Apache Kafka-生产者_批量发送消息的核心参数及功能实现

叁歲伎倆 2022-10-30 06:25 10阅读 0赞

### 文章目录 ###

*  概述
 *  参数设置
 *  Code
 *   *  POM依赖
     *  配置文件
     *  生产者
     *  消费者
     *  单元测试
     *  测试结果
 *  源码地址

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9hcnRpc2FuLmJsb2cuY3Nkbi5uZXQ_size_16_color_FFFFFF_t_70]

--------------------

# 概述 #

kafka中有个 micro batch 的概念 ，为了提高Producer 发送的性能。

不同于RocketMQ 提供了一个可以批量发送多条消息的 API 。 Kafka 的做法是：提供了一个 **RecordAccumulator 消息收集器**，将发送给相同 Topic 的相同 Partition 分区的消息们，缓冲一下，当满足条件时候，一次性批量将缓冲的消息提交给 Kafka Broker 。

--------------------

# 参数设置 #

[https://kafka.apache.org/24/documentation.html\#producerconfigs][https_kafka.apache.org_24_documentation.html_producerconfigs]

主要涉及的参数 ，三个条件，满足任一即会批量发送：

*  batch-size ：超过收集的消息数量的最大量。默认16KB

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70]

*  buffer-memory ：超过收集的消息占用的最大内存 , 默认32M

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 1]

*  linger.ms ：超过收集的时间的最大等待时长，单位：毫秒。

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 2]

--------------------

# Code #

## POM依赖 ##

<dependencies>
    		<dependency>
    			<groupId>org.springframework.boot</groupId>
    			<artifactId>spring-boot-starter-web</artifactId>
    		</dependency>
    
    		
    		<dependency>
    			<groupId>org.springframework.kafka</groupId>
    			<artifactId>spring-kafka</artifactId>
    		</dependency>
    
    		<dependency>
    			<groupId>org.springframework.boot</groupId>
    			<artifactId>spring-boot-starter-test</artifactId>
    			<scope>test</scope>
    		</dependency>
    		<dependency>
    			<groupId>junit</groupId>
    			<artifactId>junit</artifactId>
    			<scope>test</scope>
    		</dependency>
    	</dependencies>

--------------------

## 配置文件 ##

spring:
      # Kafka 配置项，对应 KafkaProperties 配置类
      kafka:
        bootstrap-servers: 192.168.126.140:9092 # 指定 Kafka Broker 地址，可以设置多个，以逗号分隔
        # Kafka Producer 配置项
        producer:
          acks: 1 # 0-不应答。1-leader 应答。all-所有 leader 和 follower 应答。
          retries: 3 # 发送失败时，重试发送的次数
          key-serializer: org.apache.kafka.common.serialization.StringSerializer # 消息的 key 的序列化
          value-serializer: org.springframework.kafka.support.serializer.JsonSerializer # 消息的 value 的序列化
          batch-size: 16384 # 每次批量发送消息的最大数量 单位 字节 默认 16K
          buffer-memory: 33554432 # 每次批量发送消息的最大内存 单位 字节 默认 32M
          properties:
            linger:
              ms: 10000 # 批处理延迟时间上限。[实际不会配这么长，这里用于测速]这里配置为 10 * 1000 ms 过后，不管是否消息数量是否到达 batch-size 或者消息大小到达 buffer-memory 后，都直接发送一次请求。
        # Kafka Consumer 配置项
        consumer:
          auto-offset-reset: earliest # 设置消费者分组最初的消费进度为 earliest
          key-deserializer: org.apache.kafka.common.serialization.StringDeserializer
          value-deserializer: org.springframework.kafka.support.serializer.JsonDeserializer
          properties:
            spring:
              json:
                trusted:
                  packages: com.artisan.springkafka.domain
        # Kafka Consumer Listener 监听器配置
        listener:
          missing-topics-fatal: false # 消费监听接口监听的主题不存在时，默认会报错。所以通过设置为 false ，解决报错
    
    logging:
      level:
        org:
          springframework:
            kafka: ERROR # spring-kafka
          apache:
            kafka: ERROR # kafka

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 3]

--------------------

## 生产者 ##

package com.artisan.springkafka.producer;
    
    import com.artisan.springkafka.constants.TOPIC;
    import com.artisan.springkafka.domain.MessageMock;
    import org.springframework.beans.factory.annotation.Autowired;
    import org.springframework.kafka.core.KafkaTemplate;
    import org.springframework.kafka.support.SendResult;
    import org.springframework.stereotype.Component;
    import org.springframework.util.concurrent.ListenableFuture;
    
    import java.util.Random;
    import java.util.concurrent.ExecutionException;
    
    /** * @author 小工匠 * @version 1.0 * @description: TODO * @date 2021/2/17 22:25 * @mark: show me the code , change the world */
    
