Docker-compose搭建zookeeper-zookeeper集群

前言

最近在研究kafka与rabbitMQ对比，而因为自己电脑性能不是那么好，要搭建一个kafka集群和zookeeper集群最少也要开三台虚拟机，而且也挺麻烦，所以毋庸置疑选择了docker-compose来搭建一个伪分布式的集群。参考了很多资料，很多乱七八糟都是转发的，真正去研究的很少，所以在这里记录一下自己的搭建过程

软件版本

• docker： Docker version 1.13.1, build b2f74b2/1.13.1
• docke-compose：docker-compose version 1.23.2, build 1110ad0
• docker_ zookeeper_image : zookeeper:3.5.7
• docker_ kafka_iamge: wurstmeister/kafka:latest
• docker_ kafka-manager_image: sheepkiller/kafka-manager:latest

Compose file format compatibility matrix

Compose file format	Docker Engine
1	1.9.0+
2.0	1.10.0+
2.1	1.12.0+
2.2, 3.0, 3.1, 3.2	1.13.0+
2.3, 3.3, 3.4, 3.5	17.06.0+
2.4	17.12.0+
3.6	18.02.0+
3.7	18.06.0+

怎么安装的在这里就不多赘述了，相信必应、百度都有很多

准备

如果你对防火墙不感冒，那么在测试的时候建议你关闭防火墙；反之请开放如下端口

宿主机防火墙开放：9000/tcp 、2181/tcp、2182/tcp、2183/tcp、9092/tcp、9093/tcp、9094/tcp端口

Kafka是依赖zookeeper来存放一些数据的，所以kafka集群的启动必须依赖zookeeper集群

docker-compose.yml

version: '2'

services:
  zookeeper1:
    image: zookeeper:3.5.7
    hostname: zookeeper1
    container_name: zookeeper1
    privileged: true
    ports:
      - "2181:2181"
    volumes:
      - ./zookeeper1/conf:/conf
      - ./zookeeper1/data:/data
      - ./zookeeper1/datalog:/datalog
    environment:
      ZOO_MY_ID: 1
      ZOO_SERVERS: server.1=0.0.0.0:2888:3888;2181 server.2=zookeeper2:2888:3888;2181 server.3=zookeeper3:2888:3888;2181
    networks:
      net_zoo_kafka:
        ipv4_address: 172.23.0.11

  zookeeper2:
    image: zookeeper:3.5.7
    hostname: zookeeper2
    container_name: zookeeper2
    privileged: true
    ports:
      - "2182:2181"
    volumes:
      - ./zookeeper2/conf:/conf
      - ./zookeeper2/data:/data
      - ./zookeeper2/datalog:/datalog
    environment:
      ZOO_MY_ID: 2
      ZOO_SERVERS: server.1=zookeeper1:2888:3888;2181 server.2=0.0.0.0:2888:3888;2181 server.3=zookeeper3:2888:3888;2181
    networks:
      net_zoo_kafka:
        ipv4_address: 172.23.0.12

  zookeeper3:
    image: zookeeper:3.5.7
    hostname: zookeeper3
    container_name: zookeeper3
    privileged: true
    ports:
      - "2183:2181"
    volumes:
      - ./zookeeper3/conf:/conf
      - ./zookeeper3/data:/data
      - ./zookeeper3/datalog:/datalog
    environment:
      ZOO_MY_ID: 3
      ZOO_SERVERS: server.1=zookeeper1:2888:3888;2181 server.2=zookeeper2:2888:3888;2181 server.3=0.0.0.0:2888:3888;2181
    networks:
      net_zoo_kafka:
        ipv4_address: 172.23.0.13

  kafka1:
    image: docker.io/wurstmeister/kafka
    hostname: kafka1
    container_name: kafka1
    privileged: true
    ports:
      - "9092:9092"
    environment:
      KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://192.168.94.133:9092
      KAFKA_ADVERTISED_HOST_NAME: 192.168.94.133
      KAFKA_ADVERTISED_PORT: 9092
      KAFKA_ZOOKEEPER_CONNECT: zookeeper1:2181/kafka,zookeeper2:2182/kafka,zookeeper3:2183/kafka
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ./kafka1:/kafka
    links:
      - zookeeper1
      - zookeeper2
      - zookeeper3
    networks:
      net_zoo_kafka:
        ipv4_address: 172.23.0.14

  kafka2:
    image: docker.io/wurstmeister/kafka
    hostname: kafka2
    container_name: kafka2
    privileged: true
    ports:
      - "9093:9092"
    environment:
      KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://192.168.94.133:9093
      KAFKA_ADVERTISED_HOST_NAME: 192.168.94.133
      KAFKA_ADVERTISED_PORT: 9093
      KAFKA_ZOOKEEPER_CONNECT: zookeeper1:2181/kafka,zookeeper2:2182/kafka,zookeeper3:2183/kafka
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ./kafka2:/kafka
    links:
      - zookeeper1
      - zookeeper2
      - zookeeper3
    networks:
      net_zoo_kafka:
        ipv4_address: 172.23.0.15

