当前位置：首页 > news >正文

docker-compose部署kafka

news 2026/4/2 18:06:32

docker-compose.yml配置

version: "3"
services:kafka:image: 'bitnami/kafka:latest'ports:- '7050:7050'environment:- KAFKA_ENABLE_KRAFT=yes- KAFKA_CFG_PROCESS_ROLES=broker,controller- KAFKA_CFG_CONTROLLER_LISTENER_NAMES=CONTROLLER- KAFKA_CFG_LISTENERS=PLAINTEXT://:7050,CONTROLLER://:7051- KAFKA_CFG_LISTENER_SECURITY_PROTOCOL_MAP=CONTROLLER:PLAINTEXT,PLAINTEXT:PLAINTEXT- KAFKA_CFG_ADVERTISED_LISTENERS=PLAINTEXT://183.56.203.157:7050- KAFKA_BROKER_ID=1- KAFKA_CFG_CONTROLLER_QUORUM_VOTERS=1@0.0.0.0:7051- ALLOW_PLAINTEXT_LISTENER=yes

kafka UI界面

docker run -d --name kafka-map -p 8049:8080 -e DEFAULT_USERNAME=admin -e DEFAULT_PASSWORD=admin dushixiang/kafka-map:latest

docker run -p 8080:8080 -e KAFKA_BROKERS=host.docker.internal:9092 docker.redpanda.com/vectorized/console:master-173596f

UI界面总览

https://towardsdatascience.com/overview-of-ui-tools-for-monitoring-and-management-of-apache-kafka-clusters-8c383f897e80

