it is now easier to add a Validator to validate @KafkaListener @Payload arguments. Previously, you had to configure a custom DefaultMessageHandlerMethodFactory and add it to the registrar. Now, you can add the validator to the registrar itself. The following code shows how to do so:

@Configuration
@EnableKafka
public class Config implements KafkaListenerConfigurer {

    ...

    @Override
    public void configureKafkaListeners(KafkaListenerEndpointRegistrar registrar) {
      registrar.setValidator(new MyValidator());
    }
}
When you use Spring Boot with the validation starter, a LocalValidatorFactoryBean is auto-configured, as the following example shows:
@Configuration
@EnableKafka
public class Config implements KafkaListenerConfigurer {

    @Autowired
    private LocalValidatorFactoryBean validator;
    ...

    @Override
    public void configureKafkaListeners(KafkaListenerEndpointRegistrar registrar) {
      registrar.setValidator(this.validator);
    }
}

The following examples show how to validate:

public static class ValidatedClass {

  @Max(10)
  private int bar;

  public int getBar() {
    return this.bar;
  }

  public void setBar(int bar) {
    this.bar = bar;
  }

}
@KafkaListener(id="validated", topics = "annotated35", errorHandler = "validationErrorHandler",
      containerFactory = "kafkaJsonListenerContainerFactory")
public void validatedListener(@Payload @Valid ValidatedClass val) {
    ...
}

@Bean
public KafkaListenerErrorHandler validationErrorHandler() {
    return (m, e) -> {
        ...
    };
}
Rebalancing Listeners

ContainerProperties has a property called consumerRebalanceListener, which takes an implementation of the Kafka client’s ConsumerRebalanceListener interface. If this property is not provided, the container configures a logging listener that logs rebalance events at the INFO level. The framework also adds a sub-interface ConsumerAwareRebalanceListener. The following listing shows the ConsumerAwareRebalanceListener interface definition:

public interface ConsumerAwareRebalanceListener extends ConsumerRebalanceListener {

    void onPartitionsRevokedBeforeCommit(Consumer<?, ?> consumer, Collection<TopicPartition> partitions);

    void onPartitionsRevokedAfterCommit(Consumer<?, ?> consumer, Collection<TopicPartition> partitions);

    void onPartitionsAssigned(Consumer<?, ?> consumer, Collection<TopicPartition> partitions);

    void onPartitionsLost(Consumer<?, ?> consumer, Collection<TopicPartition> partitions);

}

Notice that there are two callbacks when partitions are revoked. The first is called immediately. The second is called after any pending offsets are committed. This is useful if you wish to maintain offsets in some external repository, as the following example shows:

containerProperties.setConsumerRebalanceListener(new ConsumerAwareRebalanceListener() {

    @Override
    public void onPartitionsRevokedBeforeCommit(Consumer<?, ?> consumer, Collection<TopicPartition> partitions) {
        // acknowledge any pending Acknowledgments (if using manual acks)
    }

    @Override
    public void onPartitionsRevokedAfterCommit(Consumer<?, ?> consumer, Collection<TopicPartition> partitions) {
        // ...
            store(consumer.position(partition));
        // ...
    }

    @Override
    public void onPartitionsAssigned(Collection<TopicPartition> partitions) {
        // ...
            consumer.seek(partition, offsetTracker.getOffset() + 1);
        // ...
    }
});
Starting with version 2.4, a new method onPartitionsLost() has been added (similar to a method with the same name in ConsumerRebalanceLister). The default implementation on ConsumerRebalanceLister simply calls onPartionsRevoked. The default implementation on ConsumerAwareRebalanceListener does nothing. When supplying the listener container with a custom listener (of either type), it is important that your implementation not call onPartitionsRevoked from onPartitionsLost. If you implement ConsumerRebalanceListener you should override the default method. This is because the listener container will call its own onPartitionsRevoked from its implementation of onPartitionsLost after calling the method on your implementation. If you implementation delegates to the default behavior, onPartitionsRevoked will be called twice each time the Consumer calls that method on the container’s listener.
Forwarding Listener Results using @SendTo

Starting with version 2.0, if you also annotate a @KafkaListener with a @SendTo annotation and the method invocation returns a result, the result is forwarded to the topic specified by the @SendTo.

The @SendTo value can have several forms:

  • @SendTo("someTopic") routes to the literal topic

  • @SendTo("#{someExpression}") routes to the topic determined by evaluating the expression once during application context initialization.

  • @SendTo("!{someExpression}") routes to the topic determined by evaluating the expression at runtime. The #root object for the evaluation has three properties:

    • request: The inbound ConsumerRecord (or ConsumerRecords object for a batch listener))

    • source: The org.springframework.messaging.Message<?> converted from the request.

    • result: The method return result.

  • @SendTo (no properties): This is treated as !{source.headers['kafka_replyTopic']} (since version 2.1.3).

Starting with versions 2.1.11 and 2.2.1, property placeholders are resolved within @SendTo values.

The result of the expression evaluation must be a String that represents the topic name. The following examples show the various ways to use @SendTo:

@KafkaListener(topics = "annotated21")
@SendTo("!{request.value()}") // runtime SpEL
public String replyingListener(String in) {
    ...
}

@KafkaListener(topics = "${some.property:annotated22}")
@SendTo("#{myBean.replyTopic}") // config time SpEL
public Collection<String> replyingBatchListener(List<String> in) {
    ...
}

@KafkaListener(topics = "annotated23", errorHandler = "replyErrorHandler")
@SendTo("annotated23reply") // static reply topic definition
public String replyingListenerWithErrorHandler(String in) {
    ...
}
...
@KafkaListener(topics = "annotated25")
@SendTo("annotated25reply1")
public class MultiListenerSendTo {

    @KafkaHandler
    public String foo(String in) {
        ...
    }

    @KafkaHandler
    @SendTo("!{'annotated25reply2'}")
    public String bar(@Payload(required = false) KafkaNull nul,
            @Header(KafkaHeaders.RECEIVED_MESSAGE_KEY) int key) {
        ...
    }

}
In order to support @SendTo, the listener container factory must be provided with a KafkaTemplate (in its replyTemplate property), which is used to send the reply. This should be a KafkaTemplate and not a ReplyingKafkaTemplate which is used on the client-side for request/reply processing. When using Spring Boot, boot will auto-configure the template into the factory; when configuring your own factory, it must be set as shown in the examples below.

Starting with version 2.2, you can add a ReplyHeadersConfigurer to the listener container factory. This is consulted to determine which headers you want to set in the reply message. The following example shows how to add a ReplyHeadersConfigurer:

@Bean
public ConcurrentKafkaListenerContainerFactory<Integer, String> kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
        new ConcurrentKafkaListenerContainerFactory<>();
    factory.setConsumerFactory(cf());
    factory.setReplyTemplate(template());
    factory.setReplyHeadersConfigurer((k, v) -> k.equals("cat"));
    return factory;
}

You can also add more headers if you wish. The following example shows how to do so:

@Bean
public ConcurrentKafkaListenerContainerFactory<Integer, String> kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
        new ConcurrentKafkaListenerContainerFactory<>();
    factory.setConsumerFactory(cf());
    factory.setReplyTemplate(template());
    factory.setReplyHeadersConfigurer(new ReplyHeadersConfigurer() {

      @Override
      public boolean shouldCopy(String headerName, Object headerValue) {
        return false;
      }

      @Override
      public Map<String, Object> additionalHeaders() {
        return Collections.singletonMap("qux", "fiz");
      }

    });
    return factory;
}

When you use @SendTo, you must configure the ConcurrentKafkaListenerContainerFactory with a KafkaTemplate in its replyTemplate property to perform the send.

Unless you use request/reply semantics only the simple send(topic, value) method is used, so you may wish to create a subclass to generate the partition or key. The following example shows how to do so:
@Bean
public KafkaTemplate<String, String> myReplyingTemplate() {
    return new KafkaTemplate<Integer, String>(producerFactory()) {

        @Override
        public ListenableFuture<SendResult<String, String>> send(String topic, String data) {
            return super.send(topic, partitionForData(data), keyForData(data), data);
        }

        ...

    };
}

If the listener method returns Message<?> or Collection<Message<?>>, the listener method is responsible for setting up the message headers for the reply. For example, when handling a request from a ReplyingKafkaTemplate, you might do the following:

@KafkaListener(id = "messageReturned", topics = "someTopic")
public Message<?> listen(String in, @Header(KafkaHeaders.REPLY_TOPIC) byte[] replyTo,
        @Header(KafkaHeaders.CORRELATION_ID) byte[] correlation) {
    return MessageBuilder.withPayload(in.toUpperCase())
            .setHeader(KafkaHeaders.TOPIC, replyTo)
            .setHeader(KafkaHeaders.MESSAGE_KEY, 42)
            .setHeader(KafkaHeaders.CORRELATION_ID, correlation)
            .setHeader("someOtherHeader", "someValue")
            .build();
}

When using request/reply semantics, the target partition can be requested by the sender.

You can annotate a @KafkaListener method with @SendTo even if no result is returned. This is to allow the configuration of an errorHandler that can forward information about a failed message delivery to some topic. The following example shows how to do so:

@KafkaListener(id = "voidListenerWithReplyingErrorHandler", topics = "someTopic",
        errorHandler = "voidSendToErrorHandler")
@SendTo("failures")
public void voidListenerWithReplyingErrorHandler(String in) {
    throw new RuntimeException("fail");
}

@Bean
public KafkaListenerErrorHandler voidSendToErrorHandler() {
    return (m, e) -> {
        return ... // some information about the failure and input data
    };
}

See Handling Exceptions for more information.

If a listener method returns an Iterable, by default a record for each element as the value is sent. Starting with version 2.3.5, set the splitIterables property on @KafkaListener to false and the entire result will be sent as the value of a single ProducerRecord. This requires a suitable serializer in the reply template’s producer configuration. However, if the reply is Iterable<Message<?>> the property is ignored and each message is sent separately.
Filtering Messages

In certain scenarios, such as rebalancing, a message that has already been processed may be redelivered. The framework cannot know whether such a message has been processed or not. That is an application-level function. This is known as the Idempotent Receiver pattern and Spring Integration provides an implementation of it.

The Spring for Apache Kafka project also provides some assistance by means of the FilteringMessageListenerAdapter class, which can wrap your MessageListener. This class takes an implementation of RecordFilterStrategy in which you implement the filter method to signal that a message is a duplicate and should be discarded. This has an additional property called ackDiscarded, which indicates whether the adapter should acknowledge the discarded record. It is false by default.

When you use @KafkaListener, set the RecordFilterStrategy (and optionally ackDiscarded) on the container factory so that the listener is wrapped in the appropriate filtering adapter.

In addition, a FilteringBatchMessageListenerAdapter is provided, for when you use a batch message listener.

The FilteringBatchMessageListenerAdapter is ignored if your @KafkaListener receives a ConsumerRecords<?, ?> instead of List<ConsumerRecord<?, ?>>, because ConsumerRecords is immutable.
Retrying Deliveries

If your listener throws an exception, the default behavior is to invoke the Container Error Handlers, if configured, or logged otherwise.

NOTE: To retry deliveries, a convenient listener adapter RetryingMessageListenerAdapter is provided.

You can configure it with a RetryTemplate and RecoveryCallback<Void> - see the spring-retry project for information about these components. If a recovery callback is not provided, the exception is thrown to the container after retries are exhausted. In that case, the ErrorHandler is invoked, if configured, or logged otherwise.

When you use @KafkaListener, you can set the RetryTemplate (and optionally recoveryCallback) on the container factory. When you do so, the listener is wrapped in the appropriate retrying adapter.

The contents of the RetryContext passed into the RecoveryCallback depend on the type of listener. The context always has a record attribute, which is the record for which the failure occurred. If your listener is acknowledging or consumer aware, additional acknowledgment or consumer attributes are available. For convenience, the RetryingMessageListenerAdapter provides static constants for these keys. See its Javadoc for more information.

A retry adapter is not provided for any of the batch message listeners, because the framework has no knowledge of where in a batch the failure occurred. If you need retry capabilities when you use a batch listener, we recommend that you use a RetryTemplate within the listener itself.

Stateful Retry
Now that the SeekToCurrentErrorHandler can be configured with a BackOff and has the ability to retry only certain exceptions (since version 2.3), the use of stateful retry, via the listener adapter retry configuration, is no longer necessary. You can provide the same functionality with appropriate configuration of the error handler and remove all retry configuration from the listener adatper. See Seek To Current Container Error Handlers for more information.

You should understand that the retry discussed in the preceding section suspends the consumer thread (if a BackOffPolicy is used). There are no calls to Consumer.poll() during the retries. Kafka has two properties to determine consumer health. The session.timeout.ms is used to determine if the consumer is active. Since kafka-clients version 0.10.1.0, heartbeats are sent on a background thread, so a slow consumer no longer affects that. max.poll.interval.ms (default: five minutes) is used to determine if a consumer appears to be hung (taking too long to process records from the last poll). If the time between poll() calls exceeds this, the broker revokes the assigned partitions and performs a rebalance. For lengthy retry sequences, with back off, this can easily happen.

Since version 2.1.3, you can avoid this problem by using stateful retry in conjunction with a SeekToCurrentErrorHandler. In this case, each delivery attempt throws the exception back to the container, the error handler re-seeks the unprocessed offsets, and the same message is redelivered by the next poll(). This avoids the problem of exceeding the max.poll.interval.ms property (as long as an individual delay between attempts does not exceed it). So, when you use an ExponentialBackOffPolicy, you must ensure that the maxInterval is less than the max.poll.interval.ms property. To enable stateful retry, you can use the RetryingMessageListenerAdapter constructor that takes a stateful boolean argument (set it to true). When you configure the listener container factory (for @KafkaListener), set the factory’s statefulRetry property to true.

Version 2.2 added recovery to the SeekToCurrentErrorHandler, such as sending a failed record to a dead-letter topic. When using stateful retry, you must perform the recovery in the retry RecoveryCallback and NOT in the error handler. Otherwise, if the recovery is done in the error handler, the retry template’s state will never be cleared. Also, you must ensure that the maxFailures in the SeekToCurrentErrorHandler must be at least as many as configured in the retry policy, again to ensure that the retries are exhausted and the state cleared. Here is an example for retry configuration when used with a SeekToCurrentErrorHandler where factory is the ConcurrentKafkaListenerContainerFactory.
@Autowired
DeadLetterPublishingRecoverer recoverer;

...
    factory.setRetryTemplate(new RetryTemplate()); // 3 retries by default
    factory.setStatefulRetry(true);
    factory.setRecoveryCallback(context -> {
        recoverer.accept((ConsumerRecord<?, ?>) context.getAttribute("record"),
                (Exception) context.getLastThrowable());
        return null;
    });
...

@Bean
public SeekToCurrentErrorHandler eh() {
    return new SeekToCurrentErrorHandler(new FixedBackOff(0L, 3L)); // at least 3
}

However, see the note at the beginning of this section; you can avoid using the RetryTemplate altogether.

If the recoverer fails (throws an exception), the failed record will be included in the seeks. Starting with version 2.5.5, if the recoverer fails, the BackOff will be reset by default and redeliveries will again go through the back offs before recovery is attempted again. With earlier versions, the BackOff was not reset and recovery was re-attempted on the next failure. To revert to the previous behavior, set the error handler’s resetStateOnRecoveryFailure to false.

Starting with version 2.6, you can now provide the error handler with a BiFunction<ConsumerRecord<?, ?>, Exception, BackOff> to determine the BackOff to use, based on the failed record and/or the exception:

handler.setBackOffFunction((record, ex) -> { ... });

If the function returns null, the handler’s default BackOff will be used.

Starting with version 2.6.3, set resetStateOnExceptionChange to true and the retry sequence will be restarted (including the selection of a new BackOff, if so configured) if the exception type changes between failures. By default, the exception type is not considered.

4.1.5. Listener Container Properties

Table 1. ContainerProperties Properties
Property Default Description
ackCount

1

The number of records before committing pending offsets when the ackMode is COUNT or COUNT_TIME.

adviceChain

null

A chain of Advice objects (e.g. MethodInterceptor around advice) wrapping the message listener, invoked in order.

ackMode

BATCH

Controls how often offsets are committed - see Committing Offsets.

ackOnError

false

[DEPRECATED in favor of ErrorHandler.isAckAfterHandle()]

ackTime

5000

The time in milliseconds after which pending offsets are committed when the ackMode is TIME or COUNT_TIME.

assignment
CommitOption

LATEST_ONLY _NO_TX

Whether or not to commit the initial position on assignment; by default, the initial offset will only be committed if the ConsumerConfig.AUTO_OFFSET_RESET_CONFIG is latest and it won’t run in a transaction even if there is a transaction manager present. See the javadocs for ContainerProperties.AssignmentCommitOption for more information about the available options.

authorizationException
RetryInterval

null

When not null, a Duration to sleep between polls when an AuthorizationException is thrown by the Kafka client. When null, such exceptions are considered fatal and the container will stop.

clientId

(empty string)

A prefix for the client.id consumer property. Overrides the consumer factory client.id property; in a concurrent container, -n is added as a suffix for each consumer instance.

commitCallback

null

When present and syncCommits is false a callback invoked after the commit completes.

commitLogLevel

DEBUG

The logging level for logs pertaining to committing offsets.

consumerRebalanceListener

null

A rebalance listener; see Rebalancing Listeners.

consumerStartTimout

30s

The time to wait for the consumer to start before logging an error; this might happen if, say, you use a task executor with insufficient threads.

consumerTaskExecutor

SimpleAsync TaskExecutor

A task executor to run the consumer threads. The default executor creates threads named <name>-C-n; with the KafkaMessageListenerContainer, the name is the bean name; with the ConcurrentMessageListenerContainer the name is the bean name suffixed with -n where n is incremented for each child container.

deliveryAttemptHeader

false

See Delivery Attempts Header.

eosMode

BETA

Exactly Once Semantics mode; see Exactly Once Semantics.

fixTxOffsets

false

When consuming records produced by a transactional producer, and the consumer is positioned at the end of a partition, the lag can incorrectly be reported as greater than zero, due to the pseudo record used to indicate transaction commit/rollback and, possibly, the presence of rolled-back records. This does not functionally affect the consumer but some users have expressed concern that the "lag" is non-zero. Set this property to true and the container will correct such mis-reported offsets. The check is performed before the next poll to avoid adding significant complexity to the commit processing. At the time of writing, the lag will only be corrected if the consumer is configured with isolation.level=read_committed and max.poll.records is greater than 1. See KAFKA-10683 for more information.

groupId

null

Overrides the consumer group.id property; automatically set by the @KafkaListener id or groupId property.

idleBetweenPolls

0

Used to slow down deliveries by sleeping the thread between polls. The time to process a batch of records plus this value must be less than the max.poll.interval.ms consumer property.

idleEventInterval

null

When set, enables publication of ListenerContainerIdleEvent s, see Application Events and Detecting Idle and Non-Responsive Consumers.

kafkaConsumerProperties

None

Used to override any arbitrary consumer properties configured on the consumer factory.

logContainerConfig

false

Set to true to log at INFO level all container properties.

messageListener

null

The message listener.

micrometerEnabled

true

Whether or not to maintain Micrometer timers for the consumer threads.

missingTopicsFatal

false

When true prevents the container from starting if the confifgured topic(s) are not present on the broker.

monitorInterval

30s

How often to check the state of the consumer threads for NonResponsiveConsumerEvent s. See noPollThreshold and pollTimeout.

noPollThreshold

3.0

Multiplied by pollTimeOut to determine whether to publish a NonResponsiveConsumerEvent. See monitorInterval.

onlyLogRecord
Metadata

false

Set to true to show only the topic-partition@offset for a record instead of the whole consumer record (in error, debug logs etc).

pollTimeout

5000

The timeout passed into Consumer.poll().

scheduler

ThreadPool TaskScheduler

A scheduler on which to run the consumer monitor task.

shutdownTimeout

10000

The maximum time in ms to block the stop() method until all consumers stop and before publishing the container stopped event.

stopContainerWhenFenced

false

Stop the listener container if a ProducerFencedException is thrown. See After-rollback Processor for more information.

subBatchPerPartition

See desc.

When using a batch listener, if this is true, the listener is called with the results of the poll split into sub batches, one per partition. Default false except when using transactions with EOSMode.ALPHA - see Exactly Once Semantics.

syncCommitTimeout

null

The timeout to use when syncCommits is true. When not set, the container will attempt to determine the default.api.timeout.ms consumer property and use that; otherwise it will use 60 seconds.

syncCommits

true

Whether to use sync or async commits for offsets; see commitCallback.

topics
topicPattern
topicPartitions

n/a

The configured topics, topic pattern or explicitly assigned topics/partitions. Mutually exclusive; at least one must be provided; enforced by ContainerProperties constructors.

transaction
Definition

null

Set transaction properties; see Using ChainedKafkaTransactionManager for more information.

transactionManager

null

See Transactions.

Table 2. AbstractListenerContainer Properties
Property Default Description
afterRollback
Processor

DefaultAfter Rollback Processor

An AfterRollbackProcessor to invoke after a transaction is rolled back.

applicationEventPublisher

application context

The event publisher.

batchError
Handler

See desc.

An error handler for a batch listener; defaults to a RecoveringBatchErrorHandler or null if transactions are being used (errors are handled by the AfterRollbackProcessor).

beanName

bean name

The bean name of the container; suffixed with -n for child containers.

containerProperties

Container Properties

The container properties instance.

errorHandler

See desc.

An error handler for a record listener; defaults to a SeekToCurrentErrorHandler or null if transactions are being used (errors are handled by the AfterRollbackProcessor).

genericErrorHandler

See desc.

Either a batch or record error handler - see batchErrorHandler and errorHandler.

groupId

See desc.

The containerProperties.groupId, if present, otherwise the group.id property from the consumer factory.

intercept
BeforeTx

false

Determines whether the recordInterceptor is called before or after a transaction starts.

listenerId

See desc.

The bean name for user-configured containers or the id attribute of @KafkaListener s.

pause
Requested

(read only)

True if a consumer pause has been requested.

record
Interceptor

null

Set a RecordInterceptor to call before invoking the listener; does not apply to batch listeners. Also see interceptBeforeTx.

topicCheck
Timeout

30s

When the missingTopicsFatal container property is true, how long to wait, in seconds, for the describeTopics operation to complete.

Table 3. KafkaMessageListenerContainer Properties
Property Default Description
assigned
Partitions

(read only)

The partitions currently assigned to this container (explicitly or not).

assigned
Partitions
ByClientId

(read only)

The partitions currently assigned to this container (explicitly or not).

clientId
Suffix

null

Used by the concurrent container to give each child container’s consumer a unique client.id.

containerPaused

n/a

True if pause has been requested and the consumer has actually paused.

Table 4. ConcurrentMessageListenerContainer Properties
Property Default Description
alwaysClientId
Suffix

true

Set to false to suppress adding a suffix to the client.id consumer property, when the concurrency is only 1.

assigned
Partitions

(read only)

The aggregate of partitions currently assigned to this container’s child KafkaMessageListenerContainer s (explicitly or not).

assigned
Partitions
ByClientId

(read only)

The partitions currently assigned to this container’s child KafkaMessageListenerContainer s (explicitly or not), keyed by the child container’s consumer’s client.id property.

concurrency

1

The number of child KafkaMessageListenerContainer s to manage.

containerPaused

n/a

True if pause has been requested and all child containers' consumer has actually paused.

containers

n/a

A reference to all child KafkaMessageListenerContainer s.