    @Component
    public class ArtisanProducerMock { 
    
    
        @Autowired
        private KafkaTemplate<Object,Object> kafkaTemplate ;
    
    
        /** * 同步发送 * @return * @throws ExecutionException * @throws InterruptedException */
        public SendResult sendMsgSync() throws ExecutionException, InterruptedException { 
            // 模拟发送的消息
            Integer id = new Random().nextInt(100);
            MessageMock messageMock = new MessageMock(id,"artisanTestMessage-" + id);
            // 同步等待
           return  kafkaTemplate.send(TOPIC.TOPIC, messageMock).get();
        }
    
    
    
        public ListenableFuture<SendResult<Object, Object>> sendMsgASync() throws ExecutionException, InterruptedException { 
            // 模拟发送的消息
            Integer id = new Random().nextInt(100);
            MessageMock messageMock = new MessageMock(id,"messageSendByAsync-" + id);
            // 异步发送消息
            ListenableFuture<SendResult<Object, Object>> result = kafkaTemplate.send(TOPIC.TOPIC, messageMock);
            return result ;
    
        }
    
    }

--------------------

## 消费者 ##

package com.artisan.springkafka.consumer;
    
    import com.artisan.springkafka.domain.MessageMock;
    import com.artisan.springkafka.constants.TOPIC;
    import org.slf4j.Logger;
    import org.slf4j.LoggerFactory;
    import org.springframework.kafka.annotation.KafkaListener;
    import org.springframework.stereotype.Component;
    
    /** * @author 小工匠 * @version 1.0 * @description: TODO * @date 2021/2/17 22:33 * @mark: show me the code , change the world */
    
    @Component
    public class ArtisanCosumerMock { 
    
    
        private Logger logger = LoggerFactory.getLogger(getClass());
        private static final String CONSUMER_GROUP_PREFIX = "MOCK-A" ;
    
        @KafkaListener(topics = TOPIC.TOPIC ,groupId = CONSUMER_GROUP_PREFIX + TOPIC.TOPIC)
        public void onMessage(MessageMock messageMock){ 
            logger.info("【接受到消息][线程:{} 消息内容：{}]", Thread.currentThread().getName(), messageMock);
        }
    
    }

package com.artisan.springkafka.consumer;
    
    import com.artisan.springkafka.domain.MessageMock;
    import com.artisan.springkafka.constants.TOPIC;
    import org.slf4j.Logger;
    import org.slf4j.LoggerFactory;
    import org.springframework.kafka.annotation.KafkaListener;
    import org.springframework.stereotype.Component;
    
    /** * @author 小工匠 * @version 1.0 * @description: TODO * @date 2021/2/17 22:33 * @mark: show me the code , change the world */
    
    @Component
    public class ArtisanCosumerMockDiffConsumeGroup { 
    
    
        private Logger logger = LoggerFactory.getLogger(getClass());
    
        private static final String CONSUMER_GROUP_PREFIX = "MOCK-B" ;
    
        @KafkaListener(topics = TOPIC.TOPIC ,groupId = CONSUMER_GROUP_PREFIX + TOPIC.TOPIC)
        public void onMessage(MessageMock messageMock){ 
            logger.info("【接受到消息][线程:{} 消息内容：{}]", Thread.currentThread().getName(), messageMock);
        }
    
    }

--------------------

## 单元测试 ##

package com.artisan.springkafka.produceTest;
    
    import com.artisan.springkafka.SpringkafkaApplication;
    import com.artisan.springkafka.producer.ArtisanProducerMock;
    import org.junit.Test;
    import org.junit.runner.RunWith;
    import org.slf4j.Logger;
    import org.slf4j.LoggerFactory;
    import org.springframework.beans.factory.annotation.Autowired;
    import org.springframework.boot.test.context.SpringBootTest;
    import org.springframework.kafka.support.SendResult;
    import org.springframework.test.context.junit4.SpringRunner;
    import org.springframework.util.concurrent.ListenableFuture;
    import org.springframework.util.concurrent.ListenableFutureCallback;
    
    import java.util.concurrent.CountDownLatch;
    import java.util.concurrent.ExecutionException;
    import java.util.concurrent.TimeUnit;
    
    /** * @author 小工匠 * * @version 1.0 * @description: TODO * @date 2021/2/17 22:40 * @mark: show me the code , change the world */
    
    @RunWith(SpringRunner.class)
    @SpringBootTest(classes = SpringkafkaApplication.class)
    public class ProduceMockTest { 
    
        private Logger logger = LoggerFactory.getLogger(getClass());
    
    
        @Autowired
        private ArtisanProducerMock artisanProducerMock;
    
    
    
        @Test
        public void testAsynSend() throws ExecutionException, InterruptedException { 
            logger.info("开始发送");
    
            for (int i = 0; i < 2; i++) { 
                artisanProducerMock.sendMsgASync().addCallback(new ListenableFutureCallback<SendResult<Object, Object>>() { 
                    @Override
                    public void onFailure(Throwable throwable) { 
                        logger.info(" 发送异常{}]]", throwable);
    