kafka学习

生产者

import org.apache.kafka.clients.producer.Callback
import org.apache.kafka.clients.producer.KafkaProducer
import org.apache.kafka.clients.producer.ProducerConfig
import org.apache.kafka.clients.producer.ProducerRecord
import org.apache.kafka.common.serialization.StringSerializer
import org.junit.Test
import java.util.*/*** @Description :* @Author  xiaomh* @date  2022/8/5 15:58*/
class CustomProducer {//异步发送@Testfun customProducer() {//配置val properties = Properties()//链接kafkaproperties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)//发送数据for (i in 0 until 5) {//黏性发送，达到设置的数据最大值/时间后，切换分区（不会是当前分区）kafkaProducer.send(ProducerRecord("xiao1", "customProducer,count::$i"))}//关闭资源kafkaProducer.close()}//同步发送@Testfun customProducerSync() {//配置val properties = Properties()//链接kafkaproperties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)//发送数据for (i in 0 until 5) {//黏性发送，达到设置的数据最大值/时间后，切换分区（不会是当前分区）kafkaProducer.send(ProducerRecord("xiao1", "customProducerSync,count::$i")).get()}//关闭资源kafkaProducer.close()}//回调异步发送@Testfun customProducerCallback() {//配置val properties = Properties()//链接kafkaproperties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)//发送数据for (i in 0 until 500) {//黏性发送，达到设置的数据最大值/时间后，切换分区（不会是当前分区）kafkaProducer.send(ProducerRecord("xiao1", "customProducerCallback,count::$i"), Callback{ metadata, exception ->if (exception == null) {println("主题：${metadata.topic()},分区：${metadata.partition()}")}})//测试分区策略Thread.sleep(1)}//关闭资源kafkaProducer.close()}//回调异步发送+使用分区@Testfun customProducerCallbackPartitions1() {//配置val properties = Properties()//链接kafkaproperties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)//发送数据for (i in 0 until 5) {//1.没有指明partition值但有key的情况下，将key的hash值与topic的partition数进行取余得到partition值//2.既没有partition值又没有key值的情况下，Kafka采用Sticky Partition（黏性分区器）//key可以作为producer数据名，让consumer通过key找到kafkaProducer.send(ProducerRecord("xiao1", 1, "", "customProducerCallbackPartitions,count::$i"), Callback{ metadata, exception ->if (exception == null) {println("主题：${metadata.topic()},分区：${metadata.partition()}")}})}//关闭资源kafkaProducer.close()}//回调异步发送+自定义分区@Testfun customProducerCallbackPartitions2() {//配置val properties = Properties()//链接kafka，集群链接使用"183.56.203.157:7050,183.56.203.157:7051"properties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//关联自定义分区器properties[ProducerConfig.PARTITIONER_CLASS_CONFIG] ="com.umh.medicalbookingplatform.b2bapi.config.MyPartitioner"//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)//发送数据for (i in 0 until 50) {//1.没有指明partition值但有key的情况下，将key的hash值与topic的partition数进行取余得到partition值//2.既没有partition值又没有key值的情况下，Kafka采用Sticky Partition（黏性分区器）//key可以作为producer数据名，让consumer通过key找到kafkaProducer.send(ProducerRecord("xiao1", "felix is strong,count::$i"), Callback{ metadata, exception ->if (exception == null) {println("主题：${metadata.topic()},分区：${metadata.partition()}")}})}//关闭资源kafkaProducer.close()}//自定义配置缓冲区、批次、等待时间、压缩@Testfun customProducerParameters() {//配置val properties = Properties()properties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//缓冲区大小。默认32，64=33554432x2properties[ProducerConfig.BUFFER_MEMORY_CONFIG] = 33554432//批次大小。默认16kproperties[ProducerConfig.BATCH_SIZE_CONFIG] = 16384//等待时间。默认0properties[ProducerConfig.LINGER_MS_CONFIG] = 1//压缩.压缩，默认 none，可配置值 gzip、snappy、lz4 和 zstdproperties[ProducerConfig.COMPRESSION_TYPE_CONFIG] = "snappy"//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)for (i in 0 until 10) {//1.没有指明partition值但有key的情况下，将key的hash值与topic的partition数进行取余得到partition值//2.既没有partition值又没有key值的情况下，Kafka采用Sticky Partition（黏性分区器）//key可以作为producer数据名，让consumer通过key找到kafkaProducer.send(ProducerRecord("xiao1", "customProducerParameters::$i"), Callback{ metadata, exception ->if (exception == null) {println("主题：${metadata.topic()},分区：${metadata.partition()}")}})}//关闭资源kafkaProducer.close()}//ack、重试次数配置@Testfun customProducerAck() {//配置val properties = Properties()properties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//ackproperties[ProducerConfig.ACKS_CONFIG] = "1"//重试次数properties[ProducerConfig.RETRIES_CONFIG] = 30//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)for (i in 0 until 10) {//1.没有指明partition值但有key的情况下，将key的hash值与topic的partition数进行取余得到partition值//2.既没有partition值又没有key值的情况下，Kafka采用Sticky Partition（黏性分区器）//key可以作为producer数据名，让consumer通过key找到kafkaProducer.send(ProducerRecord("xiao1", "customProducerAck::$i"), Callback{ metadata, exception ->if (exception == null) {println("主题：${metadata.topic()},分区：${metadata.partition()}")}})}//关闭资源kafkaProducer.close()}//事物@Testfun customProducerTransaction() {//配置val properties = Properties()properties[ProducerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//指定对应key和value的序列化类型(二选一)
//        properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = "org.apache.kafka.common.serialization.StringSerializer"properties[ProducerConfig.KEY_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.nameproperties[ProducerConfig.VALUE_SERIALIZER_CLASS_CONFIG] = StringSerializer::class.java.name//指定事务id,一定要指定！！properties[ProducerConfig.TRANSACTIONAL_ID_CONFIG] = UUID.randomUUID().toString()//创建kafka生产者对象val kafkaProducer = KafkaProducer<String, String>(properties)//开启事务kafkaProducer.initTransactions()kafkaProducer.beginTransaction()try {for (i in 0 until 10) {//1.没有指明partition值但有key的情况下，将key的hash值与topic的partition数进行取余得到partition值//2.既没有partition值又没有key值的情况下，Kafka采用Sticky Partition（黏性分区器）//key可以作为producer数据名，让consumer通过key找到kafkaProducer.send(ProducerRecord("xiao1", "customProducerTransaction::$i"), Callback{ metadata, exception ->if (exception == null) {println("主题：${metadata.topic()},分区：${metadata.partition()}")}})}
//            val test: Int = 1 / 0kafkaProducer.commitTransaction()} catch (e: Exception) {kafkaProducer.abortTransaction()} finally {//关闭资源kafkaProducer.close()}}}