4.1.6. Application Events

The following Spring application events are published by listener containers and their consumers:

  • ConsumerStartingEvent - published when a consumer thread is first started, before it starts polling.

  • ConsumerStartedEvent - published when a consumer is about to start polling.

  • ConsumerFailedToStartEvent - published if no ConsumerStartingEvent is published within the consumerStartTimeout container property. This event might signal that the configured task executor has insufficient threads to support the containers it is used in and their concurrency. An error message is also logged when this condition occurs.

  • ListenerContainerIdleEvent: published when no messages have been received in idleInterval (if configured).

  • ListenerContainerNoLongerIdleEvent: published when a record is consumed after previously publishing a ListenerContainerIdleEvent.

  • NonResponsiveConsumerEvent: published when the consumer appears to be blocked in the poll method.

  • ConsumerPausedEvent: published by each consumer when the container is paused.

  • ConsumerResumedEvent: published by each consumer when the container is resumed.

  • ConsumerStoppingEvent: published by each consumer just before stopping.

  • ConsumerStoppedEvent: published after the consumer is closed. See Thread Safety.

  • ContainerStoppedEvent: published when all consumers have stopped.

By default, the application context’s event multicaster invokes event listeners on the calling thread. If you change the multicaster to use an async executor, you must not invoke any Consumer methods when the event contains a reference to the consumer.

The ListenerContainerIdleEvent has the following properties:

  • source: The listener container instance that published the event.

  • container: The listener container or the parent listener container, if the source container is a child.

  • id: The listener ID (or container bean name).

  • idleTime: The time the container had been idle when the event was published.

  • topicPartitions: The topics and partitions that the container was assigned at the time the event was generated.

  • consumer: A reference to the Kafka Consumer object. For example, if the consumer’s pause() method was previously called, it can resume() when the event is received.

  • paused: Whether the container is currently paused. See Pausing and Resuming Listener Containers for more information.

The ListenerContainerNoLongerIdleEvent has the same properties, except idleTime and paused.

The NonResponsiveConsumerEvent has the following properties:

  • source: The listener container instance that published the event.

  • container: The listener container or the parent listener container, if the source container is a child.

  • id: The listener ID (or container bean name).

  • timeSinceLastPoll: The time just before the container last called poll().

  • topicPartitions: The topics and partitions that the container was assigned at the time the event was generated.

  • consumer: A reference to the Kafka Consumer object. For example, if the consumer’s pause() method was previously called, it can resume() when the event is received.

  • paused: Whether the container is currently paused. See Pausing and Resuming Listener Containers for more information.

The ConsumerPausedEvent, ConsumerResumedEvent, and ConsumerStopping events have the following properties:

  • source: The listener container instance that published the event.

  • container: The listener container or the parent listener container, if the source container is a child.

  • partitions: The TopicPartition instances involved.

The ConsumerStartingEvent, ConsumerStartingEvent, ConsumerFailedToStartEvent, ConsumerStoppedEvent and ContainerStoppedEvent events have the following properties:

  • source: The listener container instance that published the event.

  • container: The listener container or the parent listener container, if the source container is a child.

All containers (whether a child or a parent) publish ContainerStoppedEvent. For a parent container, the source and container properties are identical.

In addition, the ConsumerStoppedEvent has the following additional property:

  • reason

    • NORMAL - the consumer stopped normally (container was stopped).

    • ERROR - a java.lang.Error was thrown.

    • FENCED - the transactional producer was fenced and the stopContainerWhenFenced container property is true.

You can use this event to restart the container after such a condition:

if (event.getReason.equals(Reason.FENCED)) {
    event.getSource(MessageListenerContainer.class).start();
}
Detecting Idle and Non-Responsive Consumers

While efficient, one problem with asynchronous consumers is detecting when they are idle. You might want to take some action if no messages arrive for some period of time.

You can configure the listener container to publish a ListenerContainerIdleEvent when some time passes with no message delivery. While the container is idle, an event is published every idleEventInterval milliseconds.

To configure this feature, set the idleEventInterval on the container. The following example shows how to do so:

@Bean
public KafkaMessageListenerContainer(ConsumerFactory<String, String> consumerFactory) {
    ContainerProperties containerProps = new ContainerProperties("topic1", "topic2");
    ...
    containerProps.setIdleEventInterval(60000L);
    ...
    KafkaMessageListenerContainer<String, String> container = new KafKaMessageListenerContainer<>(...);
    return container;
}

The following example shows how to set the idleEventInterval for a @KafkaListener:

@Bean
public ConcurrentKafkaListenerContainerFactory kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<String, String> factory =
                new ConcurrentKafkaListenerContainerFactory<>();
    ...
    factory.getContainerProperties().setIdleEventInterval(60000L);
    ...
    return factory;
}

In each of these cases, an event is published once per minute while the container is idle.

In addition, if the broker is unreachable, the consumer poll() method does not exit, so no messages are received and idle events cannot be generated. To solve this issue, the container publishes a NonResponsiveConsumerEvent if a poll does not return within 3x the pollTimeout property. By default, this check is performed once every 30 seconds in each container. You can modify this behavior by setting the monitorInterval (default 30 seconds) and noPollThreshold (default 3.0) properties in the ContainerProperties when configuring the listener container. The noPollThreshold should be greater than 1.0 to avoid getting spurious events due to a race condition. Receiving such an event lets you stop the containers, thus waking the consumer so that it can stop.

Starting with version 2.6.2, if a container has published a ListenerContainerIdleEvent, it will publish a ListenerContainerNoLongerIdleEvent when a record is subsequently received.

Event Consumption

You can capture these events by implementing ApplicationListener — either a general listener or one narrowed to only receive this specific event. You can also use @EventListener, introduced in Spring Framework 4.2.

The next example combines @KafkaListener and @EventListener into a single class. You should understand that the application listener gets events for all containers, so you may need to check the listener ID if you want to take specific action based on which container is idle. You can also use the @EventListener condition for this purpose.

See Application Events for information about event properties.

The event is normally published on the consumer thread, so it is safe to interact with the Consumer object.

The following example uses both @KafkaListener and @EventListener:

public class Listener {

    @KafkaListener(id = "qux", topics = "annotated")
    public void listen4(@Payload String foo, Acknowledgment ack) {
        ...
    }

    @EventListener(condition = "event.listenerId.startsWith('qux-')")
    public void eventHandler(ListenerContainerIdleEvent event) {
        ...
    }

}
Event listeners see events for all containers. Consequently, in the preceding example, we narrow the events received based on the listener ID. Since containers created for the @KafkaListener support concurrency, the actual containers are named id-n where the n is a unique value for each instance to support the concurrency. That is why we use startsWith in the condition.
If you wish to use the idle event to stop the lister container, you should not call container.stop() on the thread that calls the listener. Doing so causes delays and unnecessary log messages. Instead, you should hand off the event to a different thread that can then stop the container. Also, you should not stop() the container instance if it is a child container. You should stop the concurrent container instead.
Current Positions when Idle

Note that you can obtain the current positions when idle is detected by implementing ConsumerSeekAware in your listener. See onIdleContainer() in Seeking to a Specific Offset.

4.1.7. Topic/Partition Initial Offset

There are several ways to set the initial offset for a partition.

When manually assigning partitions, you can set the initial offset (if desired) in the configured TopicPartitionOffset arguments (see Message Listener Containers). You can also seek to a specific offset at any time.

When you use group management where the broker assigns partitions:

  • For a new group.id, the initial offset is determined by the auto.offset.reset consumer property (earliest or latest).

  • For an existing group ID, the initial offset is the current offset for that group ID. You can, however, seek to a specific offset during initialization (or at any time thereafter).

4.1.8. Seeking to a Specific Offset

In order to seek, your listener must implement ConsumerSeekAware, which has the following methods:

void registerSeekCallback(ConsumerSeekCallback callback);

void onPartitionsAssigned(Map<TopicPartition, Long> assignments, ConsumerSeekCallback callback);

void onPartitionsRevoked(Collection<TopicPartition> partitions)

void onIdleContainer(Map<TopicPartition, Long> assignments, ConsumerSeekCallback callback);

The registerSeekCallback is called when the container is started and whenever partitions are assigned. You should use this callback when seeking at some arbitrary time after initialization. You should save a reference to the callback. If you use the same listener in multiple containers (or in a ConcurrentMessageListenerContainer), you should store the callback in a ThreadLocal or some other structure keyed by the listener Thread.

When using group management, onPartitionsAssigned is called when partitions are assigned. You can use this method, for example, for setting initial offsets for the partitions, by calling the callback. You can also use this method to associate this thread’s callback with the assigned partitions (see the example below). You must use the callback argument, not the one passed into registerSeekCallback. Starting with version 2.5.5, this method is called, even when using manual partition assignment.

onPartitionsRevoked is called when the container is stopped or Kafka revokes assignments. You should discard this thread’s callback and remove any associations to the revoked partitions.

The callback has the following methods:

void seek(String topic, int partition, long offset);

void seekToBeginning(String topic, int partition);

void seekToBeginning(Collection=<TopicPartitions> partitions);

void seekToEnd(String topic, int partition);

void seekToEnd(Collection=<TopicPartitions> partitions);

void seekRelative(String topic, int partition, long offset, boolean toCurrent);

void seekToTimestamp(String topic, int partition, long timestamp);

void seekToTimestamp(Collection<TopicPartition> topicPartitions, long timestamp);

seekRelative was added in version 2.3, to perform relative seeks.

  • offset negative and toCurrent false - seek relative to the end of the partition.

  • offset positive and toCurrent false - seek relative to the beginning of the partition.

  • offset negative and toCurrent true - seek relative to the current position (rewind).

  • offset positive and toCurrent true - seek relative to the current position (fast forward).

The seekToTimestamp methods were also added in version 2.3.

When seeking to the same timestamp for multiple partitions in the onIdleContainer or onPartitionsAssigned methods, the second method is preferred because it is more efficient to find the offsets for the timestamps in a single call to the consumer’s offsetsForTimes method. When called from other locations, the container will gather all timestamp seek requests and make one call to offsetsForTimes.

You can also perform seek operations from onIdleContainer() when an idle container is detected. See Detecting Idle and Non-Responsive Consumers for how to enable idle container detection.

The seekToBeginning method that accepts a collection is useful, for example, when processing a compacted topic and you wish to seek to the beginning every time the application is started:
public class MyListener implements ConsumerSeekAware {

...

    @Override
    public void onPartitionsAssigned(Map<TopicPartition, Long> assignments, ConsumerSeekCallback callback) {
        callback.seekToBeginning(assignments.keySet());
    }

}

To arbitrarily seek at runtime, use the callback reference from the registerSeekCallback for the appropriate thread.

Here is a trivial Spring Boot application that demonstrates how to use the callback; it sends 10 records to the topic; hitting <Enter> in the console causes all partitions to seek to the beginning.

@SpringBootApplication
public class SeekExampleApplication {

    public static void main(String[] args) {
        SpringApplication.run(SeekExampleApplication.class, args);
    }

    @Bean
    public ApplicationRunner runner(Listener listener, KafkaTemplate<String, String> template) {
        return args -> {
            IntStream.range(0, 10).forEach(i -> template.send(
                new ProducerRecord<>("seekExample", i % 3, "foo", "bar")));
            while (true) {
                System.in.read();
                listener.seekToStart();
            }
        };
    }

    @Bean
    public NewTopic topic() {
        return new NewTopic("seekExample", 3, (short) 1);
    }

}

@Component
class Listener implements ConsumerSeekAware {

    private static final Logger logger = LoggerFactory.getLogger(Listener.class);

    private final ThreadLocal<ConsumerSeekCallback> callbackForThread = new ThreadLocal<>();

    private final Map<TopicPartition, ConsumerSeekCallback> callbacks = new ConcurrentHashMap<>();

    @Override
    public void registerSeekCallback(ConsumerSeekCallback callback) {
        this.callbackForThread.set(callback);
    }

    @Override
    public void onPartitionsAssigned(Map<TopicPartition, Long> assignments, ConsumerSeekCallback callback) {
        assignments.keySet().forEach(tp -> this.callbacks.put(tp, this.callbackForThread.get()));
    }

    @Override
    public void onPartitionsRevoked(Collection<TopicPartition> partitions) {
        partitions.forEach(tp -> this.callbacks.remove(tp));
        this.callbackForThread.remove();
    }

    @Override
    public void onIdleContainer(Map<TopicPartition, Long> assignments, ConsumerSeekCallback callback) {
    }

    @KafkaListener(id = "seekExample", topics = "seekExample", concurrency = "3")
    public void listen(ConsumerRecord<String, String> in) {
        logger.info(in.toString());
    }

    public void seekToStart() {
        this.callbacks.forEach((tp, callback) -> callback.seekToBeginning(tp.topic(), tp.partition()));
    }

}

To make things simpler, version 2.3 added the AbstractConsumerSeekAware class, which keeps track of which callback is to be used for a topic/partition. The following example shows how to seek to the last record processed, in each partition, each time the container goes idle. It also has methods that allow arbitrary external calls to rewind partitions by one record.

public class SeekToLastOnIdleListener extends AbstractConsumerSeekAware {

    @KafkaListener(id = "seekOnIdle", topics = "seekOnIdle")
    public void listen(String in) {
        ...
    }

    @Override
    public void onIdleContainer(Map<org.apache.kafka.common.TopicPartition, Long> assignments,
            ConsumerSeekCallback callback) {

            assignments.keySet().forEach(tp -> callback.seekRelative(tp.topic(), tp.partition(), -1, true));
    }

    /**
    * Rewind all partitions one record.
    */
    public void rewindAllOneRecord() {
        getSeekCallbacks()
            .forEach((tp, callback) ->
                callback.seekRelative(tp.topic(), tp.partition(), -1, true));
    }

    /**
    * Rewind one partition one record.
    */
    public void rewindOnePartitionOneRecord(String topic, int partition) {
        getSeekCallbackFor(new org.apache.kafka.common.TopicPartition(topic, partition))
            .seekRelative(topic, partition, -1, true);
    }

}

Version 2.6 added convenience methods to the abstract class:

  • seekToBeginning() - seeks all assigned partitions to the beginning

  • seekToEnd() - seeks all assigned partitions to the end

  • seekToTimestamp(long time) - seeks all assigned partitions to the offset represented by that timestamp.

Example:

public class MyListener extends AbstractConsumerSeekAware {

    @KafkaListener(...)
    void listn(...) {
        ...
    }
}

public class SomeOtherBean {

    MyListener listener;

    ...

    void someMethod() {
        this.listener.seekToTimestamp(System.currentTimeMillis - 60_000);
    }

}

4.1.9. Container factory

As discussed in @KafkaListener Annotation, a ConcurrentKafkaListenerContainerFactory is used to create containers for annotated methods.

Starting with version 2.2, you can use the same factory to create any ConcurrentMessageListenerContainer. This might be useful if you want to create several containers with similar properties or you wish to use some externally configured factory, such as the one provided by Spring Boot auto-configuration. Once the container is created, you can further modify its properties, many of which are set by using container.getContainerProperties(). The following example configures a ConcurrentMessageListenerContainer:

@Bean
public ConcurrentMessageListenerContainer<String, String>(
        ConcurrentKafkaListenerContainerFactory<String, String> factory) {

    ConcurrentMessageListenerContainer<String, String> container =
        factory.createContainer("topic1", "topic2");
    container.setMessageListener(m -> { ... } );
    return container;
}
Containers created this way are not added to the endpoint registry. They should be created as @Bean definitions so that they are registered with the application context.

Starting with version 2.3.4, you can add a ContainerCustomizer to the factory to further configure each container after it has been created and configured.

@Bean
public KafkaListenerContainerFactory<?, ?> kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
            new ConcurrentKafkaListenerContainerFactory<>();
    ...
    factory.setContainerCustomizer(container -> { /* customize the container */ });
    return factory;
}

4.1.10. Thread Safety

When using a concurrent message listener container, a single listener instance is invoked on all consumer threads. Listeners, therefore, need to be thread-safe, and it is preferable to use stateless listeners. If it is not possible to make your listener thread-safe or adding synchronization would significantly reduce the benefit of adding concurrency, you can use one of a few techniques:

  • Use n containers with concurrency=1 with a prototype scoped MessageListener bean so that each container gets its own instance (this is not possible when using @KafkaListener).

  • Keep the state in ThreadLocal<?> instances.

  • Have the singleton listener delegate to a bean that is declared in SimpleThreadScope (or a similar scope).

To facilitate cleaning up thread state (for the second and third items in the preceding list), starting with version 2.2, the listener container publishes a ConsumerStoppedEvent when each thread exits. You can consume these events with an ApplicationListener or @EventListener method to remove ThreadLocal<?> instances or remove() thread-scoped beans from the scope. Note that SimpleThreadScope does not destroy beans that have a destruction interface (such as DisposableBean), so you should destroy() the instance yourself.

By default, the application context’s event multicaster invokes event listeners on the calling thread. If you change the multicaster to use an async executor, thread cleanup is not effective.

4.1.11. Monitoring

Monitoring Listener Performance

Starting with version 2.3, the listener container will automatically create and update Micrometer Timer s for the listener, if Micrometer is detected on the class path, and a single MeterRegistry is present in the application context. The timers can be disabled by setting the ContainerProperty micrometerEnabled to false.

Two timers are maintained - one for successful calls to the listener and one for failures.

The timers are named spring.kafka.listener and have the following tags:

  • name : (container bean name)

  • result : success or failure

  • exception : none or ListenerExecutionFailedException

You can add additional tags using the ContainerProperties micrometerTags property.

With the concurrent container, timers are created for each thread and the name tag is suffixed with -n where n is 0 to concurrency-1.
Monitoring KafkaTemplate Performance

Starting with version 2.5, the template will automatically create and update Micrometer Timer s for send operations, if Micrometer is detected on the class path, and a single MeterRegistry is present in the application context. The timers can be disabled by setting the template’s micrometerEnabled property to false.

Two timers are maintained - one for successful calls to the listener and one for failures.

The timers are named spring.kafka.template and have the following tags:

  • name : (template bean name)

  • result : success or failure

  • exception : none or the exception class name for failures

You can add additional tags using the template’s micrometerTags property.

Micrometer Native Metrics

Starting with version 2.5, the framework provides Factory Listeners to manage a Micrometer KafkaClientMetrics instance whenever producers and consumers are created and closed.

To enable this feature, simply add the listeners to your producer and consumer factories:

@Bean
public ConsumerFactory<String, String> myConsumerFactory() {
    Map<String, Object> configs = consumerConfigs();
    ...
    DefaultKafkaConsumerFactory<String, String> cf = new DefaultKafkaConsumerFactory<>(configs);
    ...
    cf.addListener(new MicrometerConsumerListener<String, String>(meterRegistry(),
            Collections.singletonList(new ImmutableTag("customTag", "customTagValue"))));
    ...
    return cf;
}

@Bean
public ProducerFactory<String, String> myProducerFactory() {
    Map<String, Object> configs = producerConfigs();
    configs.put(ProducerConfig.CLIENT_ID_CONFIG, "myClientId");
    ...
    DefaultKafkaProducerFactory<String, String> pf = new DefaultKafkaProducerFactory<>(configs);
    ...
    pf.addListener(new MicrometerProducerListener<String, String>(meterRegistry(),
            Collections.singletonList(new ImmutableTag("customTag", "customTagValue"))));
    ...
    return pf;
}

The consumer/producer id passed to the listener is added to the meter’s tags with tag name spring.id.

An example of obtaining one of the Kafka metrics
double count = this.meterRegistry.get("kafka.producer.node.incoming.byte.total")
                .tag("customTag", "customTagValue")
                .tag("spring.id", "myProducerFactory.myClientId-1")
                .functionCounter()
                .count()

A similar listener is provided for the StreamsBuilderFactoryBean - see KafkaStreams Micrometer Support.

4.1.12. Transactions

This section describes how Spring for Apache Kafka supports transactions.

Overview

The 0.11.0.0 client library added support for transactions. Spring for Apache Kafka adds support in the following ways:

  • KafkaTransactionManager: Used with normal Spring transaction support (@Transactional, TransactionTemplate etc).

  • Transactional KafkaMessageListenerContainer

  • Local transactions with KafkaTemplate

  • Transaction synchronization with other transaction managers

  • ChainedKafkaTransactionManager

Transactions are enabled by providing the DefaultKafkaProducerFactory with a transactionIdPrefix. In that case, instead of managing a single shared Producer, the factory maintains a cache of transactional producers. When the user calls close() on a producer, it is returned to the cache for reuse instead of actually being closed. The transactional.id property of each producer is transactionIdPrefix + n, where n starts with 0 and is incremented for each new producer, unless the transaction is started by a listener container with a record-based listener. In that case, the transactional.id is <transactionIdPrefix>.<group.id>.<topic>.<partition>. This is to properly support fencing zombies, as described here. This new behavior was added in versions 1.3.7, 2.0.6, 2.1.10, and 2.2.0. If you wish to revert to the previous behavior, you can set the producerPerConsumerPartition property on the DefaultKafkaProducerFactory to false.

While transactions are supported with batch listeners, by default, zombie fencing is not supported because a batch may contain records from multiple topics or partitions. However, starting with version 2.3.2, zombie fencing is supported if you set the container property subBatchPerPartition to true. In that case, the batch listener is invoked once per partition received from the last poll, as if each poll only returned records for a single partition. This is true by default since version 2.5 when transactions are enabled with EOSMode.ALPHA; set it to false if you are using transactions but are not concerned about zombie fencing. Also see Exactly Once Semantics.

With Spring Boot, it is only necessary to set the spring.kafka.producer.transaction-id-prefix property - Boot will automatically configure a KafkaTransactionManager bean and wire it into the listener container.

Starting with version 2.5.8, you can now configure the maxAge property on the producer factory. This is useful when using transactional producers that might lay idle for the broker’s transactional.id.expiration.ms. With current kafka-clients, this can cause a ProducerFencedException without a rebalance. By setting the maxAge to less than transactional.id.expiration.ms, the factory will refresh the producer if it is past it’s max age.
Using KafkaTransactionManager

The KafkaTransactionManager is an implementation of Spring Framework’s PlatformTransactionManager. It is provided with a reference to the producer factory in its constructor. If you provide a custom producer factory, it must support transactions. See ProducerFactory.transactionCapable().

You can use the KafkaTransactionManager with normal Spring transaction support (@Transactional, TransactionTemplate, and others). If a transaction is active, any KafkaTemplate operations performed within the scope of the transaction use the transaction’s Producer. The manager commits or rolls back the transaction, depending on success or failure. You must configure the KafkaTemplate to use the same ProducerFactory as the transaction manager.

Transaction Synchronization

This section refers to producer-only transactions (transactions not started by a listener container); see Using ChainedKafkaTransactionManager for information about synchronizing transactions when the container starts the transaction.

If you want to send records to kafka and perform some database updates, you can use normal Spring transaction management with, say, a DataSourceTransactionManager.

