文章目录
概述参数设置CodePOM依赖配置文件生产者消费者单元测试测试结果源码地址概述
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
/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 buffer-memory ：超过收集的消息占用的最大内存 , 默认32M linger.ms ：超过收集的时间的最大等待时长，单位：毫秒。
Code
POM依赖
<dependencies><dependency><groupId>org.springframework.boot</groupId><artifactId>spring-boot-starter-web</artifactId></dependency><!-- 引入 Spring-Kafka 依赖 --><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.mon.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 16Kbuffer-memory: 33554432 # 每次批量发送消息的最大内存 单位 字节 默认 32Mproperties: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 earliestkey-deserializer: org.mon.serialization.StringDeserializervalue-deserializer: org.springframework.kafka.support.serializer.JsonDeserializerproperties: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-kafkaapache:kafka: ERROR # kafka
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.ponent;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 /2/17 22:25* @mark: show me the code , change the world*/@Componentpublic class ArtisanProducerMock {@Autowiredprivate 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.ponent;/*** @author 小工匠* @version 1.0* @description: TODO* @date /2/17 22:33* @mark: show me the code , change the world*/@Componentpublic 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.ponent;/*** @author 小工匠* @version 1.0* @description: TODO* @date /2/17 22:33* @mark: show me the code , change the world*/@Componentpublic 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 /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());@Autowiredprivate ArtisanProducerMock artisanProducerMock;@Testpublic 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>>() {@Overridepublic void onFailure(Throwable throwable) {logger.info(" 发送异常{}]]", throwable);}@Overridepublic 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: 10000TimeUnit.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
-02-18 10:58:53.360 INFO 24736 --- [ main] c.a.s.produceTest.ProduceMockTest : 开始发送-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]-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]-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'}]-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'}]-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'}]-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'}]
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。
源码地址
/yangshangwei/boot2/tree/master/springkafkaBatchSend