消费者

1、一个consumer group中有多个consumer组成，一个 topic有多个partition组成，现在的问题是，到底由哪个consumer来消费哪个 partition的数据。

2、Kafka有四种主流的分区分配策略： Range、RoundRobin、Sticky、CooperativeSticky。可以通过配置参数partition.assignment.strategy，修改分区的分配策略。默认策略是Range + CooperativeSticky。Kafka可以同时使用多个分区分配策略。

3、每个消费者都会和coordinator保持心跳（默认3s），一旦超时（session.timeout.ms=45s），该消费者会被移除，并触发再平衡；或者消费者处理消息的过长（max.poll.interval.ms5分钟），也会触发再平衡

package com.umh.medicalbookingplatform.apiimport com.alibaba.fastjson.parser.ParserConfig
import com.fasterxml.jackson.databind.MapperFeature
import com.umh.medicalbookingplatform.core.audit.SpringSecurityAuditorAware
import com.umh.medicalbookingplatform.core.config.CoreConfiguration
import com.umh.medicalbookingplatform.core.jsonview.JsonViews
import com.umh.medicalbookingplatform.core.properties.ApplicationProperties
import com.umh.medicalbookingplatform.core.utils.ApplicationJsonObjectMapper
import org.jboss.resteasy.client.jaxrs.ResteasyClientBuilder
import org.keycloak.OAuth2Constants
import org.keycloak.admin.client.Keycloak
import org.keycloak.admin.client.KeycloakBuilder
import io.swagger.v3.oas.models.Components
import io.swagger.v3.oas.models.OpenAPI
import org.springframework.beans.factory.annotation.Autowired
import org.springframework.boot.autoconfigure.SpringBootApplication
import org.springframework.boot.runApplication
import org.springframework.boot.web.servlet.ServletComponentScan
import org.springframework.cache.annotation.EnableCaching
import org.springframework.context.annotation.Bean
import org.springframework.context.annotation.Import
import org.springframework.data.domain.AuditorAware
import org.springframework.data.jpa.repository.config.EnableJpaAuditing
import org.springframework.http.MediaType
import org.springframework.http.converter.HttpMessageConverter
import org.springframework.http.converter.ResourceHttpMessageConverter
import org.springframework.http.converter.json.MappingJackson2HttpMessageConverter
import org.springframework.scheduling.annotation.EnableScheduling
import org.springframework.web.servlet.config.annotation.WebMvcConfigurer
import java.security.Security
import java.util.*
import io.swagger.v3.oas.models.info.Info
import io.swagger.v3.oas.models.info.License
import org.apache.kafka.clients.consumer.ConsumerConfig
import org.apache.kafka.clients.consumer.ConsumerRecords
import org.apache.kafka.clients.consumer.KafkaConsumer
import org.apache.kafka.common.TopicPartition
import org.apache.kafka.common.serialization.StringDeserializer
import org.keycloak.adapters.KeycloakConfigResolver
import org.keycloak.adapters.springboot.KeycloakSpringBootConfigResolver
import org.keycloak.adapters.springboot.KeycloakSpringBootProperties
import org.springframework.http.converter.StringHttpMessageConverter
import java.time.Duration
import java.util.concurrent.TimeUnit@EnableJpaAuditing
@EnableCaching
@EnableScheduling
@SpringBootApplication
@Import(CoreConfiguration::class)
@ServletComponentScan("com.umh.medicalbookingplatform")
open class ApiApplication : WebMvcConfigurer {@Autowiredprivate lateinit var appProperties: ApplicationProperties@Autowiredprivate lateinit var keycloakSpringBootProperties: KeycloakSpringBootProperties@Beanfun keycloakConfigResolver(): KeycloakConfigResolver {return KeycloakSpringBootConfigResolver()}@Beanfun fastJson(){ParserConfig.getGlobalInstance().isAutoTypeSupport = true}@Beanfun customConsumer() {//配置val properties = Properties()//连接properties[ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//反序列化(注意写法：生产者是序列化，消费者是反序列化)properties[ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG] = StringDeserializer::class.java.nameproperties[ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG] = StringDeserializer::class.java.name//配置消费者组id(就算消费者组只有一个消费者也需要)//当消费者组ID相同时，表示他们在同一个消费者组//当有三个分区，而消费者组里又有三个消费者时，消费者会各自自动选取一个分区进行消费properties[ConsumerConfig.GROUP_ID_CONFIG] = "test"//1.创建一个消费者val kafkaConsumer = KafkaConsumer<String, String>(properties)//2.定义主题 xiao1val topics = mutableListOf<String>()topics.add("xiao1")kafkaConsumer.subscribe(topics)//3.