@Transactional
public void process(List<Thing> things) {
    things.forEach(thing -> this.kafkaTemplate.send("topic", thing));
    updateDb(things);
}

The interceptor for the @Transactional annotation starts the transaction and the KafkaTemplate will synchronize a transaction with that transaction manager; each send will participate in that transaction. When the method exits, the database transaction will commit followed by the Kafka transaction. If you wish the commits to be performed in the reverse order (Kafka first), use a ChainedTransactionManager configured with the DataSourceTransactionManager as the first TM and then the KafkaTransactionManager, in that order.

Using ChainedKafkaTransactionManager

The ChainedKafkaTransactionManager was introduced in version 2.1.3. This is a subclass of ChainedTransactionManager that can have exactly one KafkaTransactionManager. Since it is a KafkaAwareTransactionManager, the container can send the offsets to the transaction in the same way as when the container is configured with a simple KafkaTransactionManager. This provides another mechanism for synchronizing transactions without having to send the offsets to the transaction in the listener code. You should chain your transaction managers in the desired order and provide the ChainedTransactionManager in the ContainerProperties.

See Example of Transactions with ChainedKafkaTransactionManager for an example application that chains JDBC and Kafka transactions.

Starting with version 2.5.4, you can configure a TransactionDefinition in the ContainerProperties; its properties will be copied to the container’s TransactionTemplate used to start transactions. This allows, for example, setting a transaction timeout for other transaction managers within the ChainedKafkaTransactionManager. It does not, however, make sense to set a propagation behavior because the container would always need to start a new transaction; anything other than REQUIRED or REQUIRES_NEW will be rejected.

KafkaTemplate Local Transactions

You can use the KafkaTemplate to execute a series of operations within a local transaction. The following example shows how to do so:

boolean result = template.executeInTransaction(t -> {
    t.sendDefault("thing1", "thing2");
    t.sendDefault("cat", "hat");
    return true;
});

The argument in the callback is the template itself (this). If the callback exits normally, the transaction is committed. If an exception is thrown, the transaction is rolled back.

If there is a KafkaTransactionManager (or synchronized) transaction in process, it is not used. Instead, a new "nested" transaction is used.
transactionIdPrefix

As mentioned in the overview, the producer factory is configured with this property to build the producer transactional.id property. There is a dichotomy when specifying this property in that, when running multiple instances of the application with EOSMode.ALPHA, it must be the same on all instances to satisfy fencing zombies (also mentioned in the overview) when producing records on a listener container thread. However, when producing records using transactions that are not started by a listener container, the prefix has to be different on each instance. Version 2.3, makes this simpler to configure, especially in a Spring Boot application. In previous versions, you had to create two producer factories and KafkaTemplate s - one for producing records on a listener container thread and one for stand-alone transactions started by kafkaTemplate.executeInTransaction() or by a transaction interceptor on a @Transactional method.

Now, you can override the factory’s transactionalIdPrefix on the KafkaTemplate and the KafkaTransactionManager.

When using a transaction manager and template for a listener container, you would normally leave this to default to the producer factory’s property. This value should be the same for all application instances when using EOSMode.ALPHA. With EOSMode.BETA it is no longer necessary to use the same transactional.id, even for consumer-initiated transactions; in fact, it must be unique on each instance the same as producer-initiated transactions. For transactions started by the template (or the transaction manager for @Transaction) you should set the property on the template and transaction manager respectively. This property must have a different value on each application instance.

This problem (different rules for transactional.id) has been eliminated when EOSMode.BETA is being used (with broker versions >= 2.5); see Exactly Once Semantics.

KafkaTemplate Transactional and non-Transactional Publishing

Normally, when a KafkaTemplate is transactional (configured with a transaction-capable producer factory), transactions are required. The transaction can be started by a TransactionTemplate, a @Transactional method, calling executeInTransaction, or by a listener container, when configured with a KafkaTransactionManager. Any attempt to use the template outside the scope of a transaction results in the template throwing an IllegalStateException. Starting with version 2.4.3, you can set the template’s allowNonTransactional property to true. In that case, the template will allow the operation to run without a transaction, by calling the ProducerFactory 's createNonTransactionalProducer() method; the producer will be cached, or thread-bound, as normal for reuse. See Using DefaultKafkaProducerFactory.

Transactions with Batch Listeners

When a listener fails while transactions are being used, the AfterRollbackProcessor is invoked to take some action after the rollback occurs. When using the default AfterRollbackProcessor with a record listener, seeks are performed so that the failed record will be redelivered. With a batch listener, however, the whole batch will be redelivered because the framework doesn’t know which record in the batch failed. See After-rollback Processor for more information.

When using a batch listener, version 2.4.2 introduced an alternative mechanism to deal with failures while processing a batch; the BatchToRecordAdapter. When a container factory with batchListener set to true is configured with a BatchToRecordAdapter, the listener is invoked with one record at a time. This enables error handling within the batch, while still making it possible to stop processing the entire batch, depending on the exception type. A default BatchToRecordAdapter is provided, that can be configured with a standard ConsumerRecordRecoverer such as the DeadLetterPublishingRecoverer. The following test case configuration snippet illustrates how to use this feature:

public static class TestListener {

    final List<String> values = new ArrayList<>();

    @KafkaListener(id = "batchRecordAdapter", topics = "test")
    public void listen(String data) {
        values.add(data);
        if ("bar".equals(data)) {
            throw new RuntimeException("reject partial");
        }
    }

}

@Configuration
@EnableKafka
public static class Config {

    ConsumerRecord<?, ?> failed;

    @Bean
    public TestListener test() {
        return new TestListener();
    }

    @Bean
    public ConsumerFactory<?, ?> consumerFactory() {
        return mock(ConsumerFactory.class);
    }

    @Bean
    public ConcurrentKafkaListenerContainerFactory<String, String> kafkaListenerContainerFactory() {
        ConcurrentKafkaListenerContainerFactory factory = new ConcurrentKafkaListenerContainerFactory();
        factory.setConsumerFactory(consumerFactory());
        factory.setBatchListener(true);
        factory.setBatchToRecordAdapter(new DefaultBatchToRecordAdapter<>((record, ex) ->  {
            this.failed = record;
        }));
        return factory;
    }

}

4.1.13. Exactly Once Semantics

You can provide a listener container with a KafkaAwareTransactionManager instance. When so configured, the container starts a transaction before invoking the listener. Any KafkaTemplate operations performed by the listener participate in the transaction. If the listener successfully processes the record (or multiple records, when using a BatchMessageListener), the container sends the offset(s) to the transaction by using producer.sendOffsetsToTransaction()), before the transaction manager commits the transaction. If the listener throws an exception, the transaction is rolled back and the consumer is repositioned so that the rolled-back record(s) can be retrieved on the next poll. See After-rollback Processor for more information and for handling records that repeatedly fail.

Using transactions enables Exactly Once Semantics (EOS).

This means that, for a read→process-write sequence, it is guaranteed that the sequence is completed exactly once. (The read and process are have at least once semantics).

Spring for Apache Kafka version 2.5 and later supports two EOS modes:

  • ALPHA - aka transactional.id fencing (since version 0.11.0.0)

  • BETA - aka fetch-offset-request fencing (since version 2.5)

With mode ALPHA, the producer is "fenced" if another instance with the same transactional.id is started. Spring manages this by using a Producer for each group.id/topic/partition; when a rebalance occurs a new instance will use the same transactional.id and the old producer is fenced.

With mode BETA, it is not necessary to have a producer for each group.id/topic/partition because consumer metadata is sent along with the offsets to the transaction and the broker can determine if the producer is fenced using that information instead.

Starting with version 2.6, the default EOSMode is BETA.

To configure the container to use mode ALPHA, set the container property EOSMode to ALPHA, to revert to the previous behavior.

With BETA, your brokers must be version 2.5 or later, however with kafka-clients version 2.6, the producer will automatically fall back to ALPHA if the broker does not support BETA. The DefaultKafkaProducerFactory is configured to enable that behavior. If your brokers are earlier than 2.5, be sure to leave the DefaultKafkaProducerFactory producerPerConsumerPartition set to true and, if you are using a batch listener, you should set subBatchPerPartition to true.

When your brokers are upgraded to 2.5 or later, the producer will automatically switch to using mode BETA, but the number of producers will remain as before. You can then do a rolling upgrade of your application with producerPerConsumerPartition set to false to reduce the number of producers; you should also no longer set the subBatchPerPartition container property.

If your brokers are already 2.5 or newer, you should set the DefaultKafkaProducerFactory producerPerConsumerPartition property to false, to reduce the number of producers needed.

When using EOSMode.BETA with producerPerConsumerPartition=false the transactional.id must be unique across all application instances.

When using BETA mode, it is no longer necessary to set the subBatchPerPartition to true; it will default to false when the EOSMode is BETA.

Refer to KIP-447 for more information.

4.1.14. Wiring Spring Beans into Producer/Consumer Interceptors

Apache Kafka provides a mechanism to add interceptors to producers and consumers. These objects are managed by Kafka, not Spring, and so normal Spring dependency injection won’t work for wiring in dependent Spring Beans. However, you can manually wire in those dependencies using the interceptor config() method. The following Spring Boot application shows how to do this by overriding boot’s default factories to add some dependent bean into the configuration properties.

@SpringBootApplication
public class Application {

    public static void main(String[] args) {
        SpringApplication.run(Application.class, args);
    }

    @Bean
    public ConsumerFactory<?, ?> kafkaConsumerFactory(KafkaProperties properties, SomeBean someBean) {
        Map<String, Object> consumerProperties = properties.buildConsumerProperties();
        consumerProperties.put(ConsumerConfig.INTERCEPTOR_CLASSES_CONFIG, MyConsumerInterceptor.class.getName());
        consumerProperties.put("some.bean", someBean);
        return new DefaultKafkaConsumerFactory<>(consumerProperties);
    }

    @Bean
    public ProducerFactory<?, ?> kafkaProducerFactory(KafkaProperties properties, SomeBean someBean) {
        Map<String, Object> producerProperties = properties.buildProducerProperties();
        producerProperties.put(ProducerConfig.INTERCEPTOR_CLASSES_CONFIG, MyProducerInterceptor.class.getName());
        producerProperties.put("some.bean", someBean);
        DefaultKafkaProducerFactory<?, ?> factory = new DefaultKafkaProducerFactory<>(producerProperties);
        String transactionIdPrefix = properties.getProducer()
                .getTransactionIdPrefix();
        if (transactionIdPrefix != null) {
            factory.setTransactionIdPrefix(transactionIdPrefix);
        }
        return factory;
    }

    @Bean
    public SomeBean someBean() {
        return new SomeBean();
    }

    @KafkaListener(id = "kgk897", topics = "kgh897")
    public void listen(String in) {
        System.out.println("Received " + in);
    }

    @Bean
    public ApplicationRunner runner(KafkaTemplate<String, String> template) {
        return args -> template.send("kgh897", "test");
    }

    @Bean
    public NewTopic kRequests() {
        return TopicBuilder.name("kgh897")
            .partitions(1)
            .replicas(1)
            .build();
    }

}
public class SomeBean {

    public void someMethod(String what) {
        System.out.println(what + " in my foo bean");
    }

}
public class MyProducerInterceptor implements ProducerInterceptor<String, String> {

    private SomeBean bean;

    @Override
    public void configure(Map<String, ?> configs) {
        this.bean = (SomeBean) configs.get("some.bean");
    }

    @Override
    public ProducerRecord<String, String> onSend(ProducerRecord<String, String> record) {
        this.bean.someMethod("producer interceptor");
        return record;
    }

    @Override
    public void onAcknowledgement(RecordMetadata metadata, Exception exception) {
    }

    @Override
    public void close() {
    }

}
public class MyConsumerInterceptor implements ConsumerInterceptor<String, String> {

    private SomeBean bean;

    @Override
    public void configure(Map<String, ?> configs) {
        this.bean = (SomeBean) configs.get("some.bean");
    }

    @Override
    public ConsumerRecords<String, String> onConsume(ConsumerRecords<String, String> records) {
        this.bean.someMethod("consumer interceptor");
        return records;
    }

    @Override
    public void onCommit(Map<TopicPartition, OffsetAndMetadata> offsets) {
    }

    @Override
    public void close() {
    }

}

Result:

producer interceptor in my foo bean
consumer interceptor in my foo bean
Received test

4.1.15. Pausing and Resuming Listener Containers

Version 2.1.3 added pause() and resume() methods to listener containers. Previously, you could pause a consumer within a ConsumerAwareMessageListener and resume it by listening for a ListenerContainerIdleEvent, which provides access to the Consumer object. While you could pause a consumer in an idle container by using an event listener, in some cases, this was not thread-safe, since there is no guarantee that the event listener is invoked on the consumer thread. To safely pause and resume consumers, you should use the pause and resume methods on the listener containers. A pause() takes effect just before the next poll(); a resume() takes effect just after the current poll() returns. When a container is paused, it continues to poll() the consumer, avoiding a rebalance if group management is being used, but it does not retrieve any records. See the Kafka documentation for more information.

Starting with version 2.1.5, you can call isPauseRequested() to see if pause() has been called. However, the consumers might not have actually paused yet. isConsumerPaused() returns true if all Consumer instances have actually paused.

In addition (also since 2.1.5), ConsumerPausedEvent and ConsumerResumedEvent instances are published with the container as the source property and the TopicPartition instances involved in the partitions property.

The following simple Spring Boot application demonstrates by using the container registry to get a reference to a @KafkaListener method’s container and pausing or resuming its consumers as well as receiving the corresponding events:

@SpringBootApplication
public class Application implements ApplicationListener<KafkaEvent> {

    public static void main(String[] args) {
        SpringApplication.run(Application.class, args).close();
    }

    @Override
    public void onApplicationEvent(KafkaEvent event) {
        System.out.println(event);
    }

    @Bean
    public ApplicationRunner runner(KafkaListenerEndpointRegistry registry,
            KafkaTemplate<String, String> template) {
        return args -> {
            template.send("pause.resume.topic", "thing1");
            Thread.sleep(10_000);
            System.out.println("pausing");
            registry.getListenerContainer("pause.resume").pause();
            Thread.sleep(10_000);
            template.send("pause.resume.topic", "thing2");
            Thread.sleep(10_000);
            System.out.println("resuming");
            registry.getListenerContainer("pause.resume").resume();
            Thread.sleep(10_000);
        };
    }

    @KafkaListener(id = "pause.resume", topics = "pause.resume.topic")
    public void listen(String in) {
        System.out.println(in);
    }

    @Bean
    public NewTopic topic() {
        return TopicBuilder.name("pause.resume.topic")
            .partitions(2)
            .replicas(1)
            .build();
    }

}

The following listing shows the results of the preceding example:

partitions assigned: [pause.resume.topic-1, pause.resume.topic-0]
thing1
pausing
ConsumerPausedEvent [partitions=[pause.resume.topic-1, pause.resume.topic-0]]
resuming
ConsumerResumedEvent [partitions=[pause.resume.topic-1, pause.resume.topic-0]]
thing2

4.1.16. Serialization, Deserialization, and Message Conversion

Overview

Apache Kafka provides a high-level API for serializing and deserializing record values as well as their keys. It is present with the org.apache.kafka.common.serialization.Serializer<T> and org.apache.kafka.common.serialization.Deserializer<T> abstractions with some built-in implementations. Meanwhile, we can specify serializer and deserializer classes by using Producer or Consumer configuration properties. The following example shows how to do so:

props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, IntegerDeserializer.class);
props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, StringDeserializer.class);
...
props.put(ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG, IntegerSerializer.class);
props.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, StringSerializer.class);

For more complex or particular cases, the KafkaConsumer (and, therefore, KafkaProducer) provides overloaded constructors to accept Serializer and Deserializer instances for keys and values, respectively.

When you use this API, the DefaultKafkaProducerFactory and DefaultKafkaConsumerFactory also provide properties (through constructors or setter methods) to inject custom Serializer and Deserializer instances into the target Producer or Consumer. Also, you can pass in Supplier<Serializer> or Supplier<Deserializer> instances through constructors - these Supplier s are called on creation of each Producer or Consumer.

String serialization

Since version 2.5, Spring Kafka provides ToStringSerializer and ParseStringDeserializer classes that uses String representation of entities. They rely on methods toString and some Function<String> or BiFuntion<String, Headers> to parse the String and populate properties of an instance. Usually, this would invoke some static method on the class, such as parse:

ToStringSerializer<Thing> thingSerializer = new ToStringSerializer<>();
//...
ParseStringDeserializer<Thing> deserializer = new ParseStringDeserializer<>(Thing::parse);

By default, the ToStringSerializer is configured to convey type information about the serialized entity in the record Headers. You can disable this by setting the addTypeInfo property to false. This information can be used by ParseStringDeserializer on the receiving side.

  • ToStringSerializer.ADD_TYPE_INFO_HEADERS (default true): You can set it to false to disable this feature on the ToStringSerializer (sets the addTypeInfo property).

ParseStringDeserializer<Object> deserializer = new ParseStringDeserializer<>((str, headers) -> {
    byte[] header = headers.lastHeader(ToStringSerializer.VALUE_TYPE).value();
    String entityType = new String(header);

    if (entityType.contains("Thing")) {
        return Thing.parse(str);
    }
    else {
        // ...parsing logic
    }
});

You can configure the Charset used to convert String to/from byte[] with the default being UTF-8.

You can configure the deserializer with the name of the parser method using ConsumerConfig properties:

  • ParseStringDeserializer.KEY_PARSER

  • ParseStringDeserializer.VALUE_PARSER

The properties must contain the fully qualified name of the class followed by the method name, separated by a period .. The method must be static and have a signature of either (String, Headers) or (String).

A ToFromStringSerde is also provided, for use with Kafka Streams.

JSON

Spring for Apache Kafka also provides JsonSerializer and JsonDeserializer implementations that are based on the Jackson JSON object mapper. The JsonSerializer allows writing any Java object as a JSON byte[]. The JsonDeserializer requires an additional Class<?> targetType argument to allow the deserialization of a consumed byte[] to the proper target object. The following example shows how to create a JsonDeserializer:

JsonDeserializer<Thing> thingDeserializer = new JsonDeserializer<>(Thing.class);

You can customize both JsonSerializer and JsonDeserializer with an ObjectMapper. You can also extend them to implement some particular configuration logic in the configure(Map<String, ?> configs, boolean isKey) method.

Starting with version 2.3, all the JSON-aware components are configured by default with a JacksonUtils.enhancedObjectMapper() instance, which comes with the MapperFeature.DEFAULT_VIEW_INCLUSION and DeserializationFeature.FAIL_ON_UNKNOWN_PROPERTIES features disabled. Also such an instance is supplied with well-known modules for custom data types, such a Java time and Kotlin support. See JacksonUtils.enhancedObjectMapper() JavaDocs for more information. This method also registers a org.springframework.kafka.support.JacksonMimeTypeModule for org.springframework.util.MimeType objects serialization into the plain string for inter-platform compatibility over the network. A JacksonMimeTypeModule can be registered as a bean in the application context and it will be auto-configured into Spring Boot ObjectMapper instance.

Also starting with version 2.3, the JsonDeserializer provides TypeReference-based constructors for better handling of target generic container types.

Starting with version 2.1, you can convey type information in record Headers, allowing the handling of multiple types. In addition, you can configure the serializer and deserializer by using the following Kafka properties. They have no effect if you have provided Serializer and Deserializer instances for KafkaConsumer and KafkaProducer, respectively.

Configuration Properties
  • JsonSerializer.ADD_TYPE_INFO_HEADERS (default true): You can set it to false to disable this feature on the JsonSerializer (sets the addTypeInfo property).

  • JsonSerializer.TYPE_MAPPINGS (default empty): See Mapping Types.

  • JsonDeserializer.USE_TYPE_INFO_HEADERS (default true): You can set it to false to ignore headers set by the serializer.

  • JsonDeserializer.REMOVE_TYPE_INFO_HEADERS (default true): You can set it to false to retain headers set by the serializer.

  • JsonDeserializer.KEY_DEFAULT_TYPE: Fallback type for deserialization of keys if no header information is present.

  • JsonDeserializer.VALUE_DEFAULT_TYPE: Fallback type for deserialization of values if no header information is present.

  • JsonDeserializer.TRUSTED_PACKAGES (default java.util, java.lang): Comma-delimited list of package patterns allowed for deserialization. * means deserialize all.

  • JsonDeserializer.TYPE_MAPPINGS (default empty): See Mapping Types.

  • JsonDeserializer.KEY_TYPE_METHOD (default empty): See Using Methods to Determine Types.

  • JsonDeserializer.VALUE_TYPE_METHOD (default empty): See Using Methods to Determine Types.

Starting with version 2.2, the type information headers (if added by the serializer) are removed by the deserializer. You can revert to the previous behavior by setting the removeTypeHeaders property to false, either directly on the deserializer or with the configuration property described earlier.

Mapping Types

Starting with version 2.2, when using JSON, you can now provide type mappings by using the properties in the preceding list. Previously, you had to customize the type mapper within the serializer and deserializer. Mappings consist of a comma-delimited list of token:className pairs. On outbound, the payload’s class name is mapped to the corresponding token. On inbound, the token in the type header is mapped to the corresponding class name.

The following example creates a set of mappings:

senderProps.put(ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG, JsonSerializer.class);
senderProps.put(JsonSerializer.TYPE_MAPPINGS, "cat:com.mycat.Cat, hat:com.myhat.hat");
...
consumerProps.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, JsonDeserializer.class);
consumerProps.put(JsonDeSerializer.TYPE_MAPPINGS, "cat:com.yourcat.Cat, hat:com.yourhat.hat");
The corresponding objects must be compatible.

If you use Spring Boot, you can provide these properties in the application.properties (or yaml) file. The following example shows how to do so:

spring.kafka.producer.value-serializer=org.springframework.kafka.support.serializer.JsonSerializer
spring.kafka.producer.properties.spring.json.type.mapping=cat:com.mycat.Cat,hat:com.myhat.Hat

You can perform only simple configuration with properties. For more advanced configuration (such as using a custom ObjectMapper in the serializer and deserializer), you should use the producer and consumer factory constructors that accept a pre-built serializer and deserializer. The following Spring Boot example overrides the default factories:

@Bean
public ConsumerFactory<Foo, Bar> kafkaConsumerFactory(KafkaProperties properties,
    JsonDeserializer customDeserializer) {

    return new DefaultKafkaConsumerFactory<>(properties.buildConsumerProperties(),
        customDeserializer, customDeserializer);
}

@Bean
public ProducerFactory<Foo, Bar> kafkaProducerFactory(KafkaProperties properties,
    JsonSerializer customSerializer) {

    return new DefaultKafkaProducerFactory<>(properties.buildProducerProperties(),
        customSerializer, customSerializer);
}

Setters are also provided, as an alternative to using these constructors.

Starting with version 2.2, you can explicitly configure the deserializer to use the supplied target type and ignore type information in headers by using one of the overloaded constructors that have a boolean useHeadersIfPresent (which is true by default). The following example shows how to do so:

DefaultKafkaConsumerFactory<Integer, Cat1> cf = new DefaultKafkaConsumerFactory<>(props,
        new IntegerDeserializer(), new JsonDeserializer<>(Cat1.class, false));
Using Methods to Determine Types

Starting with version 2.5, you can now configure the deserializer, via properties, to invoke a method to determine the target type. If present, this will override any of the other techniques discussed above. This can be useful if the data is published by an application that does not use the Spring serializer and you need to deserialize to different types depending on the data, or other headers. Set these properties to the method name - a fully qualified class name followed by the method name, separated by a period .. The method must be declared as public static, have one of three signatures (String topic, byte[] data, Headers headers), (byte[] data, Headers headers) or (byte[] data) and return a Jackson JavaType.