  kafka3:
    image: docker.io/wurstmeister/kafka
    hostname: kafka3
    container_name: kafka3
    privileged: true
    ports:
      - "9094:9092"
    environment:
      KAFKA_ADVERTISED_LISTENERS: PLAINTEXT://192.168.94.133:9094
      KAFKA_ADVERTISED_HOST_NAME: 192.168.94.133
      KAFKA_ADVERTISED_PORT: 9094
      KAFKA_ZOOKEEPER_CONNECT: zookeeper1:2181/kafka,zookeeper2:2182/kafka,zookeeper3:2183/kafka
    volumes:
      - /var/run/docker.sock:/var/run/docker.sock
      - ./kafka3:/kafka
    links:
      - zookeeper1
      - zookeeper2
      - zookeeper3
    networks:
      net_zoo_kafka:
        ipv4_address: 172.23.0.16

  kafka-manager:
    image: docker.io/sheepkiller/kafka-manager
    container_name: kafka-manager
    hostname: kafka-manager
    privileged: true
    ports:
      - "9000:9000"
    links:            # 连接本compose文件创建的container
      - zookeeper1
      - zookeeper2
      - zookeeper3
      - kafka1
      - kafka2
      - kafka3
    # external_links:   # 连接compose以外的container
    environment:
      ZK_HOSTS: zookeeper1:2181/kafka,zookeeper2:2182/kafka,zookeeper3:2183/kafka
      KAFKA_BROKERS: kafka1:9092,kafka2:9093,kafka3:9094
      APPLICATION_SECRET: letmein
      KM_ARGS: -Djava.net.preferIPv4Stack=true
    networks:
      net_zoo_kafka:
        ipv4_address: 172.23.0.17

networks:
  net_zoo_kafka:
    driver: bridge
    ipam:
      config:
        - subnet: 172.23.0.0/25
          gateway: 172.23.0.1

详解

1. version 选择 2 版本，因为3版本不支持gateway, 我这里为了方便，将网络的创建放在了compose文件中，你也可以在外部用docker network create --driver bridge --subnet 172.23.0.0/25 --gateway 172.23.0.1 zookeeper_network的方式创建一个bridge的网桥，compose文件中用external指定外部网络
2. 所有容器都在创建的网桥的子网内，避免访问不到的问题
3. zookeeper集群只需要暴露每个容器的2181端口出去，而2888/3888端口，因为容器之间在一个子网，可以互通，所以不需要暴露
4. kafka 集群的 KAFKA_ADVERTISED_LISTENERS / KAFKA_ADVERTISED_HOST_NAME 配置成你宿主机的IP ! 记得修改

验证zookeeper集群

启动后用 docker ps命令查看容器启动没有问题，应该成功启动了3个zookeeper容器、3个kafka容器和1个kafka-manager容器

进入zookeeper集群中任意一个容器

输入 zkServer.sh status命令查看zookeeper集群状态，此时你看到该节点 Node: follower 或者 Node: leader,表示zookeeper集群启动成功

验证kafka集群

Ctrl + p + q 退出容器

进入任意一台kafka容器 docker exec -it kafka1 /bin/bash ，进入 /opt/kafka/bin 目录

键入命令创建一个topic ./kafka-topics.sh --zookeeper zookeeper1:2181/kafka,zookeeper2:2182/kafka,zookeeper3:2183/kafka --create --replication-factor 2 --partitions 1 --topic topic1
键入命令查看所有topic ./kafka-topics.sh --zookeeper zookeeper1:2181/kafka,zookeeper2:2182/kafka,zookeeper3:2183/kafka --list
键入命令删除topic ./kafka-topics.sh --zookeeper zookeeper1:2181/kafka,zookeeper2:2182/kafka,zookeeper3:2183/kafka --delete --topic topic1

注：kafka的topic信息都是保存在zookeeper中的，所以要连接zookeeper集群，并且在上面的compose文件中我们明确的指定了kafka在zookeeper中的namespace, 即/kafka, 所以在操作topic时，务必指定namespace, 否则，会报错提示broker 为0

验证使用 Kafka-manager

再次确认宿主机 9000/tcp 端口是否开放，或者防火墙是否关闭

浏览器访问 宿主机IP:9000

添加一个Cluster

标红的地方注意一下，其他的地方根据你的需求修改；输入完成点击save

OK, 至此，整个zookeeper集群、kafka集群、kafka-manger都部署成功

结语

坑总是一个又一个，踩坑了，睡一觉，但别忘了爬出来。
如果大家有什么疑问，欢迎给我留言