消费数据while (true) {val consumerRecord: ConsumerRecords<String, String> = kafkaConsumer.poll(Duration.ofSeconds(1))for (msg in consumerRecord) {println("consumer,msg:::$msg")}}}//    @Beanfun customConsumerPartition() {//配置val properties = Properties()//连接properties[ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//反序列化(注意写法：生产者是序列化，消费者是反序列化)properties[ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG] = StringDeserializer::class.java.nameproperties[ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG] = StringDeserializer::class.java.name//配置消费者组id(就算消费者组只有一个消费者也需要)//当消费者组ID相同时，表示他们在同一个消费者组properties[ConsumerConfig.GROUP_ID_CONFIG] = UUID.randomUUID().toString()//1.创建一个消费者val kafkaConsumer = KafkaConsumer<String, String>(properties)//2.定义主题对应的分区val topicPartition = mutableListOf<TopicPartition>()topicPartition.add(TopicPartition("xiao1", 1))kafkaConsumer.assign(topicPartition)//3.消费数据while (true) {val consumerRecord: ConsumerRecords<String, String> = kafkaConsumer.poll(Duration.ofSeconds(1))for (msg in consumerRecord) {println("msg:::$msg")}}}@Bean(name = ["keycloakGlobalCmsApi"])fun keycloakGlobalCmsApiInstance(): Keycloak {return KeycloakBuilder.builder().serverUrl(appProperties.keycloakAuthServerUrl)//https://keycloak.umhgp.com/auth.realm(appProperties.keycloakGlobalCmsRealm)//global_cms.clientId(appProperties.keycloakGlobalCmsClient)//global-cms.username(appProperties.keycloakApiUsername)//medical-booking-platform-system-uat.password(appProperties.keycloakApiPassword)//Kas7aAnC76eGVHv5.grantType(OAuth2Constants.PASSWORD).resteasyClient(ResteasyClientBuilder().connectTimeout(10, TimeUnit.SECONDS).readTimeout(10, TimeUnit.SECONDS).connectionPoolSize(100).build()).build()}@Bean(name = ["keycloakGlobalProfileApi"])fun keycloakGlobalProfileApiInstance(): Keycloak {return KeycloakBuilder.builder().serverUrl(appProperties.keycloakAuthServerUrl).realm(appProperties.keycloakGlobalProfileRealm).clientId(appProperties.keycloakGlobalProfileClient).username(appProperties.keycloakApiUsername).password(appProperties.keycloakApiPassword).grantType(OAuth2Constants.PASSWORD).resteasyClient(ResteasyClientBuilder().connectTimeout(10, TimeUnit.SECONDS).readTimeout(10, TimeUnit.SECONDS).connectionPoolSize(100).build()).build()}@Bean(name = ["keycloakBookingSystemApi"])fun keycloakBookingSystemApiInstance(): Keycloak {return KeycloakBuilder.builder().serverUrl(appProperties.keycloakAuthServerUrl).realm(appProperties.keycloakBookingSystemRealm).clientId(appProperties.keycloakBookingSystemClient).username(appProperties.keycloakApiUsername).password(appProperties.keycloakApiPassword).grantType(OAuth2Constants.PASSWORD).resteasyClient(ResteasyClientBuilder().connectTimeout(10, TimeUnit.SECONDS).readTimeout(10, TimeUnit.SECONDS).connectionPoolSize(100).build()).build()}@Bean(name = ["keycloakUmhBookingSystemApi"])fun keycloakBookingSystemUmhApiInstance(): Keycloak {return KeycloakBuilder.builder().serverUrl(appProperties.keycloakAuthServerUrl).realm(appProperties.keycloakUmhBookingSystemRealm).clientId(appProperties.keycloakUmhBookingSystemClient).username(appProperties.keycloakApiUsername).password(appProperties.keycloakApiPassword).grantType(OAuth2Constants.PASSWORD).resteasyClient(ResteasyClientBuilder().connectTimeout(10, TimeUnit.SECONDS).readTimeout(10, TimeUnit.SECONDS).connectionPoolSize(100).build()).build()}@Beaninternal fun auditorProvider(): AuditorAware<UUID> {return SpringSecurityAuditorAware()}@Beanfun customOpenAPI(): OpenAPI? {return OpenAPI().components(Components()).info(Info().title("medical-booking-platform").version("1.5.8").license(License().name("Apache 2.0").url("http://springdoc.org")))}override fun configureMessageConverters(converters: MutableList<HttpMessageConverter<*>>) {
//        ActuatorMediaTypes()val supportedMediaTypes = ArrayList<MediaType>()supportedMediaTypes.add(MediaType.APPLICATION_JSON)supportedMediaTypes.add(MediaType.valueOf("application/vnd.spring-boot.actuator.v3+json"))supportedMediaTypes.add(MediaType.TEXT_PLAIN)val converter = MappingJackson2HttpMessageConverter()val objectMapper = ApplicationJsonObjectMapper()objectMapper.setConfig(objectMapper.serializationConfig.withView(JsonViews.Admin::class.java))objectMapper.configure(MapperFeature.DEFAULT_VIEW_INCLUSION, true)converter.objectMapper = objectMapperconverter.setPrettyPrint(true)converter.supportedMediaTypes = supportedMediaTypesconverters.add(0, StringHttpMessageConverter())converters.add(1, converter)converters.add(ResourceHttpMessageConverter())}}fun main(args: Array<String>) {Security.setProperty("crypto.policy", "unlimited")runApplication<ApiApplication>(*args)
}

