5-node-cluster.yml

version: '3.1'

services:
  namenode:
    image: renci/hadoop:2.9.0
    container_name: namenode
    volumes:
      - hadoop-public:/home/hadoop/public
      - ./site-files:/site-files
    restart: always
    hostname: namenode
    networks:
      - hadoop
    ports:
      - '50070:50070'
    environment:
      IS_NODE_MANAGER: 'false'
      IS_NAME_NODE: 'true'
      IS_SECONDARY_NAME_NODE: 'false'
      IS_DATA_NODE: 'false'
      IS_RESOURCE_MANAGER: 'false'
      CLUSTER_NODES: namenode resourcemanager worker1 worker2 worker3

  resourcemanager:
    image: renci/hadoop:2.9.0
    depends_on:
      - namenode
    container_name: resourcemanager
    volumes:
      - hadoop-public:/home/hadoop/public
      - ./site-files:/site-files
    restart: always
    hostname: resourcemanager
    networks:
      - hadoop
    ports:
      - '8088:8088'
    environment:
      IS_NODE_MANAGER: 'false'
      IS_NAME_NODE: 'false'
      IS_SECONDARY_NAME_NODE: 'false'
      IS_DATA_NODE: 'false'
      IS_RESOURCE_MANAGER: 'true'
      CLUSTER_NODES: namenode resourcemanager worker1 worker2 worker3

  worker1:
    image: renci/hadoop:2.9.0
    depends_on:
      - namenode
    container_name: worker1
    volumes:
      - hadoop-public:/home/hadoop/public
      - ./site-files:/site-files
    restart: always
    hostname: worker1
    networks:
      - hadoop
    ports:
      - '8042:8042'
      - '50075:50075'
    environment:
      IS_NODE_MANAGER: 'true'
      IS_NAME_NODE: 'false'
      IS_SECONDARY_NAME_NODE: 'false'
      IS_DATA_NODE: 'true'
      IS_RESOURCE_MANAGER: 'false'
      CLUSTER_NODES: namenode resourcemanager worker1 worker2 worker3

  worker2:
    image: renci/hadoop:2.9.0
    depends_on:
      - namenode
    container_name: worker2
    volumes:
      - hadoop-public:/home/hadoop/public
      - ./site-files:/site-files
    restart: always
    hostname: worker2
    networks:
      - hadoop
    ports:
      - '8043:8042'
      - '50076:50075'
    environment:
      IS_NODE_MANAGER: 'true'
      IS_NAME_NODE: 'false'
      IS_SECONDARY_NAME_NODE: 'false'
      IS_DATA_NODE: 'true'
      IS_RESOURCE_MANAGER: 'false'
      CLUSTER_NODES: namenode resourcemanager worker1 worker2 worker3

  worker3:
    image: renci/hadoop:2.9.0
    depends_on:
      - namenode
    container_name: worker3
    volumes:
      - hadoop-public:/home/hadoop/public
      - ./site-files:/site-files
    restart: always
    hostname: worker3
    networks:
      - hadoop
    ports:
      - '8044:8042'
      - '50077:50075'
    environment:
      IS_NODE_MANAGER: 'true'
      IS_NAME_NODE: 'false'
      IS_SECONDARY_NAME_NODE: 'false'
      IS_DATA_NODE: 'true'
      IS_RESOURCE_MANAGER: 'false'
      CLUSTER_NODES: namenode resourcemanager worker1 worker2 worker3

volumes:
  hadoop-public:

networks:
  hadoop: