docker-compose_cluster.yml

services:
  # Master
  master-node:
    image: "jwaresolutions/big-data-cluster:0.5.0"
    command: bash -c "/home/big_data/spark-cmd.sh start master-node"
    ports:
      - target: 8080
        published: 8080
        protocol: tcp
        mode: host
      - target: 9870
        published: 9870
        protocol: tcp
        mode: host
      - target: 18080
        published: 18080
        protocol: tcp
        mode: host
    networks:
      - cluster-net
    volumes:
      # - "./data:/home/big_data/data" # Your data
      - hdfs-master-data:/home/hadoop/data/nameNode
      - hdfs-master-checkpoint-data:/home/hadoop/data/namesecondary
    deploy:
      mode: global # Required by Docker Swarm to make published ports work with other services
      endpoint_mode: dnsrr # Required to prevent java.net.ConnectException
      placement:
        # Set node labels using `docker node update --label-add role=master <NODE ID>` from swarm manager
        constraints:
          - node.labels.role==master

  # Workers
  worker:
    image: "jwaresolutions/big-data-cluster:0.5.0"
    command: bash -c "/home/big_data/spark-cmd.sh start"
    depends_on:
      - "master-node"
    volumes:
      - hdfs-worker-data:/home/hadoop/data/dataNode
    deploy:
      placement:
        # Set node labels using `docker node update --label-add role=worker <NODE ID>` from swarm manager
        constraints:
          - node.labels.role==worker
      # Deploy N containers for this service
      replicas: 3
    networks:
      - cluster-net

volumes:
  hdfs-master-data:
    external: true
    name: 'hdsf_master_data_swarm'
  hdfs-master-checkpoint-data:
    external: true
    name: 'hdsf_master_checkpoint_data_swarm'
  hdfs-worker-data:
    external: true
    name: 'hdsf_worker_data_swarm'

# Uses cluster-net network
networks:
  cluster-net:
    external: true
    name: cluster_net_swarm