range（范围）

Kafka 默认的分区分配策略就是 Range + CooperativeSticky，所以不需要修改策略。

消费者分区操作：7分区2个消费者时

消费者1：消费分区0123

消费者2：消费分区456

在同一个消费者组，三消费者的情况下，如果其中一个宕机，45秒后会把消费者0需要处理的数据整个搬到消费者1或者消费者2.

结果：Consumer1=01234 或者 Consumer2=01256

随后如果再传输数据，消费者组会根据当前的消费者重新组织分配

Consumer0宕机45秒后再次传数据结果：Consumer1=0123 Consumer2=456

RoundRobin（轮询）

RoundRobin 针对集群中所有Topic而言。 RoundRobin 轮询分区策略，是把所有的 partition 和所有的 consumer 都列出来，然后按照 hashcode 进行排序，最后通过轮询算法来分配 partition 给到各个消费者。

策略分配的修改

    @Beanfun customConsumer() {//配置val properties = Properties()//连接properties[ConsumerConfig.BOOTSTRAP_SERVERS_CONFIG] = "183.56.218.28:8000"//反序列化(注意写法：生产者是序列化，消费者是反序列化)properties[ConsumerConfig.KEY_DESERIALIZER_CLASS_CONFIG] = StringDeserializer::class.java.nameproperties[ConsumerConfig.VALUE_DESERIALIZER_CLASS_CONFIG] = StringDeserializer::class.java.name//配置消费者组id(就算消费者组只有一个消费者也需要)//当消费者组ID相同时，表示他们在同一个消费者组//当有三个分区，而消费者组里又有三个消费者时，消费者会各自自动选取一个分区进行消费properties[ConsumerConfig.GROUP_ID_CONFIG] = "test"//设置分区分配策略properties[ConsumerConfig.PARTITION_ASSIGNMENT_STRATEGY_CONFIG] = "org.apache.kafka.clients.consumer.RoundRobinAssignor"//1.创建一个消费者val kafkaConsumer = KafkaConsumer<String, String>(properties)//2.定义主题 xiao1val topics = mutableListOf<String>()topics.add("xiao1")kafkaConsumer.subscribe(topics)//3.消费数据while (true) {val consumerRecord: ConsumerRecords<String, String> = kafkaConsumer.poll(Duration.ofSeconds(1))for (msg in consumerRecord) {println("consumer,msg:::$msg")}}}