  • JsonDeserializer.KEY_TYPE_METHOD : spring.json.key.type.method

  • JsonDeserializer.VALUE_TYPE_METHOD : spring.json.value.type.method

You can use arbitrary headers or inspect the data to determine the type.

Example
JavaType thing1Type = TypeFactory.defaultInstance().constructType(Thing1.class);

JavaType thing1Type = TypeFactory.defaultInstance().constructType(Thing2.class);

public static JavaType thingOneOrThingTwo(byte[] data, Headers headers) {
    // {"thisIsAFieldInThing1":"value", ...
    if (data[21] == '1') {
        return thing1Type;
    }
    else {
        return thing2Type;
    }
}

For more sophisticated data inspection consider using JsonPath or similar but, the simpler the test to determine the type, the more efficient the process will be.

The following is an example of creating the deserializer programmatically (when providing the consumer factory with the deserializer in the constructor):

JsonDeserializer<Object> deser = new JsonDeserializer<>()
        .trustedPackages("*")
        .typeResolver(SomeClass::thing1Thing2JavaTypeForTopic);

...

public static JavaType thing1Thing2JavaTypeForTopic(String topic, byte[] data, Headers headers) {
    ...
}
Programmatic Construction

When constructing the serializer/deserializer programmatically for use in the producer/consumer factory, since version 2.3, you can use the fluent API, which simplifies configuration.

The following example assumes you are using Spring Boot:

@Bean
public DefaultKafkaProducerFactory pf(KafkaProperties properties) {
    Map<String, Object> props = properties.buildProducerProperties();
    DefaultKafkaProducerFactory pf = new DefaultKafkaProducerFactory(props,
        new JsonSerializer<>(MyKeyType.class)
            .forKeys()
            .noTypeInfo(),
        new JsonSerializer<>(MyValueType.class)
            .noTypeInfo());
}

@Bean
public DefaultKafkaConsumerFactory pf(KafkaProperties properties) {
    Map<String, Object> props = properties.buildConsumerProperties();
    DefaultKafkaConsumerFactory pf = new DefaultKafkaConsumerFactory(props,
        new JsonDeserializer<>(MyKeyType.class)
            .forKeys()
            .ignoreTypeHeaders(),
        new JsonDeserializer<>(MyValueType.class)
            .ignoreTypeHeaders());
}

To provide type mapping programmatically, similar to Using Methods to Determine Types, use the typeFunction property.

Example
JsonDeserializer<Object> deser = new JsonDeserializer<>()
        .trustedPackages("*")
        .typeFunction(MyUtils::thingOneOrThingTwo);
Delegating Serializer and Deserializer

Version 2.3 introduced the DelegatingSerializer and DelegatingDeserializer, which allow producing and consuming records with different key and/or value types. Producers must set a header DelegatingSerializer.VALUE_SERIALIZATION_SELECTOR to a selector value that is used to select which serializer to use for the value and DelegatingSerializer.KEY_SERIALIZATION_SELECTOR for the key; if a match is not found, an IllegalStateException is thrown.

For incoming records, the deserializer uses the same headers to select the deserializer to use; if a match is not found or the header is not present, the raw byte[] is returned.

You can configure the map of selector to Serializer / Deserializer via a constructor, or you can configure it via Kafka producer/consumer properties with the keys DelegatingSerializer.VALUE_SERIALIZATION_SELECTOR_CONFIG and DelegatingSerializer.KEY_SERIALIZATION_SELECTOR_CONFIG. For the serializer, the producer property can be a Map<String, Object> where the key is the selector and the value is a Serializer instance, a serializer Class or the class name. The property can also be a String of comma-delimited map entries, as shown below.

For the deserializer, the consumer property can be a Map<String, Object> where the key is the selector and the value is a Deserializer instance, a deserializer Class or the class name. The property can also be a String of comma-delimited map entries, as shown below.

To configure using properties, use the following syntax:

producerProps.put(DelegatingSerializer.VALUE_SERIALIZATION_SELECTOR_CONFIG,
    "thing1:com.example.MyThing1Serializer, thing2:com.example.MyThing2Serializer")

consumerProps.put(DelegatingDeserializer.VALUE_SERIALIZATION_SELECTOR_CONFIG,
    "thing1:com.example.MyThing1Deserializer, thing2:com.example.MyThing2Deserializer")

Producers would then set the DelegatingSerializer.VALUE_SERIALIZATION_SELECTOR header to thing1 or thing2.

This technique supports sending different types to the same topic (or different topics).

Starting with version 2.5.1, it is not necessary to set the selector header, if the type (key or value) is one of the standard types supported by Serdes (Long, Integer, etc). Instead, the serializer will set the header to the class name of the type. It is not necessary to configure serializers or deserializers for these types, they will be created (once) dynamically.

For another technique to send different types to different topics, see Using RoutingKafkaTemplate.

Retrying Deserializer

The RetryingDeserializer uses a delegate Deserializer and RetryTemplate to retry deserialization when the delegate might have transient errors, such a network issues, during deserialization.

ConsumerFactory cf = new DefaultKafkaConsumerFactory(myConsumerConfigs,
    new RetryingDeserializer(myUnreliableKeyDeserializer, retryTemplate),
    new RetryingDeserializer(myUnreliableValueDeserializer, retryTemplate));

Refer to the spring-retry project for configuration of the RetryTemplate with a retry policy, back off policy, etc.

Spring Messaging Message Conversion

Although the Serializer and Deserializer API is quite simple and flexible from the low-level Kafka Consumer and Producer perspective, you might need more flexibility at the Spring Messaging level, when using either @KafkaListener or Spring Integration. To let you easily convert to and from org.springframework.messaging.Message, Spring for Apache Kafka provides a MessageConverter abstraction with the MessagingMessageConverter implementation and its JsonMessageConverter (and subclasses) customization. You can inject the MessageConverter into a KafkaTemplate instance directly and by using AbstractKafkaListenerContainerFactory bean definition for the @KafkaListener.containerFactory() property. The following example shows how to do so:

@Bean
public KafkaListenerContainerFactory<?, ?> kafkaJsonListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
        new ConcurrentKafkaListenerContainerFactory<>();
    factory.setConsumerFactory(consumerFactory());
    factory.setMessageConverter(new JsonMessageConverter());
    return factory;
}
...
@KafkaListener(topics = "jsonData",
                containerFactory = "kafkaJsonListenerContainerFactory")
public void jsonListener(Cat cat) {
...
}

When you use a @KafkaListener, the parameter type is provided to the message converter to assist with the conversion.

This type inference can be achieved only when the @KafkaListener annotation is declared at the method level. With a class-level @KafkaListener, the payload type is used to select which @KafkaHandler method to invoke, so it must already have been converted before the method can be chosen.

On the consumer side, you can configure a JsonMessageConverter; it can handle ConsumerRecord values of type byte[], Bytes and String so should be used in conjunction with a ByteArrayDeserializer, BytesDeserializer or StringDeserializer. (byte[] and Bytes are more efficient because they avoid an unnecessary byte[] to String conversion). You can also configure the specific subclass of JsonMessageConverter corresponding to the deserializer, if you so wish.

On the producer side, when you use Spring Integration or the KafkaTemplate.send(Message<?> message) method (see Using KafkaTemplate), you must configure a message converter that is compatible with the configured Kafka Serializer.

  • StringJsonMessageConverter with StringSerializer

  • BytesJsonMessageConverter with BytesSerializer

  • ByteArrayJsonMessageConverter with ByteArraySerializer

Again, using byte[] or Bytes is more efficient because they avoid a String to byte[] conversion.

For convenience, starting with version 2.3, the framework also provides a StringOrBytesSerializer which can serialize all three value types so it can be used with any of the message converters.

Using Spring Data Projection Interfaces

Starting with version 2.1.1, you can convert JSON to a Spring Data Projection interface instead of a concrete type. This allows very selective, and low-coupled bindings to data, including the lookup of values from multiple places inside the JSON document. For example the following interface can be defined as message payload type:

interface SomeSample {

  @JsonPath({ "$.username", "$.user.name" })
  String getUsername();

}
@KafkaListener(id="projection.listener", topics = "projection")
public void projection(SomeSample in) {
    String username = in.getUsername();
    ...
}

Accessor methods will be used to lookup the property name as field in the received JSON document by default. The @JsonPath expression allows customization of the value lookup, and even to define multiple JSON Path expressions, to lookup values from multiple places until an expression returns an actual value.

To enable this feature, use a ProjectingMessageConverter configured with an appropriate delegate converter (used for outbound conversion and converting non-projection interfaces). You must also add spring-data:spring-data-commons and com.jayway.jsonpath:json-path to the class path.

When used as the parameter to a @KafkaListener method, the interface type is automatically passed to the converter as normal.

Using ErrorHandlingDeserializer

When a deserializer fails to deserialize a message, Spring has no way to handle the problem, because it occurs before the poll() returns. To solve this problem, the ErrorHandlingDeserializer has been introduced. This deserializer delegates to a real deserializer (key or value). If the delegate fails to deserialize the record content, the ErrorHandlingDeserializer returns a null value and a DeserializationException in a header that contains the cause and the raw bytes. When you use a record-level MessageListener, if the ConsumerRecord contains a DeserializationException header for either the key or value, the container’s ErrorHandler is called with the failed ConsumerRecord. The record is not passed to the listener.

Alternatively, you can configure the ErrorHandlingDeserializer to create a custom value by providing a failedDeserializationFunction, which is a Function<FailedDeserializationInfo, T>. This function is invoked to create an instance of T, which is passed to the listener in the usual fashion. An object of type FailedDeserializationInfo, which contains all the contextual information is provided to the function. You can find the DeserializationException (as a serialized Java object) in headers. See the Javadoc for the ErrorHandlingDeserializer for more information.

When you use a BatchMessageListener, you must provide a failedDeserializationFunction. Otherwise, the batch of records are not type safe.

You can use the DefaultKafkaConsumerFactory constructor that takes key and value Deserializer objects and wire in appropriate ErrorHandlingDeserializer instances that you have configured with the proper delegates. Alternatively, you can use consumer configuration properties (which are used by the ErrorHandlingDeserializer) to instantiate the delegates. The property names are ErrorHandlingDeserializer.KEY_DESERIALIZER_CLASS and ErrorHandlingDeserializer.VALUE_DESERIALIZER_CLASS. The property value can be a class or class name. The following example shows how to set these properties:

... // other props
props.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, ErrorHandlingDeserializer.class);
props.put(ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG, ErrorHandlingDeserializer.class);
props.put(ErrorHandlingDeserializer.KEY_DESERIALIZER_CLASS, JsonDeserializer.class);
props.put(JsonDeserializer.KEY_DEFAULT_TYPE, "com.example.MyKey")
props.put(ErrorHandlingDeserializer.VALUE_DESERIALIZER_CLASS, JsonDeserializer.class.getName());
props.put(JsonDeserializer.VALUE_DEFAULT_TYPE, "com.example.MyValue")
props.put(JsonDeserializer.TRUSTED_PACKAGES, "com.example")
return new DefaultKafkaConsumerFactory<>(props);

The following example uses a failedDeserializationFunction.

public class BadFoo extends Foo {

  private final FailedDeserializationInfo failedDeserializationInfo;

  public BadFoo(FailedDeserializationInfo failedDeserializationInfo) {
    this.failedDeserializationInfo = failedDeserializationInfo;
  }

  public FailedDeserializationInfo getFailedDeserializationInfo() {
    return this.failedDeserializationInfo;
  }

}

public class FailedFooProvider implements Function<FailedDeserializationInfo, Foo> {

  @Override
  public Foo apply(FailedDeserializationInfo info) {
    return new BadFoo(info);
  }

}

The preceding example uses the following configuration:

...
consumerProps.put(ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG, ErrorHandlingDeserializer.class);
consumerProps.put(ErrorHandlingDeserializer.VALUE_DESERIALIZER_CLASS, JsonDeserializer.class);
consumerProps.put(ErrorHandlingDeserializer.VALUE_FUNCTION, FailedFooProvider.class);
...
Payload Conversion with Batch Listeners

You can also use a JsonMessageConverter within a BatchMessagingMessageConverter to convert batch messages when you use a batch listener container factory. See Serialization, Deserialization, and Message Conversion and [message-conversion] for more information.

By default, the type for the conversion is inferred from the listener argument. If you configure the JsonMessageConverter with a DefaultJackson2TypeMapper that has its TypePrecedence set to TYPE_ID (instead of the default INFERRED), the converter uses the type information in headers (if present) instead. This allows, for example, listener methods to be declared with interfaces instead of concrete classes. Also, the type converter supports mapping, so the deserialization can be to a different type than the source (as long as the data is compatible). This is also useful when you use class-level @KafkaListener instances where the payload must have already been converted to determine which method to invoke. The following example creates beans that use this method:

@Bean
public KafkaListenerContainerFactory<?, ?> kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
            new ConcurrentKafkaListenerContainerFactory<>();
    factory.setConsumerFactory(consumerFactory());
    factory.setBatchListener(true);
    factory.setMessageConverter(new BatchMessagingMessageConverter(converter()));
    return factory;
}

@Bean
public JsonMessageConverter converter() {
    return new JsonMessageConverter();
}

Note that, for this to work, the method signature for the conversion target must be a container object with a single generic parameter type, such as the following:

@KafkaListener(topics = "blc1")
public void listen(List<Foo> foos, @Header(KafkaHeaders.OFFSET) List<Long> offsets) {
    ...
}

Note that you can still access the batch headers.

If the batch converter has a record converter that supports it, you can also receive a list of messages where the payloads are converted according to the generic type. The following example shows how to do so:

@KafkaListener(topics = "blc3", groupId = "blc3")
public void listen1(List<Message<Foo>> fooMessages) {
    ...
}
ConversionService Customization

Starting with version 2.1.1, the org.springframework.core.convert.ConversionService used by the default o.s.messaging.handler.annotation.support.MessageHandlerMethodFactory to resolve parameters for the invocation of a listener method is supplied with all beans that implement any of the following interfaces:

  • org.springframework.core.convert.converter.Converter

  • org.springframework.core.convert.converter.GenericConverter

  • org.springframework.format.Formatter

This lets you further customize listener deserialization without changing the default configuration for ConsumerFactory and KafkaListenerContainerFactory.

Setting a custom MessageHandlerMethodFactory on the KafkaListenerEndpointRegistrar through a KafkaListenerConfigurer bean disables this feature.
Adding custom HandlerMethodArgumentResolver to @KafkaListener

Starting with version 2.4.2 you are able to add your own HandlerMethodArgumentResolver and resolve custom method parameters. All you need is to implement KafkaListenerConfigurer and use method setCustomMethodArgumentResolvers() from class KafkaListenerEndpointRegistrar.

@Configuration
class CustomKafkaConfig implements KafkaListenerConfigurer {

    @Override
    public void configureKafkaListeners(KafkaListenerEndpointRegistrar registrar) {
        registrar.setCustomMethodArgumentResolvers(
            new HandlerMethodArgumentResolver() {

                @Override
                public boolean supportsParameter(MethodParameter parameter) {
                    return CustomMethodArgument.class.isAssignableFrom(parameter.getParameterType());
                }

                @Override
                public Object resolveArgument(MethodParameter parameter, Message<?> message) {
                    return new CustomMethodArgument(
                        message.getHeaders().get(KafkaHeaders.RECEIVED_TOPIC, String.class)
                    );
                }
            }
        );
    }

}

4.1.17. Message Headers

The 0.11.0.0 client introduced support for headers in messages. As of version 2.0, Spring for Apache Kafka now supports mapping these headers to and from spring-messaging MessageHeaders.

Previous versions mapped ConsumerRecord and ProducerRecord to spring-messaging Message<?>, where the value property is mapped to and from the payload and other properties (topic, partition, and so on) were mapped to headers. This is still the case, but additional (arbitrary) headers can now be mapped.

Apache Kafka headers have a simple API, shown in the following interface definition:

public interface Header {

    String key();

    byte[] value();

}

The KafkaHeaderMapper strategy is provided to map header entries between Kafka Headers and MessageHeaders. Its interface definition is as follows:

public interface KafkaHeaderMapper {

    void fromHeaders(MessageHeaders headers, Headers target);

    void toHeaders(Headers source, Map<String, Object> target);

}

The DefaultKafkaHeaderMapper maps the key to the MessageHeaders header name and, in order to support rich header types for outbound messages, JSON conversion is performed. A “special” header (with a key of spring_json_header_types) contains a JSON map of <key>:<type>. This header is used on the inbound side to provide appropriate conversion of each header value to the original type.

On the inbound side, all Kafka Header instances are mapped to MessageHeaders. On the outbound side, by default, all MessageHeaders are mapped, except id, timestamp, and the headers that map to ConsumerRecord properties.

You can specify which headers are to be mapped for outbound messages, by providing patterns to the mapper. The following listing shows a number of example mappings:

public DefaultKafkaHeaderMapper() { (1)
    ...
}

public DefaultKafkaHeaderMapper(ObjectMapper objectMapper) { (2)
    ...
}

public DefaultKafkaHeaderMapper(String... patterns) { (3)
    ...
}

public DefaultKafkaHeaderMapper(ObjectMapper objectMapper, String... patterns) { (4)
    ...
}
1 Uses a default Jackson ObjectMapper and maps most headers, as discussed before the example.
2 Uses the provided Jackson ObjectMapper and maps most headers, as discussed before the example.
3 Uses a default Jackson ObjectMapper and maps headers according to the provided patterns.
4 Uses the provided Jackson ObjectMapper and maps headers according to the provided patterns.

Patterns are rather simple and can contain a leading wildcard (), a trailing wildcard, or both (for example, .cat.*). You can negate patterns with a leading !. The first pattern that matches a header name (whether positive or negative) wins.

When you provide your own patterns, we recommend including !id and !timestamp, since these headers are read-only on the inbound side.

By default, the mapper deserializes only classes in java.lang and java.util. You can trust other (or all) packages by adding trusted packages with the addTrustedPackages method. If you receive messages from untrusted sources, you may wish to add only those packages you trust. To trust all packages, you can use mapper.addTrustedPackages("*").
Mapping String header values in a raw form is useful when communicating with systems that are not aware of the mapper’s JSON format.

Starting with version 2.2.5, you can specify that certain string-valued headers should not be mapped using JSON, but to/from a raw byte[]. The AbstractKafkaHeaderMapper has new properties; mapAllStringsOut when set to true, all string-valued headers will be converted to byte[] using the charset property (default UTF-8). In addition, there is a property rawMappedHeaders, which is a map of header name : boolean; if the map contains a header name, and the header contains a String value, it will be mapped as a raw byte[] using the charset. This map is also used to map raw incoming byte[] headers to String using the charset if, and only if, the boolean in the map value is true. If the boolean is false, or the header name is not in the map with a true value, the incoming header is simply mapped as the raw unmapped header.

The following test case illustrates this mechanism.

@Test
public void testSpecificStringConvert() {
    DefaultKafkaHeaderMapper mapper = new DefaultKafkaHeaderMapper();
    Map<String, Boolean> rawMappedHeaders = new HashMap<>();
    rawMappedHeaders.put("thisOnesAString", true);
    rawMappedHeaders.put("thisOnesBytes", false);
    mapper.setRawMappedHeaders(rawMappedHeaders);
    Map<String, Object> headersMap = new HashMap<>();
    headersMap.put("thisOnesAString", "thing1");
    headersMap.put("thisOnesBytes", "thing2");
    headersMap.put("alwaysRaw", "thing3".getBytes());
    MessageHeaders headers = new MessageHeaders(headersMap);
    Headers target = new RecordHeaders();
    mapper.fromHeaders(headers, target);
    assertThat(target).containsExactlyInAnyOrder(
            new RecordHeader("thisOnesAString", "thing1".getBytes()),
            new RecordHeader("thisOnesBytes", "thing2".getBytes()),
            new RecordHeader("alwaysRaw", "thing3".getBytes()));
    headersMap.clear();
    mapper.toHeaders(target, headersMap);
    assertThat(headersMap).contains(
            entry("thisOnesAString", "thing1"),
            entry("thisOnesBytes", "thing2".getBytes()),
            entry("alwaysRaw", "thing3".getBytes()));
}

By default, the DefaultKafkaHeaderMapper is used in the MessagingMessageConverter and BatchMessagingMessageConverter, as long as Jackson is on the class path.

With the batch converter, the converted headers are available in the KafkaHeaders.BATCH_CONVERTED_HEADERS as a List<Map<String, Object>> where the map in a position of the list corresponds to the data position in the payload.

If there is no converter (either because Jackson is not present or it is explicitly set to null), the headers from the consumer record are provided unconverted in the KafkaHeaders.NATIVE_HEADERS header. This header is a Headers object (or a List<Headers> in the case of the batch converter), where the position in the list corresponds to the data position in the payload).

Certain types are not suitable for JSON serialization, and a simple toString() serialization might be preferred for these types. The DefaultKafkaHeaderMapper has a method called addToStringClasses() that lets you supply the names of classes that should be treated this way for outbound mapping. During inbound mapping, they are mapped as String. By default, only org.springframework.util.MimeType and org.springframework.http.MediaType are mapped this way.
Starting with version 2.3, handling of String-valued headers is simplified. Such headers are no longer JSON encoded, by default (i.e. they do not have enclosing "…​" added). The type is still added to the JSON_TYPES header so the receiving system can convert back to a String (from byte[]). The mapper can handle (decode) headers produced by older versions (it checks for a leading "); in this way an application using 2.3 can consume records from older versions.
To be compatible with earlier versions, set encodeStrings to true, if records produced by a version using 2.3 might be consumed by applications using earlier versions. When all applications are using 2.3 or higher, you can leave the property at its default value of false.
@Bean
MessagingMessageConverter converter() {
    MessagingMessageConverter converter = new MessagingMessageConverter();
    DefaultKafkaHeaderMapper mapper = new DefaultKafkaHeaderMapper();
    mapper.setEncodeStrings(true);
    converter.setHeaderMapper(mapper);
    return converter;
}

If using Spring Boot, it will auto configure this converter bean into the auto-configured KafkaTemplate; otherwise you should add this converter to the template.

4.1.18. Null Payloads and Log Compaction of 'Tombstone' Records

When you use Log Compaction, you can send and receive messages with null payloads to identify the deletion of a key.

You can also receive null values for other reasons, such as a Deserializer that might return null when it cannot deserialize a value.

To send a null payload by using the KafkaTemplate, you can pass null into the value argument of the send() methods. One exception to this is the send(Message<?> message) variant. Since spring-messaging Message<?> cannot have a null payload, you can use a special payload type called KafkaNull, and the framework sends null. For convenience, the static KafkaNull.INSTANCE is provided.

When you use a message listener container, the received ConsumerRecord has a null value().

