一、背景

此处简略记录一下 SpringBoot 和 Kafka 的整合。

二、实现步骤

1、引入 jar 包

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

2、编写生产者和消费者的配置

3、生产者配置

spring.application.name=kafka-springboot
# 配置 kafka 服务器的地址，多个以逗号隔开
spring.kafka.bootstrap-servers=localhost:9092,localhost:9093,localhost:9094
# 生产者配置
spring.kafka.producer.key-serializer=org.apache.kafka.common.serialization.StringSerializer
spring.kafka.producer.value-serializer=org.apache.kafka.common.serialization.StringSerializer
spring.kafka.producer.acks=1
spring.kafka.producer.retries=0
spring.kafka.producer.batch-size=16384
spring.kafka.producer.buffer-memory=33554432

4、消费者配置

# 消费者配置
# 敞开主动提交 ack
spring.kafka.consumer.enable-auto-commit=false
spring.kafka.consumer.auto-commit-interval=100
spring.kafka.consumer.auto-offset-reset=earliest
spring.kafka.consumer.max-poll-records=500
spring.kafka.consumer.key-deserializer=org.apache.kafka.common.serialization.StringDeserializer
spring.kafka.consumer.value-deserializer=org.apache.kafka.common.serialization.StringDeserializer
# 配置监听手动提交 ack , 生产一条数据完后，立刻提交
spring.kafka.listener.ack-mode=manual_immediate
# 经测试也是批量提交的 ack , 当生产完 spring.kafka.consumer.max-poll-records 这么多的数据时候，提交
#spring.kafka.listener.ack-mode=manual
spring.kafka.listener.poll-timeout=500S

5、消费者手动提交 ack

1、spring.kafka.consumer.enable-auto-commit 批改成 false
2、spring.kafka.listener.ack-mode 批改成
            |- manual: 示意手动提交，然而测试下来发现是批量提交
            |- manual_immediate: 示意手动提交，当调用 Acknowledgment#acknowledge 之后立马提交。

3、编写生产者代码

@Component
public class KafkaProducer implements CommandLineRunner {

    @Autowired
    private KafkaTemplate<String, String> kafkaTemplate;

    @Override
    public void run(String... args) {Executors.newSingleThreadScheduledExecutor().scheduleAtFixedRate(() ->
                {kafkaTemplate.send(KafkaConstant.TOPIC, String.valueOf(System.currentTimeMillis()))
                            .addCallback(new SuccessCallback<SendResult<String, String>>() {
                                @Override
                                public void onSuccess(SendResult<String, String> result) {if (null != result.getRecordMetadata()) {System.out.println("生产发送胜利 offset:" + result.getRecordMetadata().offset());
                                        return;
                                    }
                                    System.out.println("音讯发送胜利");
                                }
                            }, new FailureCallback() {
                                @Override
                                public void onFailure(Throwable throwable) {System.out.println("生产发送失败:" + throwable.getMessage());
                                }
                            });
                },
                0, 1, TimeUnit.SECONDS);
    }
}

1、生产的发送应用 KafkaTemplate。
2、依据发送的后果晓得，音讯发送胜利还是失败。

4、编写消费者代码

@Component
public class KafkaConsumer {@KafkaListener(topics = KafkaConstant.TOPIC, groupId = "kafka-springboot-001")
    public void consumer(ConsumerRecord<String, String> record, Acknowledgment ack) throws InterruptedException {System.out.println(LocalDateTime.now().format(DateTimeFormatter.ofPattern("yyyy-MM-dd HH:mm:ss")) + "接管到 kafka 音讯,partition:" + record.partition() + ",offset:" + record.offset() + "value:" + record.value());
        TimeUnit.SECONDS.sleep(1);
        ack.acknowledge();}
}

KafkaListener:
      topic: 示意须要监听的队列名称
      groupId: 示意消费者组的 id

三、运行后果

四、参考文档

1、https://docs.spring.io/spring-boot/docs/2.4.2/reference/htmlsingle/#boot-features-kafka

五、代码门路

https://gitee.com/huan1993/rabbitmq/tree/master/kafka-springboot/src/main/java/com/huan/study/kafka