注意：06为一组给到一个消费者，3为一组给到另外一个消费者。45秒后重新发送数据，consumer2：0246，consumer3：135

Sticky （黏性）

（1）停止掉 0 号消费者，快速重新发送消息观看结果（45s 以内，越快越好）。

1 号消费者：消费到 2、5、3 号分区数据。

2 号消费者：消费到 4、6 号分区数据。

0 号消费者的任务会按照粘性规则，尽可能均衡的随机分成 0 和 1 号分区数据，分别由 1 号消费者或者 2 号消费者消费。

说明：0 号消费者挂掉后，消费者组需要按照超时时间 45s 来判断它是否退出，所以需要等待，时间到了 45s 后，判断它真的退出就会把任务分配给其他 broker 执行。

（2）再次重新发送消息观看结果（45s 以后）。

1 号消费者：消费到 2、3、5 号分区数据。

2 号消费者：消费到 0、1、4、6 号分区数据。

说明：消费者 0 已经被踢出消费者组，所以重新按照粘性方式分配。

随机+均匀

宕机后分配的消费者和45秒后分配消费者一样

宕机（3消费者变2消费者）：1403，235

45秒后2消费者：1403，235

本文转自 https://blog.csdn.net/weixin_52925162/article/details/126280062?ops_request_misc=%257B%2522request%255Fid%2522%253A%2522170100111416800225544545%2522%252C%2522scm%2522%253A%252220140713.130102334.pc%255Fall.%2522%257D&request_id=170100111416800225544545&biz_id=0&utm_medium=distribute.pc_search_result.none-task-blog-2_allfirst_rank_ecpm_v1~rank_v31_ecpm-8-126280062-null-null.142^v96pc_search_result_base9&utm_term=keycloak%20docker-compose&spm=1018.2226.3001.4187，如有侵权，请联系删除。

docker-compose部署kafka

kafka学习

生产者

消费者

range（范围）

RoundRobin（轮询）

Sticky （黏性）

相关文章：

docker-compose部署kafka

Spark与Hadoop的关系和区别

蓝桥杯-Excel地址[Java]

OSPF多区域配置-新版（12）

华为---USG6000V防火墙web基本配置示例

Ksher H5页面支付实例指导 (PHP实现)

https密钥认证、上传镜像实验

three.js使用精灵模型Sprite渲染森林

什么是数据可视化？数据可视化的流程与步骤

2022年山东省职业院校技能大赛高职组云计算赛项试卷第二场-容器云

Unity3D 中播放 RTSP 监控视频

[spark] DataFrame 的 checkpoint

flask文件夹列表改进版--Bug追踪

Elasticsearch之常用DSL语句

鸿蒙实战-库的调用（ArkTS）

观察者模式学习

人工智能_机器学习078_聚类算法_概念介绍_聚类升维_降维_各类聚类算法_有监督机器学习_无监督机器学习---人工智能工作笔记0118

基于AR+地图导航的景区智慧导览设计

git基本指令

ECMAScript基础入门

WeChatExporter：免费开源工具，三步轻松备份你的微信聊天记录到电脑

手机号码智能定位引擎：从数据解析到地理可视化的全链路解决方案

001、性能优化基础：慢SQL诊断与执行计划分析

终极视频修复指南：如何使用Untrunc轻松恢复损坏的MP4/MOV文件

YOLO11 + SAHI + TensorRT：三剑合璧，实现高精度小目标视频实时检测的工程实践

从数据孤岛到智能决策中枢：一体化系统如何重构 HR 数据流

实战演练：基于快马平台codex构建可一键部署的智能api接口生成器

模型蒸馏与量化：为什么大厂急需能把大模型跑在边缘端的SDE？

PT助手Plus终极配置指南：三步实现智能自动化下载生态

【MobaXterm进阶】SSH连接稳定性优化：Keepalive与超时设置详解