To configure the @KafkaListener to handle null payloads, you must use the @Payload annotation with required = false. If it is a tombstone message for a compacted log, you usually also need the key so that your application can determine which key was “deleted”. The following example shows such a configuration:

@KafkaListener(id = "deletableListener", topics = "myTopic")
public void listen(@Payload(required = false) String value, @Header(KafkaHeaders.RECEIVED_MESSAGE_KEY) String key) {
    // value == null represents key deletion
}

When you use a class-level @KafkaListener with multiple @KafkaHandler methods, some additional configuration is needed. Specifically, you need a @KafkaHandler method with a KafkaNull payload. The following example shows how to configure one:

@KafkaListener(id = "multi", topics = "myTopic")
static class MultiListenerBean {

    @KafkaHandler
    public void listen(String cat) {
        ...
    }

    @KafkaHandler
    public void listen(Integer hat) {
        ...
    }

    @KafkaHandler
    public void delete(@Payload(required = false) KafkaNull nul, @Header(KafkaHeaders.RECEIVED_MESSAGE_KEY) int key) {
        ...
    }

}

Note that the argument is null, not KafkaNull.

4.1.19. Handling Exceptions

This section describes how to handle various exceptions that may arise when you use Spring for Apache Kafka.

Listener Error Handlers

Starting with version 2.0, the @KafkaListener annotation has a new attribute: errorHandler.

You can use the errorHandler to provide the bean name of a KafkaListenerErrorHandler implementation. This functional interface has one method, as the following listing shows:

@FunctionalInterface
public interface KafkaListenerErrorHandler {

    Object handleError(Message<?> message, ListenerExecutionFailedException exception) throws Exception;

}

You have access to the spring-messaging Message<?> object produced by the message converter and the exception that was thrown by the listener, which is wrapped in a ListenerExecutionFailedException. The error handler can throw the original or a new exception, which is thrown to the container. Anything returned by the error handler is ignored.

It has a sub-interface (ConsumerAwareListenerErrorHandler) that has access to the consumer object, through the following method:

Object handleError(Message<?> message, ListenerExecutionFailedException exception, Consumer<?, ?> consumer);

If your error handler implements this interface, you can, for example, adjust the offsets accordingly. For example, to reset the offset to replay the failed message, you could do something like the following:

@Bean
public ConsumerAwareListenerErrorHandler listen3ErrorHandler() {
    return (m, e, c) -> {
        this.listen3Exception = e;
        MessageHeaders headers = m.getHeaders();
        c.seek(new org.apache.kafka.common.TopicPartition(
                headers.get(KafkaHeaders.RECEIVED_TOPIC, String.class),
                headers.get(KafkaHeaders.RECEIVED_PARTITION_ID, Integer.class)),
                headers.get(KafkaHeaders.OFFSET, Long.class));
        return null;
    };
}

Similarly, you could do something like the following for a batch listener:

@Bean
public ConsumerAwareListenerErrorHandler listen10ErrorHandler() {
    return (m, e, c) -> {
        this.listen10Exception = e;
        MessageHeaders headers = m.getHeaders();
        List<String> topics = headers.get(KafkaHeaders.RECEIVED_TOPIC, List.class);
        List<Integer> partitions = headers.get(KafkaHeaders.RECEIVED_PARTITION_ID, List.class);
        List<Long> offsets = headers.get(KafkaHeaders.OFFSET, List.class);
        Map<TopicPartition, Long> offsetsToReset = new HashMap<>();
        for (int i = 0; i < topics.size(); i++) {
            int index = i;
            offsetsToReset.compute(new TopicPartition(topics.get(i), partitions.get(i)),
                    (k, v) -> v == null ? offsets.get(index) : Math.min(v, offsets.get(index)));
        }
        offsetsToReset.forEach((k, v) -> c.seek(k, v));
        return null;
    };
}

This resets each topic/partition in the batch to the lowest offset in the batch.

The preceding two examples are simplistic implementations, and you would probably want more checking in the error handler.
Container Error Handlers

Two error handler interfaces (ErrorHandler and BatchErrorHandler) are provided. You must configure the appropriate type to match the message listener.

Starting with version 2.5, the default error handlers, when transactions are not being used, are the SeekToCurrentErrorHandler and RecoveringBatchErrorHandler with default configuration. See Seek To Current Container Error Handlers and Recovering Batch Error Handler. To restore the previous behavior, use the LoggingErrorHandler and BatchLoggingErrorHandler instead.

When transactions are being used, no error handlers are configured, by default, so that the exception will roll back the transaction. Error handling for transactional containers are handled by the AfterRollbackProcessor. If you provide a custom error handler when using transactions, it must throw an exception if you want the transaction rolled back.

Starting with version 2.3.2, these interfaces have a default method isAckAfterHandle() which is called by the container to determine whether the offset(s) should be committed if the error handler returns without throwing an exception. Starting with version 2.4, this returns true by default.

Typically, the error handlers provided by the framework will throw an exception when the error is not "handled" (e.g. after performing a seek operation). By default, such exceptions are logged by the container at ERROR level. Starting with version 2.5, all the framework error handlers extend KafkaExceptionLogLevelAware which allows you to control the level at which these exceptions are logged.

/**
 * Set the level at which the exception thrown by this handler is logged.
 * @param logLevel the level (default ERROR).
 */
public void setLogLevel(KafkaException.Level logLevel) {
    ...
}

You can specify a global error handler to be used for all listeners in the container factory. The following example shows how to do so:

@Bean
public KafkaListenerContainerFactory<ConcurrentMessageListenerContainer<Integer, String>>
        kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
            new ConcurrentKafkaListenerContainerFactory<>();
    ...
    factory.setErrorHandler(myErrorHandler);
    ...
    return factory;
}

Similarly, you can set a global batch error handler:

@Bean
public KafkaListenerContainerFactory<ConcurrentMessageListenerContainer<Integer, String>>
        kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<Integer, String> factory =
            new ConcurrentKafkaListenerContainerFactory<>();
    ...
    factory.setBatchErrorHandler(myBatchErrorHandler);
    ...
    return factory;
}

By default, if an annotated listener method throws an exception, it is thrown to the container, and the message is handled according to the container configuration.

If you are using Spring Boot, you simply need to add the error handler as a @Bean and boot will add it to the auto-configured factory.

Consumer-Aware Container Error Handlers

The container-level error handlers (ErrorHandler and BatchErrorHandler) have sub-interfaces called ConsumerAwareErrorHandler and ConsumerAwareBatchErrorHandler. The handle method of the ConsumerAwareErrorHandler has the following signature:

void handle(Exception thrownException, ConsumerRecord<?, ?> data, Consumer<?, ?> consumer);

The handle method of the ConsumerAwareBatchErrorHandler has the following signature:

void handle(Exception thrownException, ConsumerRecords<?, ?> data, Consumer<?, ?> consumer);

Similar to the @KafkaListener error handlers, you can reset the offsets as needed, based on the data that failed.

Unlike the listener-level error handlers, however, you should set the ackOnError container property to false (default) when making adjustments. Otherwise, any pending acks are applied after your repositioning.
Seek To Current Container Error Handlers

If an ErrorHandler implements RemainingRecordsErrorHandler, the error handler is provided with the failed record and any unprocessed records retrieved by the previous poll(). Those records are not passed to the listener after the handler exits. The following listing shows the RemainingRecordsErrorHandler interface definition:

@FunctionalInterface
public interface RemainingRecordsErrorHandler extends ConsumerAwareErrorHandler {

    void handle(Exception thrownException, List<ConsumerRecord<?, ?>> records, Consumer<?, ?> consumer);

}

This interface lets implementations seek all unprocessed topics and partitions so that the current record (and the others remaining) are retrieved by the next poll. The SeekToCurrentErrorHandler does exactly this.

ackOnError must be false (which is the default). Otherwise, if the container is stopped after the seek, but before the record is reprocessed, the record will be skipped when the container is restarted.

This is now the default error handler for record listeners.

The container commits any pending offset commits before calling the error handler.

To configure the listener container with this handler, add it to the container factory.

For example, with the @KafkaListener container factory, you can add SeekToCurrentErrorHandler as follows:

@Bean
public ConcurrentKafkaListenerContainerFactory<String, String> kafkaListenerContainerFactory() {
    ConcurrentKafkaListenerContainerFactory<String, String> factory = new ConcurrentKafkaListenerContainerFactory();
    factory.setConsumerFactory(consumerFactory());
    factory.getContainerProperties().setAckOnError(false);
    factory.getContainerProperties().setAckMode(AckMode.RECORD);
    factory.setErrorHandler(new SeekToCurrentErrorHandler(new FixedBackOff(1000L, 2L)));
    return factory;
}

This will retry a delivery up to 2 times (3 delivery attempts) with a back off of 1 second, instead of the default configuration (FixedBackOff(0L, 9)). Failures are simply logged after retries are exhausted.

As an example; if the poll returns six records (two from each partition 0, 1, 2) and the listener throws an exception on the fourth record, the container acknowledges the first three messages by committing their offsets. The SeekToCurrentErrorHandler seeks to offset 1 for partition 1 and offset 0 for partition 2. The next poll() returns the three unprocessed records.

If the AckMode was BATCH, the container commits the offsets for the first two partitions before calling the error handler.

Starting with version 2.2, the SeekToCurrentErrorHandler can now recover (skip) a record that keeps failing. By default, after ten failures, the failed record is logged (at the ERROR level). You can configure the handler with a custom recoverer (BiConsumer) and maximum failures. Using a FixedBackOff with FixedBackOff.UNLIMITED_ATTEMPTS causes (effectively) infinite retries. The following example configures recovery after three tries:

SeekToCurrentErrorHandler errorHandler =
    new SeekToCurrentErrorHandler((record, exception) -> {
        // recover after 3 failures, woth no back off - e.g. send to a dead-letter topic
    }, new FixedBackOff(0L, 2L));

Starting with version 2.2.4, when the container is configured with AckMode.MANUAL_IMMEDIATE, the error handler can be configured to commit the offset of recovered records; set the commitRecovered property to true.

When using transactions, similar functionality is provided by the DefaultAfterRollbackProcessor. See After-rollback Processor.

Starting with version 2.3, the SeekToCurrentErrorHandler considers certain exceptions to be fatal, and retries are skipped for such exceptions; the recoverer is invoked on the first failure. The exceptions that are considered fatal, by default, are:

  • DeserializationException

  • MessageConversionException

  • ConversionException

  • MethodArgumentResolutionException

  • NoSuchMethodException

  • ClassCastException

since these exceptions are unlikely to be resolved on a retried delivery.

You can add more exception types to the not-retryable category, or completely replace the map of classified exceptions. See the Javadocs for SeekToCurrentErrorHandler.setClassifications() for more information, as well as those for the spring-retry BinaryExceptionClassifier.

Here is an example that adds IllegalArgumentException to the not-retryable exceptions:

@Bean
public SeekToCurrentErrorHandler errorHandler(ConsumerRecordRecoverer recoverer) {
    SeekToCurrentErrorHandler handler = new SeekToCurrentErrorHandler(recoverer);
    handler.addNotRetryableException(IllegalArgumentException.class);
    return handler;
}

The SeekToCurrentBatchErrorHandler seeks each partition to the first record in each partition in the batch, so the whole batch is replayed. Also see Committing Offsets for an alternative. Also see Retrying Batch Error Handler. This error handler does not support recovery, because the framework cannot know which message in the batch is failing.

After seeking, an exception that wraps the ListenerExecutionFailedException is thrown. This is to cause the transaction to roll back (if transactions are enabled).

Starting with version 2.3, a BackOff can be provided to the SeekToCurrentErrorHandler and DefaultAfterRollbackProcessor so that the consumer thread can sleep for some configurable time between delivery attempts. Spring Framework provides two out of the box BackOff s, FixedBackOff and ExponentialBackOff. The maximum back off time must not exceed the max.poll.interval.ms consumer property, to avoid a rebalance.

Previously, the configuration was "maxFailures" (which included the first delivery attempt). When using a FixedBackOff, its maxAttempts properties represents the number of delivery retries (one less than the old maxFailures property). Also, maxFailures=-1 meant retry indefinitely with the old configuration, with a BackOff you would set the maxAttempts to Long.MAX_VALUE for a FixedBackOff and leave the maxElapsedTime to its default in an ExponentialBackOff.

The SeekToCurrentBatchErrorHandler can also be configured with a BackOff to add a delay between delivery attempts. Generally, you should configure the BackOff to never return STOP. However, since this error handler has no mechanism to "recover" after retries are exhausted, if the BackOffExecution returns STOP, the previous interval will be used for all subsequent delays. Again, the maximum delay must be less than the max.poll.interval.ms consumer property. Also see Retrying Batch Error Handler.

If the recoverer fails (throws an exception), the failed record will be included in the seeks. Starting with version 2.5.5, if the recoverer fails, the BackOff will be reset by default and redeliveries will again go through the back offs before recovery is attempted again. With earlier versions, the BackOff was not reset and recovery was re-attempted on the next failure. To revert to the previous behavior, set the error handler’s resetStateOnRecoveryFailure to false.

Starting with version 2.3.2, after a record has been recovered, its offset will be committed (if one of the container AckMode s is configured). To revert to the previous behavior, set the error handler’s ackAfterHandle property to false.

Starting with version 2.6, you can now provide the error handler with a BiFunction<ConsumerRecord<?, ?>, Exception, BackOff> to determine the BackOff to use, based on the failed record and/or the exception:

handler.setBackOffFunction((record, ex) -> { ... });

If the function returns null, the handler’s default BackOff will be used.

Starting with version 2.6.3, set resetStateOnExceptionChange to true and the retry sequence will be restarted (including the selection of a new BackOff, if so configured) if the exception type changes between failures. By default, the exception type is not considered.

Retrying Batch Error Handler

As discussed above, the SeekToCurrentBatchErrorHandler has no mechanism to recover after a certain number of failures. One reason for this is there is no guarantee that, when a batch is redelivered, the batch has the same number of records and/or the redelivered records are in the same order. It is impossible, therefore, to maintain retry state for a batch. The RetryingBatchErrorHandler takes a different approach. If a batch listener throws an exception, and this error handler is configured, the retries are performed from the in-memory batch of records. In order to avoid a rebalance during an extended retry sequence, the error handler pauses the consumer, polls it before sleeping for the back off, for each retry, and calls the listener again. If/when retries are exhausted, the ConsumerRecordRecoverer is called for each record in the batch. If the recoverer throws an exception, or the thread is interrupted during its sleep, a SeekToCurrentErrorHandler is invoked so that the batch of records will be redelivered on the next poll. Before exiting, regardless of the outcome, the consumer is resumed.

This error handler cannot be used with transactions.
Recovering Batch Error Handler

As an alternative to the Retrying Batch Error Handler, version 2.5 introduced the RecoveringBatchErrorHandler.

This is now the default error handler for batch listeners. The default configuration retries 9 times (10 delivery attempts) with no back off between deliveries.

This error handler works in conjunction with the listener throwing a BatchListenerFailedException providing the index in the batch where the failure occurred (or the failed record itself). If the listener throws a different exception, or the index is out of range, the error handler falls back to invoking a SeekToCurrentBatchErrorHandler and the whole batch is retried, with no recovery available. The sequence of events is:

  • Commit the offsets of the records before the index.

  • If retries are not exhausted, perform seeks so that all the remaining records (including the failed record) will be redelivered.

  • If retries are exhausted, attempt recovery of the failed record (default log only) and perform seeks so that the remaining records (excluding the failed record) will be redelivered. The recovered record’s offset is committed

  • If retries are exhausted and recovery fails, seeks are performed as if retries are not exhausted.

The default recoverer logs the failed record after retries are exhausted. You can use a custom recoverer, or one provided by the framework such as the DeadLetterPublishingRecoverer.

In all cases, a BackOff can be configured to enable a delay between delivery attempts.

Example:

@Bean
public RecoveringBatchErrorHandler(KafkaTemplate<String, String> template) {
    DeadLetterPublishingRecoverer recoverer =
            new DeadLetterPublishingRecoverer(template);
    RecoveringBatchErrorHandler errorHandler =
            new RecoveringBatchErrorHandler(recoverer, new FixedBackOff(2L, 5000));
}
@KafkaListener(id = "recovering", topics = "someTopic")
public void listen(List<ConsumerRecord<String, String>> records) {
    records.forEach(record -> {
        try {
            process(record);
        }
        catch (Exception e) {
            throw new BatchListenerFailedException("Failed to process", record);
        }
    });
}

For example; say 10 records are in the original batch and no more records are added to the topic during the retries, and the failed record is at index 4 in the list. After the first delivery fails, the offsets for the first 4 records will be committed; the remaing 6 will be redelivered after 5 seconds. Most likely (but not necessarily) the failed record will be at index 0 in the redelivery. If it fails again, it will be retried one more time and, if it again fails, it will be sent to a dead letter topic.

When using a POJO batch listener (e.g. List<Thing>), and you don’t have the full consumer record to add to the exception, you can just add the index of the record that failed:

@KafkaListener(id = "recovering", topics = "someTopic")
public void listen(List<Thing> things) {
    for (int i = 0; i < records.size(); i++) {
        try {
            process(things.get(i));
        }
        catch (Exception e) {
            throw new BatchListenerFailedException("Failed to process", i);
        }
    }
}
This error handler cannot be used with transactions
If the recoverer fails (throws an exception), the failed record will be included in the seeks. Starting with version 2.5.5, if the recoverer fails, the BackOff will be reset by default and redeliveries will again go through the back offs before recovery is attempted again. With earlier versions, the BackOff was not reset and recovery was re-attempted on the next failure. To revert to the previous behavior, set the error handler’s resetStateOnRecoveryFailure to false.

Starting with version 2.6, you can now provide the error handler with a BiFunction<ConsumerRecord<?, ?>, Exception, BackOff> to determine the BackOff to use, based on the failed record and/or the exception:

handler.setBackOffFunction((record, ex) -> { ... });

If the function returns null, the handler’s default BackOff will be used.

Starting with version 2.6.3, set resetStateOnExceptionChange to true and the retry sequence will be restarted (including the selection of a new BackOff, if so configured) if the exception type changes between failures. By default, the exception type is not considered.

Container Stopping Error Handlers

The ContainerStoppingErrorHandler (used with record listeners) stops the container if the listener throws an exception. When the AckMode is RECORD, offsets for already processed records are committed. When the AckMode is any manual value, offsets for already acknowledged records are committed. When the AckMode is BATCH, the entire batch is replayed when the container is restarted (unless transactions are enabled — in which case, only the unprocessed records are re-fetched).

The ContainerStoppingBatchErrorHandler (used with batch listeners) stops the container, and the entire batch is replayed when the container is restarted.

After the container stops, an exception that wraps the ListenerExecutionFailedException is thrown. This is to cause the transaction to roll back (if transactions are enabled).

After-rollback Processor

When using transactions, if the listener throws an exception (and an error handler, if present, throws an exception), the transaction is rolled back. By default, any unprocessed records (including the failed record) are re-fetched on the next poll. This is achieved by performing seek operations in the DefaultAfterRollbackProcessor. With a batch listener, the entire batch of records is reprocessed (the container has no knowledge of which record in the batch failed). To modify this behavior, you can configure the listener container with a custom AfterRollbackProcessor. For example, with a record-based listener, you might want to keep track of the failed record and give up after some number of attempts, perhaps by publishing it to a dead-letter topic.

Starting with version 2.2, the DefaultAfterRollbackProcessor can now recover (skip) a record that keeps failing. By default, after ten failures, the failed record is logged (at the ERROR level). You can configure the processor with a custom recoverer (BiConsumer) and maximum failures. Setting the maxFailures property to a negative number causes infinite retries. The following example configures recovery after three tries:

AfterRollbackProcessor<String, String> processor =
    new DefaultAfterRollbackProcessor((record, exception) -> {
        // recover after 3 failures, with no back off - e.g. send to a dead-letter topic
    }, new FixedBackOff(0L, 2L));

When you do not use transactions, you can achieve similar functionality by configuring a SeekToCurrentErrorHandler. See Seek To Current Container Error Handlers.

Recovery is not possible with a batch listener, since the framework has no knowledge about which record in the batch keeps failing. In such cases, the application listener must handle a record that keeps failing.

Starting with version 2.2.5, the DefaultAfterRollbackProcessor can be invoked in a new transaction (started after the failed transaction rolls back). Then, if you are using the DeadLetterPublishingRecoverer to publish a failed record, the processor will send the recovered record’s offset in the original topic/partition to the transaction. To enable this feature, set the commitRecovered and kafkaTemplate properties on the DefaultAfterRollbackProcessor.

If the recoverer fails (throws an exception), the failed record will be included in the seeks. Starting with version 2.5.5, if the recoverer fails, the BackOff will be reset by default and redeliveries will again go through the back offs before recovery is attempted again. With earlier versions, the BackOff was not reset and recovery was re-attempted on the next failure. To revert to the previous behavior, set the processor’s resetStateOnRecoveryFailure property to false.

Starting with version 2.6, you can now provide the processor with a BiFunction<ConsumerRecord<?, ?>, Exception, BackOff> to determine the BackOff to use, based on the failed record and/or the exception:

handler.setBackOffFunction((record, ex) -> { ... });

If the function returns null, the processor’s default BackOff will be used.

Starting with version 2.6.3, set resetStateOnExceptionChange to true and the retry sequence will be restarted (including the selection of a new BackOff, if so configured) if the exception type changes between failures. By default, the exception type is not considered.

Starting with version 2.3.1, similar to the SeekToCurrentErrorHandler, the DefaultAfterRollbackProcessor considers certain exceptions to be fatal, and retries are skipped for such exceptions; the recoverer is invoked on the first failure. The exceptions that are considered fatal, by default, are:

  • DeserializationException

  • MessageConversionException

  • ConversionException

  • MethodArgumentResolutionException

  • NoSuchMethodException

  • ClassCastException

since these exceptions are unlikely to be resolved on a retried delivery.

You can add more exception types to the not-retryable category, or completely replace the map of classified exceptions. See the Javadocs for SeekToCurrentErrorHandler.setClassifications() for more information, as well as those for the spring-retry BinaryExceptionClassifier.

Here is an example that adds IllegalArgumentException to the not-retryable exceptions:

@Bean
public DefaultAfterRollbackProcessor errorHandler(BiConsumer<ConsumerRecord<?, ?>, Exception> recoverer) {
    DefaultAfterRollbackProcessor processor = new DefaultAfterRollbackProcessor(recoverer);
    processor.addNotRetryableException(IllegalArgumentException.class);
    return processor;
}
With current kafka-clients, the container cannot detect whether a ProducerFencedException is caused by a rebalance or if the producer’s transactional.id has been revoked due to a timeout or expiry. Because, in most cases, it is caused by a rebalance, the container does not call the AfterRollbackProcessor (because it’s not appropriate to seek the partitions because we no longer are assigned them). If you ensure the timeout is large enough to process each transaction and periodically perform an "empty" transaction (e.g. via a ListenerContainerIdleEvent) you can avoid fencing due to timeout and expiry. Or, you can set the stopContainerWhenFenced container property to true and the container will stop, avoiding the loss of records. You can consume a ConsumerStoppedEvent and check the Reason property for FENCED to detect this condition. Since the event also has a reference to the container, you can restart the container using this event.
Delivery Attempts Header

The following applies to record listeners only, not batch listeners.