                    }
                    @Override
                    public void onSuccess(SendResult<Object, Object> objectObjectSendResult) { 
                        logger.info("回调结果 Result = topic:[{}] , partition:[{}], offset:[{}]",
                             objectObjectSendResult.getRecordMetadata().topic(),
                                objectObjectSendResult.getRecordMetadata().partition(),
                                objectObjectSendResult.getRecordMetadata().offset());
                    }
                });
                // 发送2次 每次间隔5秒， 凑够我们配置的 linger: ms: 10000
                TimeUnit.SECONDS.sleep(5);
            }
    
            // 阻塞等待，保证消费
            new CountDownLatch(1).await();
    
        }
    
    }

异步发送2条消息，每次发送消息之间， sleep 5 秒，以便达到配置的 linger.ms 最大等待时长10秒。

--------------------

## 测试结果 ##

2021-02-18 10:58:53.360  INFO 24736 --- [           main] c.a.s.produceTest.ProduceMockTest        : 开始发送
    2021-02-18 10:59:03.555  INFO 24736 --- [ad | producer-1] c.a.s.produceTest.ProduceMockTest        : 回调结果 Result =  topic:[MOCK_TOPIC] , partition:[0], offset:[30]
    2021-02-18 10:59:03.556  INFO 24736 --- [ad | producer-1] c.a.s.produceTest.ProduceMockTest        : 回调结果 Result =  topic:[MOCK_TOPIC] , partition:[0], offset:[31]
    2021-02-18 10:59:03.595  INFO 24736 --- [ntainer#0-0-C-1] c.a.s.consumer.ArtisanCosumerMock        : 【接受到消息][线程:org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1 消息内容：MessageMock{ id=6, name='messageSendByAsync-6'}]
    2021-02-18 10:59:03.595  INFO 24736 --- [ntainer#1-0-C-1] a.s.c.ArtisanCosumerMockDiffConsumeGroup : 【接受到消息][线程:org.springframework.kafka.KafkaListenerEndpointContainer#1-0-C-1 消息内容：MessageMock{ id=6, name='messageSendByAsync-6'}]
    2021-02-18 10:59:03.595  INFO 24736 --- [ntainer#1-0-C-1] a.s.c.ArtisanCosumerMockDiffConsumeGroup : 【接受到消息][线程:org.springframework.kafka.KafkaListenerEndpointContainer#1-0-C-1 消息内容：MessageMock{ id=94, name='messageSendByAsync-94'}]
    2021-02-18 10:59:03.595  INFO 24736 --- [ntainer#0-0-C-1] c.a.s.consumer.ArtisanCosumerMock        : 【接受到消息][线程:org.springframework.kafka.KafkaListenerEndpointContainer#0-0-C-1 消息内容：MessageMock{ id=94, name='messageSendByAsync-94'}]

![在这里插入图片描述][watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 4]

10 秒后，满足批量消息的最大等待时长，所以 2 条消息被 Producer 批量发送。同时我们配置的是 acks=1 ，需要等待发送成功后，才会回调 ListenableFutureCallback 的方法。

当然了，我们这里都是为了测试，设置的这么长的间隔，实际中需要根据具体的业务场景设置一个合理的值。  
![在这里插入图片描述][20210218113735320.gif_pic_center]

--------------------

# 源码地址 #

[https://github.com/yangshangwei/boot2/tree/master/springkafkaBatchSend][https_github.com_yangshangwei_boot2_tree_master_springkafkaBatchSend]

[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9hcnRpc2FuLmJsb2cuY3Nkbi5uZXQ_size_16_color_FFFFFF_t_70]: /images/20221024/9839d778953d426ab8ade781cf861cb7.png
[https_kafka.apache.org_24_documentation.html_producerconfigs]: https://kafka.apache.org/24/documentation.html#producerconfigs
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70]: /images/20221024/850c19f5b7d349ada2915c817555f4a7.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 1]: /images/20221024/b9751f2035604debbb72fcffae0b37b5.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 2]: /images/20221024/54a4eb8eb55741b1b72a733bfe8ff204.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 3]: /images/20221024/dc12ad7e74044fc6833075bd4c6b900c.png
[watermark_type_ZmFuZ3poZW5naGVpdGk_shadow_10_text_aHR0cHM6Ly9ibG9nLmNzZG4ubmV0L3lhbmdzaGFuZ3dlaQ_size_16_color_FFFFFF_t_70 4]: /images/20221024/46bd15d8c63947bea4e0a94744630c47.png
[20210218113735320.gif_pic_center]: /images/20221024/06ac6d0d74694a88a3036c74eb88ddc8.png
[https_github.com_yangshangwei_boot2_tree_master_springkafkaBatchSend]: https://github.com/yangshangwei/boot2/tree/master/springkafkaBatchSend