version: "3.6"
networks:
  hadoop_network:

x-template: &template
  image: neshkeev/hadoop
  restart: on-failure
  networks:
    - hadoop_network
  healthcheck: &hc
    interval: 5s
    timeout: 3s
    start_period: 10s
    retries: 20
  environment: &env
    CORE-SITE.XML_fs.defaultFS: hdfs://namenode:9000

    HDFS-SITE.XML_dfs.replication: 3
    HDFS-SITE.XML_dfs.permissions: false
    HDFS-SITE.XML_dfs.datanode.hostname: localhost

    MAPRED-SITE.XML_mapreduce.framework.name: yarn
    MAPRED-SITE.XML_mapreduce.application.classpath: $$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/*:$$HADOOP_MAPRED_HOME/share/hadoop/mapreduce/lib/*

    MAPRED-SITE.XML_mapreduce.jobhistory.address: host.docker.internal:10020
    MAPRED-SITE.XML_mapreduce.jobhistory.webapp.address: host.docker.internal:19888
    MAPRED-SITE.XML_mapreduce.jobhistory.bind-host: 0.0.0.0

    YARN-SITE.XML_yarn.resourcemanager.hostname: resourcemanager
    YARN-SITE.XML_yarn.resourcemanager.webapp.address: host.docker.internal:8088
    YARN-SITE.XML_yarn.resourcemanager.bind-host: 0.0.0.0

    YARN-SITE.XML_yarn.nodemanager.aux-services: mapreduce_shuffle
    YARN-SITE.XML_yarn.nodemanager.env-whitelist: JAVA_HOME,HADOOP_COMMON_HOME,HADOOP_HDFS_HOME,HADOOP_CONF_DIR,CLASSPATH_PREPEND_DISTCACHE,HADOOP_YARN_HOME,HADOOP_HOME,PATH,LANG,TZ,HADOOP_MAPRED_HOME

    YARN-SITE.XML_yarn.web-proxy.address: host.docker.internal:9099
    YARN-SITE.XML_yarn.web-proxy.bind-host: 0.0.0.0

services:
  namenode:
    <<: *template
    hostname: &name namenode
    container_name: *name
    command:
      - hdfs
      - namenode
    healthcheck:
      <<: *hc
      test: bash /bin/check-ports 9870 9000
    ports:
      - "9000:9000"
      - "9870:9870"

  historyserver:
    <<: *template
    hostname: &name historyserver
    container_name: *name
    command:
      - mapred
      - historyserver
    healthcheck:
      <<: *hc
      test: bash /bin/check-ports 10020 19888
    ports:
      - "19888:19888"

  resourcemanager:
    <<: *template
    hostname: &name resourcemanager
    container_name: *name
    command:
      - yarn
      - resourcemanager
    healthcheck:
      <<: *hc
      test: bash /bin/check-ports 8088
    ports:
      - "8088:8088"

  proxyserver:
    <<: *template
    hostname: &name proxyserver
    container_name: *name
    command:
      - yarn
      - proxyserver
    healthcheck:
      <<: *hc
      test: bash /bin/check-ports 9099
    ports:
      - "9099:9099"
    depends_on:
      resourcemanager:
        condition: service_healthy

  worker-1: &dn
    <<: *template
    hostname: &name worker-1
    container_name: *name
    command:
      - bash
      - /bin/worker
    healthcheck:
      <<: *hc
      test: bash /bin/check-ports 18042 19864 19866
    ports:
      - "18042:18042"
      - "19864:19864"
    depends_on:
      namenode:
        condition: service_healthy
      resourcemanager:
        condition: service_healthy
    environment:
      <<: *env
      HDFS-SITE.XML_dfs.datanode.http.address: 0.0.0.0:19864
      HDFS-SITE.XML_dfs.datanode.address: 0.0.0.0:19866

      YARN-SITE.XML_yarn.nodemanager.webapp.address: $${yarn.nodemanager.hostname}:18042

  worker-2:
    << : *dn
    hostname: &name worker-2
    container_name: *name
    healthcheck:
      <<: *hc
      test: bash /bin/check-ports 28042 29864 29866
    ports:
      - "28042:28042"
      - "29864:29864"
    environment:
      <<: *env
      HDFS-SITE.XML_dfs.datanode.http.address: 0.0.0.0:29864
      HDFS-SITE.XML_dfs.datanode.address: 0.0.0.0:29866

      YARN-SITE.XML_yarn.nodemanager.webapp.address: $${yarn.nodemanager.hostname}:28042

  worker-3:
    << : *dn
    hostname: &name worker-3
    container_name: *name
    healthcheck:
      <<: *hc
      test: bash /bin/check-ports 38042 39864 39866
    ports:
      - "38042:38042"
      - "39864:39864"
    environment:
      <<: *env
      HDFS-SITE.XML_dfs.datanode.http.address: 0.0.0.0:39864
      HDFS-SITE.XML_dfs.datanode.address: 0.0.0.0:39866

      YARN-SITE.XML_yarn.nodemanager.webapp.address: $${yarn.nodemanager.hostname}:38042