Starting with version 2.5, when using an ErrorHandler or AfterRollbackProcessor that implements DeliveryAttemptAware, it is possible to enable the addition of the KafkaHeaders.DELIVERY_ATTEMPT header (kafka_deliveryAttempt) to the record. The value of this header is an incrementing integer starting at 1. When receiving a raw ConsumerRecord<?, ?> the integer is in a byte[4].

int delivery = ByteBuffer.wrap(record.headers()
    .lastHeader(KafkaHeaders.DELIVERY_ATTEMPT).value())
    .getInt()

When using @KafkaListener with the DefaultKafkaHeaderMapper or SimpleKafkaHeaderMapper, it can be obtained by adding @Header(KafkaHeaders.DELIVERY_ATTEMPT) int delivery as a parameter to the listener method.

To enable population of this header, set the container property deliveryAttemptHeader to true. It is disabled by default to avoid the (small) overhead of looking up the state for each record and adding the header.

The SeekToCurrentErrorHandler and DefaultAfterRollbackProcessor support this feature.

Publishing Dead-letter Records

As discussed earlier, you can configure the SeekToCurrentErrorHandler and DefaultAfterRollbackProcessor (as well as the RecoveringBatchErrorHandler) with a record recoverer when the maximum number of failures is reached for a record. The framework provides the DeadLetterPublishingRecoverer, which publishes the failed message to another topic. The recoverer requires a KafkaTemplate<Object, Object>, which is used to send the record. You can also, optionally, configure it with a BiFunction<ConsumerRecord<?, ?>, Exception, TopicPartition>, which is called to resolve the destination topic and partition. By default, the dead-letter record is sent to a topic named <originalTopic>.DLT (the original topic name suffixed with .DLT) and to the same partition as the original record. Therefore, when you use the default resolver, the dead-letter topic must have at least as many partitions as the original topic. If the returned TopicPartition has a negative partition, the partition is not set in the ProducerRecord, so the partition is selected by Kafka. Starting with version 2.2.4, any ListenerExecutionFailedException (thrown, for example, when an exception is detected in a @KafkaListener method) is enhanced with the groupId property. This allows the destination resolver to use this, in addition to the information in the ConsumerRecord to select the dead letter topic.

The following example shows how to wire a custom destination resolver:

DeadLetterPublishingRecoverer recoverer = new DeadLetterPublishingRecoverer(template,
        (r, e) -> {
            if (e instanceof FooException) {
                return new TopicPartition(r.topic() + ".Foo.failures", r.partition());
            }
            else {
                return new TopicPartition(r.topic() + ".other.failures", r.partition());
            }
        });
ErrorHandler errorHandler = new SeekToCurrentErrorHandler(recoverer, new FixedBackOff(0L, 2L));

The record sent to the dead-letter topic is enhanced with the following headers:

  • KafkaHeaders.DLT_EXCEPTION_FQCN: The Exception class name.

  • KafkaHeaders.DLT_EXCEPTION_STACKTRACE: The Exception stack trace.

  • KafkaHeaders.DLT_EXCEPTION_MESSAGE: The Exception message.

  • KafkaHeaders.DLT_ORIGINAL_TOPIC: The original topic.

  • KafkaHeaders.DLT_ORIGINAL_PARTITION: The original partition.

  • KafkaHeaders.DLT_ORIGINAL_OFFSET: The original offset.

  • KafkaHeaders.DLT_ORIGINAL_TIMESTAMP: The original timestamp.

  • KafkaHeaders.DLT_ORIGINAL_TIMESTAMP_TYPE: The original timestamp type.

There are two mechanisms to add more headers.

  1. Subclass the recoverer and override createProducerRecord() - call super.createProducerRecord() and add more headers.

  2. Provide a BiFunction to receive the consumer record and exception, returning a Headers object; headers from there will be copied to the final producer record. Use setHeadersFunction() to set the BiFunction.

The second is simpler to implement but the first has more information available, including the already assembled standard headers.

Starting with version 2.3, when used in conjunction with an ErrorHandlingDeserializer, the publisher will restore the record value(), in the dead-letter producer record, to the original value that failed to be deserialized. Previously, the value() was null and user code had to decode the DeserializationException from the message headers. In addition, you can provide multiple KafkaTemplate s to the publisher; this might be needed, for example, if you want to publish the byte[] from a DeserializationException, as well as values using a different serializer from records that were deserialized successfully. Here is an example of configuring the publisher with KafkaTemplate s that use a String and byte[] serializer:

@Bean
public DeadLetterPublishingRecoverer publisher(KafkaTemplate<?, ?> stringTemplate,
        KafkaTemplate<?, ?> bytesTemplate) {

    Map<Class<?>, KafkaTemplate<?, ?>> templates = new LinkedHashMap<>();
    templates.put(String.class, stringTemplate);
    templates.put(byte[].class, bytesTemplate);
    return new DeadLetterPublishingRecoverer(templates);
}

The publisher uses the map keys to locate a template that is suitable for the value() about to be published. A LinkedHashMap is recommended so that the keys are examined in order.

When publishing null values, when there are multiple templates, the recoverer will look for a template for the Void class; if none is present, the first template from the values().iterator() will be used.

If the recoverer fails (throws an exception), the failed record will be included in the seeks. Starting with version 2.5.5, if the recoverer fails, the BackOff will be reset by default and redeliveries will again go through the back offs before recovery is attempted again. With earlier versions, the BackOff was not reset and recovery was re-attempted on the next failure. To revert to the previous behavior, set the error handler’s resetStateOnRecoveryFailure property to false.

Starting with version 2.6.3, set resetStateOnExceptionChange to true and the retry sequence will be restarted (including the selection of a new BackOff, if so configured) if the exception type changes between failures. By default, the exception type is not considered.

Starting with version 2.3, the recoverer can also be used with Kafka Streams - see Recovery from Deserialization Exceptions for more information.

The ErrorHandlingDeserializer adds the deserialization exception(s) in headers ErrorHandlingDeserializer.VALUE_DESERIALIZER_EXCEPTION_HEADER and ErrorHandlingDeserializer.KEY_DESERIALIZER_EXCEPTION_HEADER (using java serialization). By default, these headers are not retained in the message published to the dead letter topic, unless both the key and value fail deserialization. In that case, the DLT_* headers are based on the value deserialization and the key DeserializationException is retained in the header.

If incoming records are dependent on each other, but may arrive out of order, it may be useful to republish a failed record to the tail of the original topic (for some number of times), instead of sending it directly to the dead letter topic. See this Stack Overflow Question for an example.

The following error handler configuration will do exactly that:

@Bean
public ErrorHandler eh(KafkaOperations<String, String> template) {
    return new SeekToCurrentErrorHandler(new DeadLetterPublishingRecoverer(template,
            (rec, ex) -> {
                org.apache.kafka.common.header.Header retries = rec.headers().lastHeader("retries");
                if (retries == null) {
                    retries = new RecordHeader("retries", new byte[] { 1 });
                    rec.headers().add(retries);
                }
                else {
                    retries.value()[0]++;
                }
                return retries.value()[0] > 5
                        ? new TopicPartition("topic.DLT", rec.partition())
                        : new TopicPartition("topic", rec.partition());
            }), new FixedBackOff(0L, 0L));
}

4.1.20. JAAS and Kerberos

Starting with version 2.0, a KafkaJaasLoginModuleInitializer class has been added to assist with Kerberos configuration. You can add this bean, with the desired configuration, to your application context. The following example configures such a bean:

@Bean
public KafkaJaasLoginModuleInitializer jaasConfig() throws IOException {
    KafkaJaasLoginModuleInitializer jaasConfig = new KafkaJaasLoginModuleInitializer();
    jaasConfig.setControlFlag("REQUIRED");
    Map<String, String> options = new HashMap<>();
    options.put("useKeyTab", "true");
    options.put("storeKey", "true");
    options.put("keyTab", "/etc/security/keytabs/kafka_client.keytab");
    options.put("principal", "[email protected]");
    jaasConfig.setOptions(options);
    return jaasConfig;
}

4.2. Kafka Streams Support

Starting with version 1.1.4, Spring for Apache Kafka provides first-class support for Kafka Streams. To use it from a Spring application, the kafka-streams jar must be present on classpath. It is an optional dependency of the spring-kafka project and is not downloaded transitively.

4.2.1. Basics

The reference Apache Kafka Streams documentation suggests the following way of using the API:

// Use the builders to define the actual processing topology, e.g. to specify
// from which input topics to read, which stream operations (filter, map, etc.)
// should be called, and so on.

StreamsBuilder builder = ...;  // when using the Kafka Streams DSL

// Use the configuration to tell your application where the Kafka cluster is,
// which serializers/deserializers to use by default, to specify security settings,
// and so on.
StreamsConfig config = ...;

KafkaStreams streams = new KafkaStreams(builder, config);

// Start the Kafka Streams instance
streams.start();

// Stop the Kafka Streams instance
streams.close();

So, we have two main components:

  • StreamsBuilder: With an API to build KStream (or KTable) instances.

  • KafkaStreams: To manage the lifecycle of those instances.

All KStream instances exposed to a KafkaStreams instance by a single StreamsBuilder are started and stopped at the same time, even if they have different logic. In other words, all streams defined by a StreamsBuilder are tied with a single lifecycle control. Once a KafkaStreams instance has been closed by streams.close(), it cannot be restarted. Instead, a new KafkaStreams instance to restart stream processing must be created.

4.2.2. Spring Management

To simplify using Kafka Streams from the Spring application context perspective and use the lifecycle management through a container, the Spring for Apache Kafka introduces StreamsBuilderFactoryBean. This is an AbstractFactoryBean implementation to expose a StreamsBuilder singleton instance as a bean. The following example creates such a bean:

@Bean
public FactoryBean<StreamsBuilder> myKStreamBuilder(KafkaStreamsConfiguration streamsConfig) {
    return new StreamsBuilderFactoryBean(streamsConfig);
}
Starting with version 2.2, the stream configuration is now provided as a KafkaStreamsConfiguration object rather than a StreamsConfig.

The StreamsBuilderFactoryBean also implements SmartLifecycle to manage the lifecycle of an internal KafkaStreams instance. Similar to the Kafka Streams API, you must define the KStream instances before you start the KafkaStreams. That also applies for the Spring API for Kafka Streams. Therefore, when you use default autoStartup = true on the StreamsBuilderFactoryBean, you must declare KStream instances on the StreamsBuilder before the application context is refreshed. For example, KStream can be a regular bean definition, while the Kafka Streams API is used without any impacts. The following example shows how to do so:

@Bean
public KStream<?, ?> kStream(StreamsBuilder kStreamBuilder) {
    KStream<Integer, String> stream = kStreamBuilder.stream(STREAMING_TOPIC1);
    // Fluent KStream API
    return stream;
}

If you would like to control the lifecycle manually (for example, stopping and starting by some condition), you can reference the StreamsBuilderFactoryBean bean directly by using the factory bean (&) prefix. Since StreamsBuilderFactoryBean use its internal KafkaStreams instance, it is safe to stop and restart it again. A new KafkaStreams is created on each start(). You might also consider using different StreamsBuilderFactoryBean instances, if you would like to control the lifecycles for KStream instances separately.

You also can specify KafkaStreams.StateListener, Thread.UncaughtExceptionHandler, and StateRestoreListener options on the StreamsBuilderFactoryBean, which are delegated to the internal KafkaStreams instance. Also, apart from setting those options indirectly on StreamsBuilderFactoryBean, starting with version 2.1.5, you can use a KafkaStreamsCustomizer callback interface to configure an inner KafkaStreams instance. Note that KafkaStreamsCustomizer overrides the options provided by StreamsBuilderFactoryBean. If you need to perform some KafkaStreams operations directly, you can access that internal KafkaStreams instance by using StreamsBuilderFactoryBean.getKafkaStreams(). You can autowire StreamsBuilderFactoryBean bean by type, but you should be sure to use the full type in the bean definition, as the following example shows:

@Bean
public StreamsBuilderFactoryBean myKStreamBuilder(KafkaStreamsConfiguration streamsConfig) {
    return new StreamsBuilderFactoryBean(streamsConfig);
}
...
@Autowired
private StreamsBuilderFactoryBean myKStreamBuilderFactoryBean;

Alternatively, you can add @Qualifier for injection by name if you use interface bean definition. The following example shows how to do so:

@Bean
public FactoryBean<StreamsBuilder> myKStreamBuilder(KafkaStreamsConfiguration streamsConfig) {
    return new StreamsBuilderFactoryBean(streamsConfig);
}
...
@Autowired
@Qualifier("&myKStreamBuilder")
private StreamsBuilderFactoryBean myKStreamBuilderFactoryBean;

Starting with version 2.4.1, the factory bean has a new property infrastructureCustomizer with type KafkaStreamsInfrastructureCustomizer; this allows customization of the StreamsBuilder (e.g. to add a state store) and/or the Topology before the stream is created.

public interface KafkaStreamsInfrastructureCustomizer {

	void configureBuilder(StreamsBuilder builder);

	void configureTopology(Topology topology);

}

Default no-op implementations are provided to avoid having to implement both methods if one is not required.

A CompositeKafkaStreamsInfrastructureCustomizer is provided, for when you need to apply multiple customizers.

4.2.3. KafkaStreams Micrometer Support

Introduced in version 2.5.3, you can configure a KafkaStreamsMicrometerListener to automatically register micrometer meters for the KafkaStreams object managed by the factory bean:

streamsBuilderFactoryBean.setListener(new KafkaStreamsMicrometerListener(meterRegistry,
        Collections.singletonList(new ImmutableTag("customTag", "customTagValue"))));

4.2.4. Streams JSON Serialization and Deserialization

For serializing and deserializing data when reading or writing to topics or state stores in JSON format, Spring Kafka provides a JsonSerde implementation that uses JSON, delegating to the JsonSerializer and JsonDeserializer described in Serialization, Deserialization, and Message Conversion. The JsonSerde implementation provides the same configuration options through its constructor (target type or ObjectMapper). In the following example, we use the JsonSerde to serialize and deserialize the Cat payload of a Kafka stream (the JsonSerde can be used in a similar fashion wherever an instance is required):

stream.through(Serdes.Integer(), new JsonSerde<>(Cat.class), "cats");

When constructing the serializer/deserializer programmatically for use in the producer/consumer factory, since version 2.3, you can use the fluent API, which simplifies configuration.

stream.through(new JsonSerde<>(MyKeyType.class)
        .forKeys()
        .noTypeInfo(),
    new JsonSerde<>(MyValueType.class)
        .noTypeInfo(),
    "myTypes");

4.2.5. Using KafkaStreamBrancher

The KafkaStreamBrancher class introduces a more convenient way to build conditional branches on top of KStream.

Consider the following example that does not use KafkaStreamBrancher:

KStream<String, String>[] branches = builder.stream("source").branch(
      (key, value) -> value.contains("A"),
      (key, value) -> value.contains("B"),
      (key, value) -> true
     );
branches[0].to("A");
branches[1].to("B");
branches[2].to("C");

The following example uses KafkaStreamBrancher:

new KafkaStreamBrancher<String, String>()
   .branch((key, value) -> value.contains("A"), ks -> ks.to("A"))
   .branch((key, value) -> value.contains("B"), ks -> ks.to("B"))
   //default branch should not necessarily be defined in the end of the chain!
   .defaultBranch(ks -> ks.to("C"))
   .onTopOf(builder.stream("source"));
   //onTopOf method returns the provided stream so we can continue with method chaining

4.2.6. Configuration

To configure the Kafka Streams environment, the StreamsBuilderFactoryBean requires a KafkaStreamsConfiguration instance. See the Apache Kafka documentation for all possible options.

Starting with version 2.2, the stream configuration is now provided as a KafkaStreamsConfiguration object, rather than as a StreamsConfig.

To avoid boilerplate code for most cases, especially when you develop microservices, Spring for Apache Kafka provides the @EnableKafkaStreams annotation, which you should place on a @Configuration class. All you need is to declare a KafkaStreamsConfiguration bean named defaultKafkaStreamsConfig. A StreamsBuilderFactoryBean bean, named defaultKafkaStreamsBuilder, is automatically declared in the application context. You can declare and use any additional StreamsBuilderFactoryBean beans as well. Starting with version 2.3, you can perform additional customization of that bean, by providing a bean that implements StreamsBuilderFactoryBeanCustomizer. There must only be one such bean, or one must be marked @Primary.

By default, when the factory bean is stopped, the KafkaStreams.cleanUp() method is called. Starting with version 2.1.2, the factory bean has additional constructors, taking a CleanupConfig object that has properties to let you control whether the cleanUp() method is called during start() or stop() or neither.

4.2.7. Header Enricher

Version 2.3 added the HeaderEnricher implementation of Transformer. This can be used to add headers within the stream processing; the header values are SpEL expressions; the root object of the expression evaluation has 3 properties:

  • context - the ProcessorContext, allowing access to the current record metadata

  • key - the key of the current record

  • value - the value of the current record

The expressions must return a byte[] or a String (which will be converted to byte[] using UTF-8).

To use the enricher within a stream:

.transform(() -> enricher)

The transformer does not change the key or value; it simply adds headers.

4.2.8. MessagingTransformer

Version 2.3 added the MessagingTransformer this allows a Kafka Streams topology to interact with a Spring Messaging component, such as a Spring Integration flow. The transformer requires an implementation of MessagingFunction.

@FunctionalInterface
public interface MessagingFunction {

	Message<?> exchange(Message<?> message);

}

Spring Integration automatically provides an implementation using its GatewayProxyFactoryBean. It also requires a MessagingMessageConverter to convert the key, value and metadata (including headers) to/from a Spring Messaging Message<?>. See [streams-integration] for more information.

4.2.9. Recovery from Deserialization Exceptions

Version 2.3 introduced the RecoveringDeserializationExceptionHandler which can take some action when a deserialization exception occurs. Refer to the Kafka documentation about DeserializationExceptionHandler, of which the RecoveringDeserializationExceptionHandler is an implementation. The RecoveringDeserializationExceptionHandler is configured with a ConsumerRecordRecoverer implementation. The framework provides the DeadLetterPublishingRecoverer which sends the failed record to a dead-letter topic. See Publishing Dead-letter Records for more information about this recoverer.

To configure the recoverer, add the following properties to your streams configuration:

@Bean(name = KafkaStreamsDefaultConfiguration.DEFAULT_STREAMS_CONFIG_BEAN_NAME)
public KafkaStreamsConfiguration kStreamsConfigs() {
    Map<String, Object> props = new HashMap<>();
    ...
    props.put(StreamsConfig.DEFAULT_DESERIALIZATION_EXCEPTION_HANDLER_CLASS_CONFIG,
            RecoveringDeserializationExceptionHandler.class);
    props.put(RecoveringDeserializationExceptionHandler.KSTREAM_DESERIALIZATION_RECOVERER, recoverer());
    ...
    return new KafkaStreamsConfiguration(props);
}

@Bean
public DeadLetterPublishingRecoverer recoverer() {
    return new DeadLetterPublishingRecoverer(kafkaTemplate(),
            (record, ex) -> new TopicPartition("recovererDLQ", -1));
}

Of course, the recoverer() bean can be your own implementation of ConsumerRecordRecoverer.

4.2.10. Kafka Streams Example

The following example combines all the topics we have covered in this chapter:

@Configuration
@EnableKafka
@EnableKafkaStreams
public static class KafkaStreamsConfig {

    @Bean(name = KafkaStreamsDefaultConfiguration.DEFAULT_STREAMS_CONFIG_BEAN_NAME)
    public KafkaStreamsConfiguration kStreamsConfigs() {
        Map<String, Object> props = new HashMap<>();
        props.put(StreamsConfig.APPLICATION_ID_CONFIG, "testStreams");
        props.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, "localhost:9092");
        props.put(StreamsConfig.DEFAULT_KEY_SERDE_CLASS_CONFIG, Serdes.Integer().getClass().getName());
        props.put(StreamsConfig.DEFAULT_VALUE_SERDE_CLASS_CONFIG, Serdes.String().getClass().getName());
        props.put(StreamsConfig.DEFAULT_TIMESTAMP_EXTRACTOR_CLASS_CONFIG, WallclockTimestampExtractor.class.getName());
        return new KafkaStreamsConfiguration(props);
    }

    @Bean
    public StreamsBuilderFactoryBeanCustomizer customizer() {
        return fb -> fb.setStateListener((newState, oldState) -> {
            System.out.println("State transition from " + oldState + " to " + newState);
        });
    }

    @Bean
    public KStream<Integer, String> kStream(StreamsBuilder kStreamBuilder) {
        KStream<Integer, String> stream = kStreamBuilder.stream("streamingTopic1");
        stream
                .mapValues((ValueMapper<String, String>) String::toUpperCase)
                .groupByKey()
                .windowedBy(TimeWindows.of(Duration.ofMillis(1000)))
                .reduce((String value1, String value2) -> value1 + value2,
                		Named.as("windowStore"))
                .toStream()
                .map((windowedId, value) -> new KeyValue<>(windowedId.key(), value))
                .filter((i, s) -> s.length() > 40)
                .to("streamingTopic2");

        stream.print(Printed.toSysOut());

        return stream;
    }

}

4.3. Testing Applications

The spring-kafka-test jar contains some useful utilities to assist with testing your applications.

4.3.1. JUnit

o.s.kafka.test.utils.KafkaTestUtils provides some static methods to set up producer and consumer properties. The following listing shows those method signatures:

/**
 * Set up test properties for an {@code <Integer, String>} consumer.
 * @param group the group id.
 * @param autoCommit the auto commit.
 * @param embeddedKafka a {@link EmbeddedKafkaBroker} instance.
 * @return the properties.
 */
public static Map<String, Object> consumerProps(String group, String autoCommit,
                                       EmbeddedKafkaBroker embeddedKafka) { ... }

/**
 * Set up test properties for an {@code <Integer, String>} producer.
 * @param embeddedKafka a {@link EmbeddedKafkaBroker} instance.
 * @return the properties.
 */
public static Map<String, Object> producerProps(EmbeddedKafkaBroker embeddedKafka) { ... }
Starting with version 2.5, the consumerProps method sets the ConsumerConfig.AUTO_OFFSET_RESET_CONFIG to earliest. This is because, in most cases, you want the consumer to consume any messages sent in a test case. The ConsumerConfig default is latest which means that messages already sent by a test, before the consumer starts, will not receive those records. To revert to the previous behavior, set the property to latest after calling the method.

A JUnit 4 @Rule wrapper for the EmbeddedKafkaBroker is provided to create an embedded Kafka and an embedded Zookeeper server. (See @EmbeddedKafka Annotation for information about using @EmbeddedKafka with JUnit 5). The following listing shows the signatures of those methods:

/**
 * Create embedded Kafka brokers.
 * @param count the number of brokers.
 * @param controlledShutdown passed into TestUtils.createBrokerConfig.
 * @param topics the topics to create (2 partitions per).
 */
public EmbeddedKafkaRule(int count, boolean controlledShutdown, String... topics) { ... }

/**
 *
 * Create embedded Kafka brokers.
 * @param count the number of brokers.
 * @param controlledShutdown passed into TestUtils.createBrokerConfig.
 * @param partitions partitions per topic.
 * @param topics the topics to create.
 */
public EmbeddedKafkaRule(int count, boolean controlledShutdown, int partitions, String... topics) { ... }

The EmbeddedKafkaBroker class has a utility method that lets you consume for all the topics it created. The following example shows how to use it:

Map<String, Object> consumerProps = KafkaTestUtils.consumerProps("testT", "false", embeddedKafka);
DefaultKafkaConsumerFactory<Integer, String> cf = new DefaultKafkaConsumerFactory<Integer, String>(
        consumerProps);
Consumer<Integer, String> consumer = cf.createConsumer();
embeddedKafka.consumeFromAllEmbeddedTopics(consumer);

The KafkaTestUtils has some utility methods to fetch results from the consumer. The following listing shows those method signatures:

/**
 * Poll the consumer, expecting a single record for the specified topic.
 * @param consumer the consumer.
 * @param topic the topic.
 * @return the record.
 * @throws org.junit.ComparisonFailure if exactly one record is not received.
 */
public static <K, V> ConsumerRecord<K, V> getSingleRecord(Consumer<K, V> consumer, String topic) { ... }

/**
 * Poll the consumer for records.
 * @param consumer the consumer.
 * @return the records.
 */
public static <K, V> ConsumerRecords<K, V> getRecords(Consumer<K, V> consumer) { ... }

The following example shows how to use KafkaTestUtils:

...
template.sendDefault(0, 2, "bar");
ConsumerRecord<Integer, String> received = KafkaTestUtils.getSingleRecord(consumer, "topic");
...

When the embedded Kafka and embedded Zookeeper server are started by the EmbeddedKafkaBroker, a system property named spring.embedded.kafka.brokers is set to the address of the Kafka brokers and a system property named spring.embedded.zookeeper.connect is set to the address of Zookeeper. Convenient constants (EmbeddedKafkaBroker.SPRING_EMBEDDED_KAFKA_BROKERS and EmbeddedKafkaBroker.SPRING_EMBEDDED_ZOOKEEPER_CONNECT) are provided for this property.

With the EmbeddedKafkaBroker.brokerProperties(Map<String, String>), you can provide additional properties for the Kafka servers. See Kafka Config for more information about possible broker properties.

4.3.2. Configuring Topics

The following example configuration creates topics called cat and hat with five partitions, a topic called thing1 with 10 partitions, and a topic called thing2 with 15 partitions:

public class MyTests {

    @ClassRule
    private static EmbeddedKafkaRule embeddedKafka = new EmbeddedKafkaRule(1, false, 5, "cat", "hat");

    @Test
    public void test() {
        embeddedKafkaRule.getEmbeddedKafka()
              .addTopics(new NewTopic("thing1", 10, (short) 1), new NewTopic("thing2", 15, (short) 1));
        ...
      }

}

By default, addTopics will throw an exception when problems arise (such as adding a topic that already exists). Version 2.6 added a new version of that method that returns a Map<String, Exception>; the key is the topic name and the value is null for success, or an Exception for a failure.

4.3.3. Using the Same Brokers for Multiple Test Classes

There is no built-in support for doing so, but you can use the same broker for multiple test classes with something similar to the following:

public final class EmbeddedKafkaHolder {

    private static EmbeddedKafkaRule embeddedKafka = new EmbeddedKafkaRule(1, false);

    private static boolean started;

    public static EmbeddedKafkaRule getEmbeddedKafka() {
        if (!started) {
            try {
                embeddedKafka.before();
            }
            catch (Exception e) {
                throw new KafkaException(e);
            }
            started = true;
        }
        return embeddedKafka;
    }

    private EmbeddedKafkaHolder() {
        super();
    }

}

Then, in each test class, you can use something similar to the following:

static {
    EmbeddedKafkaHolder.getEmbeddedKafka().addTopics(topic1, topic2);
}

private static EmbeddedKafkaRule embeddedKafka = EmbeddedKafkaHolder.getEmbeddedKafka();
The preceding example provides no mechanism for shutting down the brokers when all tests are complete. This could be a problem if, say, you run your tests in a Gradle daemon. You should not use this technique in such a situation, or you should use something to call destroy() on the EmbeddedKafkaBroker when your tests are complete.

4.3.4. @EmbeddedKafka Annotation

We generally recommend that you use the rule as a @ClassRule to avoid starting and stopping the broker between tests (and use a different topic for each test). Starting with version 2.0, if you use Spring’s test application context caching, you can also declare a EmbeddedKafkaBroker bean, so a single broker can be used across multiple test classes. For convenience, we provide a test class-level annotation called @EmbeddedKafka to register the EmbeddedKafkaBroker bean. The following example shows how to use it:

@RunWith(SpringRunner.class)
@DirtiesContext
@EmbeddedKafka(partitions = 1,
         topics = {
                 KafkaStreamsTests.STREAMING_TOPIC1,
                 KafkaStreamsTests.STREAMING_TOPIC2 })
public class KafkaStreamsTests {

    @Autowired
    private EmbeddedKafkaBroker embeddedKafka;

    @Test
    public void someTest() {
        Map<String, Object> consumerProps = KafkaTestUtils.consumerProps("testGroup", "true", this.embeddedKafka);
        consumerProps.put(ConsumerConfig.AUTO_OFFSET_RESET_CONFIG, "earliest");
        ConsumerFactory<Integer, String> cf = new DefaultKafkaConsumerFactory<>(consumerProps);
        Consumer<Integer, String> consumer = cf.createConsumer();
        this.embeddedKafka.consumeFromAnEmbeddedTopic(consumer, KafkaStreamsTests.STREAMING_TOPIC2);
        ConsumerRecords<Integer, String> replies = KafkaTestUtils.getRecords(consumer);
        assertThat(replies.count()).isGreaterThanOrEqualTo(1);
    }

    @Configuration
    @EnableKafkaStreams
    public static class KafkaStreamsConfiguration {

        @Value("${" + EmbeddedKafkaBroker.SPRING_EMBEDDED_KAFKA_BROKERS + "}")
        private String brokerAddresses;

        @Bean(name = KafkaStreamsDefaultConfiguration.DEFAULT_STREAMS_CONFIG_BEAN_NAME)
        public KafkaStreamsConfiguration kStreamsConfigs() {
            Map<String, Object> props = new HashMap<>();
            props.put(StreamsConfig.APPLICATION_ID_CONFIG, "testStreams");
            props.put(StreamsConfig.BOOTSTRAP_SERVERS_CONFIG, this.brokerAddresses);
            return new KafkaStreamsConfiguration(props);
        }

    }

}

Starting with version 2.2.4, you can also use the @EmbeddedKafka annotation to specify the Kafka ports property.

The following example sets the topics, brokerProperties, and brokerPropertiesLocation attributes of @EmbeddedKafka support property placeholder resolutions:

@TestPropertySource(locations = "classpath:/test.properties")
@EmbeddedKafka(topics = { "any-topic", "${kafka.topics.another-topic}" },
        brokerProperties = { "log.dir=${kafka.broker.logs-dir}",
                            "listeners=PLAINTEXT://localhost:${kafka.broker.port}",
                            "auto.create.topics.enable=${kafka.broker.topics-enable:true}" },
        brokerPropertiesLocation = "classpath:/broker.properties")

In the preceding example, the property placeholders ${kafka.topics.another-topic}, ${kafka.broker.logs-dir}, and ${kafka.broker.port} are resolved from the Spring Environment. In addition, the broker properties are loaded from the broker.properties classpath resource specified by the brokerPropertiesLocation. Property placeholders are resolved for the brokerPropertiesLocation URL and for any property placeholders found in the resource. Properties defined by brokerProperties override properties found in brokerPropertiesLocation.

You can use the @EmbeddedKafka annotation with JUnit 4 or JUnit 5.

4.3.5. @EmbeddedKafka Annotation with JUnit5

Starting with version 2.3, there are two ways to use the @EmbeddedKafka annotation with JUnit5. When used with the @SpringJunitConfig annotation, the embedded broker is added to the test application context. You can auto wire the broker into your test, at the class or method level, to get the broker address list.

When not using the spring test context, the EmbdeddedKafkaCondition creates a broker; the condition includes a parameter resolver so you can access the broker in your test method…​

@EmbeddedKafka
public class EmbeddedKafkaConditionTests {

	@Test
	public void test(EmbeddedKafkaBroker broker) {
		String brokerList = broker.getBrokersAsString();
        ...
	}

}

A stand-alone (not Spring test context) broker will be created if the class annotated with @EmbeddedBroker is not also annotated (or meta annotated) with ExtendedWith(SpringExtension.class). @SpringJunitConfig and @SpringBootTest are so meta annotated and the context-based broker will be used when either of those annotations are also present.

When there is a Spring test application context available, the topics and broker properties can contain property placeholders, which will be resolved as long as the property is defined somewhere. If there is no Spring context available, these placeholders won’t be resolved.

4.3.6. Embedded Broker in @SpringBootTest Annotations

Spring Initializr now automatically adds the spring-kafka-test dependency in test scope to the project configuration.

If your application uses the Kafka binder in spring-cloud-stream and if you want to use an embedded broker for tests, you must remove the spring-cloud-stream-test-support dependency, because it replaces the real binder with a test binder for test cases. If you wish some tests to use the test binder and some to use the embedded broker, tests that use the real binder need to disable the test binder by excluding the binder auto configuration in the test class. The following example shows how to do so:

@RunWith(SpringRunner.class)
@SpringBootTest(properties = "spring.autoconfigure.exclude="
    + "org.springframework.cloud.stream.test.binder.TestSupportBinderAutoConfiguration")
public class MyApplicationTests {
    ...
}

There are several ways to use an embedded broker in a Spring Boot application test.

They include:

JUnit4 Class Rule

The following example shows how to use a JUnit4 class rule to create an embedded broker:

@RunWith(SpringRunner.class)
@SpringBootTest
public class MyApplicationTests {

    @ClassRule
    public static EmbeddedKafkaRule broker = new EmbeddedKafkaRule(1,
        false, "someTopic")
            .brokerListProperty("spring.kafka.bootstrap-servers");
    }

    @Autowired
    private KafkaTemplate<String, String> template;

    @Test
    public void test() {
        ...
    }

}

Notice that, since this is a Spring Boot application, we override the broker list property to set Boot’s property.

@EmbeddedKafka Annotation or EmbeddedKafkaBroker Bean

The following example shows how to use an @EmbeddedKafka Annotation to create an embedded broker:

@RunWith(SpringRunner.class)
@EmbeddedKafka(topics = "someTopic",
        bootstrapServersProperty = "spring.kafka.bootstrap-servers")
public class MyApplicationTests {

    @Autowired
    private KafkaTemplate<String, String> template;

    @Test
    public void test() {
        ...
    }

}

4.3.7. Hamcrest Matchers

The o.s.kafka.test.hamcrest.KafkaMatchers provides the following matchers:

/**
 * @param key the key
 * @param <K> the type.
 * @return a Matcher that matches the key in a consumer record.
 */
public static <K> Matcher<ConsumerRecord<K, ?>> hasKey(K key) { ... }

/**
 * @param value the value.
 * @param <V> the type.
 * @return a Matcher that matches the value in a consumer record.
 */
public static <V> Matcher<ConsumerRecord<?, V>> hasValue(V value) { ... }

/**
 * @param partition the partition.
 * @return a Matcher that matches the partition in a consumer record.
 */
public static Matcher<ConsumerRecord<?, ?>> hasPartition(int partition) { ... }

/**
 * Matcher testing the timestamp of a {@link ConsumerRecord} assuming the topic has been set with
 * {@link org.apache.kafka.common.record.TimestampType#CREATE_TIME CreateTime}.
 *
 * @param ts timestamp of the consumer record.
 * @return a Matcher that matches the timestamp in a consumer record.
 */
public static Matcher<ConsumerRecord<?, ?>> hasTimestamp(long ts) {
  return hasTimestamp(TimestampType.CREATE_TIME, ts);
}

/**
 * Matcher testing the timestamp of a {@link ConsumerRecord}
 * @param type timestamp type of the record
 * @param ts timestamp of the consumer record.
 * @return a Matcher that matches the timestamp in a consumer record.
 */
public static Matcher<ConsumerRecord<?, ?>> hasTimestamp(TimestampType type, long ts) {
  return new ConsumerRecordTimestampMatcher(type, ts);
}

4.3.8. AssertJ Conditions

You can use the following AssertJ conditions:

/**
 * @param key the key
 * @param <K> the type.
 * @return a Condition that matches the key in a consumer record.
 */
public static <K> Condition<ConsumerRecord<K, ?>> key(K key) { ... }

/**
 * @param value the value.
 * @param <V> the type.
 * @return a Condition that matches the value in a consumer record.
 */
public static <V> Condition<ConsumerRecord<?, V>> value(V value) { ... }

/**
 * @param key the key.
 * @param value the value.
 * @param <K> the key type.
 * @param <V> the value type.
 * @return a Condition that matches the key in a consumer record.
 * @since 2.2.12
 */
public static <K, V> Condition<ConsumerRecord<K, V>> keyValue(K key, V value) { ... }

/**
 * @param partition the partition.
 * @return a Condition that matches the partition in a consumer record.
 */
public static Condition<ConsumerRecord<?, ?>> partition(int partition) { ... }

/**
 * @param value the timestamp.
 * @return a Condition that matches the timestamp value in a consumer record.
 */
public static Condition<ConsumerRecord<?, ?>> timestamp(long value) {
  return new ConsumerRecordTimestampCondition(TimestampType.CREATE_TIME, value);
}

/**
 * @param type the type of timestamp
 * @param value the timestamp.
 * @return a Condition that matches the timestamp value in a consumer record.
 */
public static Condition<ConsumerRecord<?, ?>> timestamp(TimestampType type, long value) {
  return new ConsumerRecordTimestampCondition(type, value);
}

4.3.9. Example

The following example brings together most of the topics covered in this chapter:

public class KafkaTemplateTests {

    private static final String TEMPLATE_TOPIC = "templateTopic";

    @ClassRule
    public static EmbeddedKafkaRule embeddedKafka = new EmbeddedKafkaRule(1, true, TEMPLATE_TOPIC);

    @Test
    public void testTemplate() throws Exception {
        Map<String, Object> consumerProps = KafkaTestUtils.consumerProps("testT", "false",
            embeddedKafka.getEmbeddedKafka());
        DefaultKafkaConsumerFactory<Integer, String> cf =
                            new DefaultKafkaConsumerFactory<Integer, String>(consumerProps);
        ContainerProperties containerProperties = new ContainerProperties(TEMPLATE_TOPIC);
        KafkaMessageListenerContainer<Integer, String> container =
                            new KafkaMessageListenerContainer<>(cf, containerProperties);
        final BlockingQueue<ConsumerRecord<Integer, String>> records = new LinkedBlockingQueue<>();
        container.setupMessageListener(new MessageListener<Integer, String>() {

            @Override
            public void onMessage(ConsumerRecord<Integer, String> record) {
                System.out.println(record);
                records.add(record);
            }

        });
        container.setBeanName("templateTests");
        container.start();
        ContainerTestUtils.waitForAssignment(container,
                            embeddedKafka.getEmbeddedKafka().getPartitionsPerTopic());
        Map<String, Object> producerProps =
                            KafkaTestUtils.producerProps(embeddedKafka.getEmbeddedKafka());
        ProducerFactory<Integer, String> pf =
                            new DefaultKafkaProducerFactory<Integer, String>(producerProps);
        KafkaTemplate<Integer, String> template = new KafkaTemplate<>(pf);
        template.setDefaultTopic(TEMPLATE_TOPIC);
        template.sendDefault("foo");
        assertThat(records.poll(10, TimeUnit.SECONDS), hasValue("foo"));
        template.sendDefault(0, 2, "bar");
        ConsumerRecord<Integer, String> received = records.poll(10, TimeUnit.SECONDS);
        assertThat(received, hasKey(2));
        assertThat(received, hasPartition(0));
        assertThat(received, hasValue("bar"));
        template.send(TEMPLATE_TOPIC, 0, 2, "baz");
        received = records.poll(10, TimeUnit.SECONDS);
        assertThat(received, hasKey(2));
        assertThat(received, hasPartition(0));
        assertThat(received, hasValue("baz"));
    }

}

The preceding example uses the Hamcrest matchers. With AssertJ, the final part looks like the following code:

assertThat(records.poll(10, TimeUnit.SECONDS)).has(value("foo"));
template.sendDefault(0, 2, "bar");
ConsumerRecord<Integer, String> received = records.poll(10, TimeUnit.SECONDS);
// using individual assertions
assertThat(received).has(key(2));
assertThat(received).has(value("bar"));
assertThat(received).has(partition(0));
template.send(TEMPLATE_TOPIC, 0, 2, "baz");
received = records.poll(10, TimeUnit.SECONDS);
// using allOf()
assertThat(received).has(allOf(keyValue(2, "baz"), partition(0)));

5. Tips, Tricks and Examples

5.1. Manually Assigning All Partitions

Let’s say you want to always read all records from all partitions (such as when using a compacted topic to load a distributed cache), it can be useful to manually assign the partitions and not use Kafka’s group management. Doing so can be unwieldy when there are many partitions, because you have to list the partitions. It’s also an issue if the number of partitions changes over time, because you would have to recompile your application each time the partition count changes.

The following is an example of how to use the power of a SpEL expression to create the partition list dynamically when the application starts:

@KafkaListener(topicPartitions = @TopicPartition(topic = "compacted",
            partitions = "#{@finder.partitions('compacted')}"),
            partitionOffsets = @PartitionOffset(partition = "*", initialOffset = "0")))
public void listen(@Header(KafkaHeaders.RECEIVED_MESSAGE_KEY) String key, String payload) {
    ...
}

@Bean
public PartitionFinder finder(ConsumerFactory<String, String> consumerFactory) {
    return new PartitionFinder(consumerFactory);
}

public static class PartitionFinder {

    private final ConsumerFactory<String, String> consumerFactory;

    public PartitionFinder(ConsumerFactory<String, String> consumerFactory) {
        this.consumerFactory = consumerFactory;
    }

    public String[] partitions(String topic) {
        try (Consumer<String, String> consumer = consumerFactory.createConsumer()) {
            return consumer.partitionsFor(topic).stream()
                .map(pi -> "" + pi.partition())
                .toArray(String[]::new);
        }
    }

}

Using this in conjunction with ConsumerConfig.AUTO_OFFSET_RESET_CONFIG=earliest will load all records each time the application is started. You should also set the container’s AckMode to MANUAL to prevent the container from committing offsets for a null consumer group. Howewever, starting with version 2.5.5, as shown above, you can apply an initial offset to all partitions; see Explicit Partition Assignment for more information.

5.2. Example of Transactions with ChainedKafkaTransactionManager

The following Spring Boot application is an example of synchronizing database and Kafka transactions.

@SpringBootApplication
public class Application {

    public static void main(String[] args) {
        SpringApplication.run(Application.class, args);
    }

    @Bean
    public ApplicationRunner runner(KafkaTemplate<String, String> template) {
        return args -> template.executeInTransaction(t -> t.send("topic1", "test"));
    }

    @Bean
    public ChainedKafkaTransactionManager<Object, Object> chainedTm(
            KafkaTransactionManager<String, String> ktm,
            DataSourceTransactionManager dstm) {

        return new ChainedKafkaTransactionManager<>(ktm, dstm);
    }

    @Bean
    public DataSourceTransactionManager dstm(DataSource dataSource) {
        return new DataSourceTransactionManager(dataSource);
    }

    @Bean
    public ConcurrentKafkaListenerContainerFactory<?, ?> kafkaListenerContainerFactory(
            ConcurrentKafkaListenerContainerFactoryConfigurer configurer,
            ConsumerFactory<Object, Object> kafkaConsumerFactory,
            ChainedKafkaTransactionManager<Object, Object> chainedTM) {

        ConcurrentKafkaListenerContainerFactory<Object, Object> factory =
                new ConcurrentKafkaListenerContainerFactory<>();
        configurer.configure(factory, kafkaConsumerFactory);
        factory.getContainerProperties().setTransactionManager(chainedTM);
        return factory;
    }

    @Component
    public static class Listener {

        private final JdbcTemplate jdbcTemplate;

        private final KafkaTemplate<String, String> kafkaTemplate;

        public Listener(JdbcTemplate jdbcTemplate, KafkaTemplate<String, String> kafkaTemplate) {
            this.jdbcTemplate = jdbcTemplate;
            this.kafkaTemplate = kafkaTemplate;
        }

        @KafkaListener(id = "group1", topics = "topic1")
        public void listen1(String in) {
            this.kafkaTemplate.send("topic2", in.toUpperCase());
            this.jdbcTemplate.execute("insert into mytable (data) values ('" + in + "')");
        }

        @KafkaListener(id = "group2", topics = "topic2")
        public void listen2(String in) {
            System.out.println(in);
        }

    }

    @Bean
    public NewTopic topic1() {
        return TopicBuilder.name("topic1").build();
    }

    @Bean
    public NewTopic topic2() {
        return TopicBuilder.name("topic2").build();
    }

}
spring.datasource.url=jdbc:mysql://localhost/integration?serverTimezone=UTC
spring.datasource.username=root
spring.datasource.driver-class-name=com.mysql.cj.jdbc.Driver

spring.kafka.consumer.auto-offset-reset=earliest
spring.kafka.consumer.enable-auto-commit=false
spring.kafka.consumer.properties.isolation.level=read_committed

spring.kafka.producer.transaction-id-prefix=tx-

#logging.level.org.springframework.transaction=trace
#logging.level.org.springframework.kafka.transaction=debug
#logging.level.org.springframework.jdbc=debug
create table mytable (data varchar(20));

6. Other Resources

In addition to this reference documentation, we recommend a number of other resources that may help you learn about Spring and Apache Kafka.

Appendix A: Override Spring Boot Dependencies

When using Spring for Apache Kafka in a Spring Boot application, the Kafka dependency versions are determined by Spring Boot’s dependency management. If you wish to use a different version, you need to override all of the associated dependencies. This is especially true when using the embedded Kafka broker in spring-kafka-test.

Backwards compatibility is not supported for all Boot versions; Spring for Apache Kafka 2.6 has been tested with Spring Boot 2.4 and 2.3.

maven

<dependency>
    <groupId>org.springframework.kafka</groupId>
    <artifactId>spring-kafka</artifactId>
    <version>2.6.3</version>
</dependency>

<dependency>
    <groupId>org.springframework.kafka</groupId>
    <artifactId>spring-kafka-test</artifactId>
    <version>2.6.3</version>
    <scope>test</scope>
</dependency>

<dependency>
    <groupId>org.apache.kafka</groupId>
    <artifactId>kafka-clients</artifactId>
    <version>{kafka-version}</version>
</dependency>

<!-- optional - only needed when using kafka-streams -->
<dependency>
    <groupId>org.apache.kafka</groupId>
    <artifactId>kafka-streams</artifactId>
    <version>{kafka-version}</version>
</dependency>

<dependency>
    <groupId>org.apache.kafka</groupId>
    <artifactId>kafka-clients</artifactId>
    <version>{kafka-version}</version>
    <classifier>test</classifier>
    <scope>test</scope>
</dependency>

<dependency>
    <groupId>org.apache.kafka</groupId>
    <artifactId>kafka_2.13</artifactId>
    <version>{kafka-version}</version>
    <scope>test</scope>
</dependency>

<dependency>
    <groupId>org.apache.kafka</groupId>
    <artifactId>kafka_2.13</artifactId>
    <version>{kafka-version}</version>
    <classifier>test</classifier>
    <scope>test</scope>
</dependency>

gradle

dependencies {

    implementation 'org.springframework.kafka:spring-kafka:2.6.3'

    implementation "org.apache.kafka:kafka-clients:$kafkaVersion"
    implementation "org.apache.kafka:kafka-streams:$kafkaVersion" // optional - only needed when using kafka-streams
    testImplementation 'org.springframework.kafka:spring-kafka-test:2.6.3'
    testImplementation "org.apache.kafka:kafka-clients:$kafkaVersion:test"
    testImplementation "org.apache.kafka:kafka_2.13:$kafkaVersion"
    testImplementation "org.apache.kafka:kafka_2.13:$kafkaVersion:test"

}

The test scope dependencies are only needed if you are using the embedded Kafka broker in tests.

Appendix B: Change History

B.1. Changes between 2.4 and 2.5

This section covers the changes made from version 2.4 to version 2.5. For changes in earlier version, see Change History.

Also see [new-in-sik].

B.1.1. Consumer/Producer Factory Changes

The default consumer and producer factories can now invoke a callback whenever a consumer or producer is created or closed. Implementations for native Micrometer metrics are provided. See Factory Listeners for more information.

You can now change bootstrap server properties at runtime, enabling failover to another Kafka cluster. See Connecting to Kafka for more information.

B.1.2. StreamsBuilderFactoryBean Changes

The factory bean can now invoke a callback whenever a KafkaStreams created or destroyed. An Implementation for native Micrometer metrics is provided. See KafkaStreams Micrometer Support for more information.

B.1.3. Kafka Client Version

This version requires the 2.5.0 kafka-clients.

B.1.4. Class/Package Changes

SeekUtils has been moved from the o.s.k.support package to o.s.k.listener.

B.1.5. Delivery Attempts Header

There is now an option to to add a header which tracks delivery attempts when using certain error handlers and after rollback processors. See Delivery Attempts Header for more information.

B.1.6. @KafkaListener Changes

Default reply headers will now be populated automatically if needed when a @KafkaListener return type is Message<?>. See Reply Type Message<?> for more information.

The KafkaHeaders.RECEIVED_MESSAGE_KEY is no longer populated with a null value when the incoming record has a null key; the header is omitted altogether.

@KafkaListener methods can now specify a ConsumerRecordMetadata parameter instead of using discrete headers for metadata such as topic, partition, etc. See Consumer Record Metadata for more information.

B.1.7. Listener Container Changes

The assignmentCommitOption container property is now LATEST_ONLY_NO_TX by default. See Listener Container Properties for more information.

The subBatchPerPartition container property is now true by default when using transactions. See Transactions for more information.

A new RecoveringBatchErrorHandler is now provided. See Recovering Batch Error Handler for more information.

Static group membership is now supported. See Message Listener Containers for more information.

When incremental/cooperative rebalancing is configured, if offsets fail to commit with a non-fatal RebalanceInProgressException, the container will attempt to re-commit the offsets for the partitions that remain assigned to this instance after the rebalance is completed.

The default error handler is now the SeekToCurrentErrorHandler for record listeners and RecoveringBatchErrorHandler for batch listeners. See Container Error Handlers for more information.

You can now control the level at which exceptions intentionally thrown by standard error handlers are logged. See Container Error Handlers for more information.

The getAssignmentsByClientId() method has been added, making it easier to determine which consumers in a concurrent container are assigned which partition(s). See Listener Container Properties for more information.

You can now suppress logging entire ConsumerRecord s in error, debug logs etc. See onlyLogRecordMetadata in Listener Container Properties.

B.1.8. KafkaTemplate Changes

The KafkaTemplate can now maintain micrometer timers. See Monitoring for more information.

The KafkaTemplate can now be configured with ProducerConfig properties to override those in the producer factory. See Using KafkaTemplate for more information.

A RoutingKafkaTemplate has now been provided. See Using RoutingKafkaTemplate for more information.

You can now use KafkaSendCallback instead of ListenerFutureCallback to get a narrower exception, making it easier to extract the failed ProducerRecord. See Using KafkaTemplate for more information.

B.1.9. Kafka String Serializer/Deserializer

New ToStringSerializer/StringDeserializer s as well as an associated SerDe are now provided. See String serialization for more information.

B.1.10. JsonDeserializer

The JsonDeserializer now has more flexibility to determine the deserialization type. See Using Methods to Determine Types for more information.

B.1.11. Delegating Serializer/Deserializer

The DelegatingSerializer can now handle "standard" types, when the outbound record has no header. See Delegating Serializer and Deserializer for more information.

B.1.12. Testing Changes

The KafkaTestUtils.consumerProps() helper record now sets ConsumerConfig.AUTO_OFFSET_RESET_CONFIG to earliest by default. See JUnit for more information.

B.2. Changes between 2.3 and 2.4

B.2.1. Kafka Client Version

This version requires the 2.4.0 kafka-clients or higher and supports the new incremental rebalancing feature.

B.2.2. ConsumerAwareRebalanceListener

Like ConsumerRebalanceListener, this interface now has an additional method onPartitionsLost. Refer to the Apache Kafka documentation for more information.

Unlike the ConsumerRebalanceListener, The default implementation does not call onPartitionsRevoked. Instead, the listener container will call that method after it has called onPartitionsLost; you should not, therefore, do the same when implementing ConsumerAwareRebalanceListener.

See the IMPORTANT note at the end of Rebalancing Listeners for more information.

B.2.3. GenericErrorHandler

The isAckAfterHandle() default implementation now returns true by default.

B.2.4. KafkaTemplate

The KafkaTemplate now supports non-transactional publishing alongside transactional. See KafkaTemplate Transactional and non-Transactional Publishing for more information.

B.2.5. AggregatingReplyingKafkaTemplate

The releaseStrategy is now a BiConsumer. It is now called after a timeout (as well as when records arrive); the second parameter is true in the case of a call after a timeout.

See Aggregating Multiple Replies for more information.

B.2.6. Listener Container

The ContainerProperties provides an authorizationExceptionRetryInterval option to let the listener container to retry after any AuthorizationException is thrown by the KafkaConsumer. See its JavaDocs and Using KafkaMessageListenerContainer for more information.

B.2.7. @KafkaListener

The @KafkaListener annotation has a new property splitIterables; default true. When a replying listener returns an Iterable this property controls whether the return result is sent as a single record or a record for each element is sent. See Forwarding Listener Results using @SendTo for more information

Batch listeners can now be configured with a BatchToRecordAdapter; this allows, for example, the batch to be processed in a transaction while the listener gets one record at a time. With the default implementation, a ConsumerRecordRecoverer can be used to handle errors within the batch, without stopping the processing of the entire batch - this might be useful when using transactions. See Transactions with Batch Listeners for more information.

B.2.8. Kafka Streams

The StreamsBuilderFactoryBean accepts a new property KafkaStreamsInfrastructureCustomizer. This allows configuration of the builder and/or topology before the stream is created. See Spring Management for more information.

B.3. Changes Between 2.2 and 2.3

This section covers the changes made from version 2.2 to version 2.3.

Also see [new-in-sik].

B.3.1. Tips, Tricks and Examples

A new chapter Tips, Tricks and Examples has been added. Please submit GitHub issues and/or pull requests for additional entries in that chapter.

B.3.2. Kafka Client Version

This version requires the 2.3.0 kafka-clients or higher.

B.3.3. Class/Package Changes

TopicPartitionInitialOffset is deprecated in favor of TopicPartitionOffset.

B.3.4. Configuration Changes

Starting with version 2.3.4, the missingTopicsFatal container property is false by default. When this is true, the application fails to start if the broker is down; many users were affected by this change; given that Kafka is a high-availability platform, we did not anticipate that starting an application with no active brokers would be a common use case.

B.3.5. Producer and Consumer Factory Changes

The DefaultKafkaProducerFactory can now be configured to create a producer per thread. You can also provide Supplier<Serializer> instances in the constructor as an alternative to either configured classes (which require no-arg constructors), or constructing with Serializer instances, which are then shared between all Producers. See Using DefaultKafkaProducerFactory for more information.

The same option is available with Supplier<Deserializer> instances in DefaultKafkaConsumerFactory. See Using KafkaMessageListenerContainer for more information.

B.3.6. Listener Container Changes

Previously, error handlers received ListenerExecutionFailedException (with the actual listener exception as the cause) when the listener was invoked using a listener adapter (such as @KafkaListener s). Exceptions thrown by native GenericMessageListener s were passed to the error handler unchanged. Now a ListenerExecutionFailedException is always the argument (with the actual listener exception as the cause), which provides access to the container’s group.id property.

Because the listener container has it’s own mechanism for committing offsets, it prefers the Kafka ConsumerConfig.ENABLE_AUTO_COMMIT_CONFIG to be false. It now sets it to false automatically unless specifically set in the consumer factory or the container’s consumer property overrides.

The ackOnError property is now false by default. See Seek To Current Container Error Handlers for more information.

It is now possible to obtain the consumer’s group.id property in the listener method. See Obtaining the Consumer group.id for more information.

The container has a new property recordInterceptor allowing records to be inspected or modified before invoking the listener. A CompositeRecordInterceptor is also provided in case you need to invoke multiple interceptors. See Message Listener Containers for more information.

The ConsumerSeekAware has new methods allowing you to perform seeks relative to the beginning, end, or current position and to seek to the first offset greater than or equal to a time stamp. See Seeking to a Specific Offset for more information.

A convenience class AbstractConsumerSeekAware is now provided to simplify seeking. See Seeking to a Specific Offset for more information.

The ContainerProperties provides an idleBetweenPolls option to let the main loop in the listener container to sleep between KafkaConsumer.poll() calls. See its JavaDocs and Using KafkaMessageListenerContainer for more information.

When using AckMode.MANUAL (or MANUAL_IMMEDIATE) you can now cause a redelivery by calling nack on the Acknowledgment. See Committing Offsets for more information.

Listener performance can now be monitored using Micrometer Timer s. See Monitoring for more information.

The containers now publish additional consumer lifecycle events relating to startup. See Application Events for more information.

Transactional batch listeners can now support zombie fencing. See Transactions for more information.

The listener container factory can now be configured with a ContainerCustomizer to further configure each container after it has been created and configured. See Container factory for more information.

B.3.7. ErrorHandler Changes

The SeekToCurrentErrorHandler now treats certain exceptions as fatal and disables retry for those, invoking the recoverer on first failure.

The SeekToCurrentErrorHandler and SeekToCurrentBatchErrorHandler can now be configured to apply a BackOff (thread sleep) between delivery attempts.

Starting with version 2.3.2, recovered records' offsets will be committed when the error handler returns after recovering a failed record.

See Seek To Current Container Error Handlers for more information.

The DeadLetterPublishingRecoverer, when used in conjunction with an ErrorHandlingDeserializer, now sets the payload of the message sent to the dead-letter topic, to the original value that could not be deserialized. Previously, it was null and user code needed to extract the DeserializationException from the message headers. See Publishing Dead-letter Records for more information.

B.3.8. TopicBuilder

A new class TopicBuilder is provided for more convenient creation of NewTopic @Bean s for automatic topic provisioning. See Configuring Topics for more information.

B.3.9. Kafka Streams Changes

You can now perform additional configuration of the StreamsBuilderFactoryBean created by @EnableKafkaStreams. See Streams Configuration for more information.

A RecoveringDeserializationExceptionHandler is now provided which allows records with deserialization errors to be recovered. It can be used in conjunction with a DeadLetterPublishingRecoverer to send these records to a dead-letter topic. See Recovery from Deserialization Exceptions for more information.

The HeaderEnricher transformer has been provided, using SpEL to generate the header values. See Header Enricher for more information.

The MessagingTransformer has been provided. This allows a Kafka streams topology to interact with a spring-messaging component, such as a Spring Integration flow. See MessagingTransformer and [streams-integration] for more information.

B.3.10. JSON Component Changes

Now all the JSON-aware components are configured by default with a Jackson ObjectMapper produced by the JacksonUtils.enhancedObjectMapper(). The JsonDeserializer now provides TypeReference-based constructors for better handling of target generic container types. Also a JacksonMimeTypeModule has been introduced for serialization of org.springframework.util.MimeType to plain string. See its JavaDocs and Serialization, Deserialization, and Message Conversion for more information.

A ByteArrayJsonMessageConverter has been provided as well as a new super class for all Json converters, JsonMessageConverter. Also, a StringOrBytesSerializer is now available; it can serialize byte[], Bytes and String values in ProducerRecord s. See Spring Messaging Message Conversion for more information.

The JsonSerializer, JsonDeserializer and JsonSerde now have fluent APIs to make programmatic configuration simpler. See the javadocs, Serialization, Deserialization, and Message Conversion, and Streams JSON Serialization and Deserialization for more informaion.

B.3.11. ReplyingKafkaTemplate

When a reply times out, the future is completed exceptionally with a KafkaReplyTimeoutException instead of a KafkaException.

Also, an overloaded sendAndReceive method is now provided that allows specifying the reply timeout on a per message basis.

B.3.12. AggregatingReplyingKafkaTemplate

Extends the ReplyingKafkaTemplate by aggregating replies from multiple receivers. See Aggregating Multiple Replies for more information.

B.3.13. Transaction Changes

You can now override the producer factory’s transactionIdPrefix on the KafkaTemplate and KafkaTransactionManager. See transactionIdPrefix for more information.

B.3.14. New Delegating Serializer/Deserializer

The framework now provides a delegating Serializer and Deserializer, utilizing a header to enable producing and consuming records with multiple key/value types. See Delegating Serializer and Deserializer for more information.

B.3.15. New Retrying Deserializer

The framework now provides a delegating RetryingDeserializer, to retry serialization when transient errors such as network problems might occur. See Retrying Deserializer for more information.

B.4. Changes Between 2.1 and 2.2

B.4.1. Kafka Client Version

This version requires the 2.0.0 kafka-clients or higher.

B.4.2. Class and Package Changes

The ContainerProperties class has been moved from org.springframework.kafka.listener.config to org.springframework.kafka.listener.

The AckMode enum has been moved from AbstractMessageListenerContainer to ContainerProperties.

The setBatchErrorHandler() and setErrorHandler() methods have been moved from ContainerProperties to both AbstractMessageListenerContainer and AbstractKafkaListenerContainerFactory.

B.4.3. After Rollback Processing

A new AfterRollbackProcessor strategy is provided. See After-rollback Processor for more information.

B.4.4. ConcurrentKafkaListenerContainerFactory Changes

You can now use the ConcurrentKafkaListenerContainerFactory to create and configure any ConcurrentMessageListenerContainer, not only those for @KafkaListener annotations. See Container factory for more information.

B.4.5. Listener Container Changes

A new container property (missingTopicsFatal) has been added. See Using KafkaMessageListenerContainer for more information.

A ConsumerStoppedEvent is now emitted when a consumer stops. See Thread Safety for more information.

Batch listeners can optionally receive the complete ConsumerRecords<?, ?> object instead of a List<ConsumerRecord<?, ?>. See Batch listeners for more information.

The DefaultAfterRollbackProcessor and SeekToCurrentErrorHandler can now recover (skip) records that keep failing, and, by default, does so after 10 failures. They can be configured to publish failed records to a dead-letter topic.

Starting with version 2.2.4, the consumer’s group ID can be used while selecting the dead letter topic name.

The ConsumerStoppingEvent has been added. See Application Events for more information.

The SeekToCurrentErrorHandler can now be configured to commit the offset of a recovered record when the container is configured with AckMode.MANUAL_IMMEDIATE (since 2.2.4). See Seek To Current Container Error Handlers for more information.

B.4.6. @KafkaListener Changes

You can now override the concurrency and autoStartup properties of the listener container factory by setting properties on the annotation. You can now add configuration to determine which headers (if any) are copied to a reply message. See @KafkaListener Annotation for more information.

You can now use @KafkaListener as a meta-annotation on your own annotations. See @KafkaListener as a Meta Annotation for more information.

It is now easier to configure a Validator for @Payload validation. See @KafkaListener @Payload Validation for more information.

You can now specify kafka consumer properties directly on the annotation; these will override any properties with the same name defined in the consumer factory (since version 2.2.4). See Annotation Properties for more information.

B.4.7. Header Mapping Changes

Headers of type MimeType and MediaType are now mapped as simple strings in the RecordHeader value. Previously, they were mapped as JSON and only MimeType was decoded. MediaType could not be decoded. They are now simple strings for interoperability.

Also, the DefaultKafkaHeaderMapper has a new addToStringClasses method, allowing the specification of types that should be mapped by using toString() instead of JSON. See Message Headers for more information.

B.4.8. Embedded Kafka Changes

The KafkaEmbedded class and its KafkaRule interface have been deprecated in favor of the EmbeddedKafkaBroker and its JUnit 4 EmbeddedKafkaRule wrapper. The @EmbeddedKafka annotation now populates an EmbeddedKafkaBroker bean instead of the deprecated KafkaEmbedded. This change allows the use of @EmbeddedKafka in JUnit 5 tests. The @EmbeddedKafka annotation now has the attribute ports to specify the port that populates the EmbeddedKafkaBroker. See Testing Applications for more information.

B.4.9. JsonSerializer/Deserializer Enhancements

You can now provide type mapping information by using producer and consumer properties.

New constructors are available on the deserializer to allow overriding the type header information with the supplied target type.

The JsonDeserializer now removes any type information headers by default.

You can now configure the JsonDeserializer to ignore type information headers by using a Kafka property (since 2.2.3).

B.4.10. Kafka Streams Changes

The streams configuration bean must now be a KafkaStreamsConfiguration object instead of a StreamsConfig object.

The StreamsBuilderFactoryBean has been moved from package …​core to …​config.

The KafkaStreamBrancher has been introduced for better end-user experience when conditional branches are built on top of KStream instance.

See Kafka Streams Support and Configuration for more information.

B.4.11. Transactional ID

When a transaction is started by the listener container, the transactional.id is now the transactionIdPrefix appended with <group.id>.<topic>.<partition>. This change allows proper fencing of zombies, as described here.

B.5. Changes Between 2.0 and 2.1

B.5.1. Kafka Client Version

This version requires the 1.0.0 kafka-clients or higher.

The 1.1.x client is supported with version 2.1.5, but you need to override dependencies as described in [deps-for-11x].

The 1.1.x client is supported natively in version 2.2.

B.5.2. JSON Improvements

The StringJsonMessageConverter and JsonSerializer now add type information in Headers, letting the converter and JsonDeserializer create specific types on reception, based on the message itself rather than a fixed configured type. See Serialization, Deserialization, and Message Conversion for more information.

B.5.3. Container Stopping Error Handlers

Container error handlers are now provided for both record and batch listeners that treat any exceptions thrown by the listener as fatal/ They stop the container. See Handling Exceptions for more information.

B.5.4. Pausing and Resuming Containers

The listener containers now have pause() and resume() methods (since version 2.1.3). See Pausing and Resuming Listener Containers for more information.

B.5.5. Stateful Retry

Starting with version 2.1.3, you can configure stateful retry. See Stateful Retry for more information.

B.5.6. Client ID

Starting with version 2.1.1, you can now set the client.id prefix on @KafkaListener. Previously, to customize the client ID, you needed a separate consumer factory (and container factory) per listener. The prefix is suffixed with -n to provide unique client IDs when you use concurrency.

B.5.7. Logging Offset Commits

By default, logging of topic offset commits is performed with the DEBUG logging level. Starting with version 2.1.2, a new property in ContainerProperties called commitLogLevel lets you specify the log level for these messages. See Using KafkaMessageListenerContainer for more information.

B.5.8. Default @KafkaHandler

Starting with version 2.1.3, you can designate one of the @KafkaHandler annotations on a class-level @KafkaListener as the default. See @KafkaListener on a Class for more information.

B.5.9. ReplyingKafkaTemplate

Starting with version 2.1.3, a subclass of KafkaTemplate is provided to support request/reply semantics. See Using ReplyingKafkaTemplate for more information.

B.5.10. ChainedKafkaTransactionManager

Version 2.1.3 introduced the ChainedKafkaTransactionManager. See Using ChainedKafkaTransactionManager for more information.

B.5.11. Migration Guide from 2.0

See the 2.0 to 2.1 Migration guide.

B.6. Changes Between 1.3 and 2.0

B.6.1. Spring Framework and Java Versions

The Spring for Apache Kafka project now requires Spring Framework 5.0 and Java 8.

B.6.2. @KafkaListener Changes

You can now annotate @KafkaListener methods (and classes and @KafkaHandler methods) with @SendTo. If the method returns a result, it is forwarded to the specified topic. See Forwarding Listener Results using @SendTo for more information.

B.6.3. Message Listeners

Message listeners can now be aware of the Consumer object. See Message Listeners for more information.

B.6.4. Using ConsumerAwareRebalanceListener

Rebalance listeners can now access the Consumer object during rebalance notifications. See Rebalancing Listeners for more information.

B.7. Changes Between 1.2 and 1.3

B.7.1. Support for Transactions

The 0.11.0.0 client library added support for transactions. The KafkaTransactionManager and other support for transactions have been added. See Transactions for more information.

B.7.2. Support for Headers

The 0.11.0.0 client library added support for message headers. These can now be mapped to and from spring-messaging MessageHeaders. See Message Headers for more information.

B.7.3. Creating Topics

The 0.11.0.0 client library provides an AdminClient, which you can use to create topics. The KafkaAdmin uses this client to automatically add topics defined as @Bean instances.

B.7.4. Support for Kafka Timestamps

KafkaTemplate now supports an API to add records with timestamps. New KafkaHeaders have been introduced regarding timestamp support. Also, new KafkaConditions.timestamp() and KafkaMatchers.hasTimestamp() testing utilities have been added. See Using KafkaTemplate, @KafkaListener Annotation, and Testing Applications for more details.

B.7.5. @KafkaListener Changes

You can now configure a KafkaListenerErrorHandler to handle exceptions. See Handling Exceptions for more information.

By default, the @KafkaListener id property is now used as the group.id property, overriding the property configured in the consumer factory (if present). Further, you can explicitly configure the groupId on the annotation. Previously, you would have needed a separate container factory (and consumer factory) to use different group.id values for listeners. To restore the previous behavior of using the factory configured group.id, set the idIsGroup property on the annotation to false.

B.7.6. @EmbeddedKafka Annotation

For convenience, a test class-level @EmbeddedKafka annotation is provided, to register KafkaEmbedded as a bean. See Testing Applications for more information.

B.7.7. Kerberos Configuration

Support for configuring Kerberos is now provided. See JAAS and Kerberos for more information.

B.8. Changes Between 1.1 and 1.2

This version uses the 0.10.2.x client.

B.9. Changes Between 1.0 and 1.1

B.9.1. Kafka Client

This version uses the Apache Kafka 0.10.x.x client.

B.9.2. Batch Listeners

Listeners can be configured to receive the entire batch of messages returned by the consumer.poll() operation, rather than one at a time.

B.9.3. Null Payloads

Null payloads are used to “delete” keys when you use log compaction.

B.9.4. Initial Offset

When explicitly assigning partitions, you can now configure the initial offset relative to the current position for the consumer group, rather than absolute or relative to the current end.

B.9.5. Seek

You can now seek the position of each topic or partition. You can use this to set the initial position during initialization when group management is in use and Kafka assigns the partitions. You can also seek when an idle container is detected or at any arbitrary point in your application’s execution. See Seeking to a Specific Offset for more information.