docker-compose-sparql.yml

version: "2"

services:
  #debugger:
  #  image: sjourdan/toolbox
  #  container_name: debugger
  #  networks:
  #    - squirrel:latest
  #  dns: 8.8.8.8
  #  command: nc -l 50070

  frontier:
    image: dicegroup/squirrel.frontier:latest
    container_name: frontier
    environment:
      - HOBBIT_RABBIT_HOST=rabbit
      - URI_WHITELIST_FILE=/var/squirrel/whitelist.txt
      - FRONTIER_CONTEXT_CONFIG_FILE=/var/squirrel/spring-config/frontier-context-sparql.xml
      - SEED_FILE=/var/squirrel/seeds.txt
      - SPARQL_URL=http://virtuosohost:8890/sparql-auth/
      - SPARQL_HOST_USER=dba
      - SPARQL_HOST_PASSWD=pw123
      - MDB_HOST_NAME=mongodb
      - MDB_PORT=27017
      - MDB_CONNECTION_TIME_OUT=5000
      - MDB_SOCKET_TIME_OUT=10000
      - MDB_SERVER_TIME_OUT=10000
      - QUEUE_FILTER_PERSIST=true
      - COMMUNICATION_WITH_WEBSERVICE=false
      - VISUALIZATION_OF_CRAWLED_GRAPH=false
      - JVM_ARGS=-Xmx8g 
      
    volumes:
      - ./data/frontier:/var/squirrel/data
      - ./seed/seeds.csv:/var/squirrel/seeds.csv:ro
      - ./seed/seeds.txt:/var/squirrel/seeds.txt:ro
      - ./whitelist/whitelist.txt:/var/squirrel/whitelist.txt:ro

  virtuosohost:
    image: openlink/virtuoso-opensource-7:latest
    container_name: virtuosohost
    ports:
      - "8890:8890"
    environment:
      - Driver=/usr/local/lib/virtodbc_32.so
      - DBA_PASSWORD=pw123

  #  sparqlhost:
  #    image: stain/jena-fuseki
  #    container_name: sparqlhost
  #    ports:
  #      - "3030:3030"
  #    environment:
  #      - ADMIN_PASSWORD=pw123
  #      - JVM_ARGS=-Xmx2g

  mongodb:
    image: mongo:4.0.0
    volumes:
      - ./data/mongodb:/data
    ports:
      - "27017:27017"

  rethinkdb:
    image: rethinkdb:2.3.5
    volumes:
      - ./data/rethinkdb:/data
    ports:
      - "8080:8080"
    command: rethinkdb --bind all

  # message bus
  rabbit:
    image: rabbitmq:management
    container_name: rabbit
    hostname: rabbit
    ports:
      - "8081:15672"
      # Forwarding the port for testing
      - "5672:5672"

  worker1:
    image: dicegroup/squirrel.worker:latest
    container_name: worker1
    environment:
      - HOBBIT_RABBIT_HOST=rabbit
      - OUTPUT_FOLDER=/var/squirrel/data
      - HTML_SCRAPER_YAML_PATH=/var/squirrel/yaml
      - CONTEXT_CONFIG_FILE=/var/squirrel/spring-config/worker-context-sparql.xml
      - SPARQL_URL=http://virtuosohost:8890/sparql-auth/
      - SPARQL_HOST_USER=dba
      - SPARQL_HOST_PASSWD=pw123
      - DEDUPLICATION_ACTIVE=false
      - JVM_ARGS=-Xmx8g
      - STORE_METADATA=true
    volumes:
      - ./data/worker1:/var/squirrel/data
      - ./yaml:/var/squirrel/yaml
      - ./spring-config:/var/squirrel/spring-config
    command: java -cp squirrel.jar org.dice_research.squirrel.components.WorkerComponentStarter


  worker2:
    image: dicegroup/squirrel.worker:latest
    container_name: worker2
    environment:
      - HOBBIT_RABBIT_HOST=rabbit
      - OUTPUT_FOLDER=/var/squirrel/data
      - HTML_SCRAPER_YAML_PATH=/var/squirrel/yaml
      - CONTEXT_CONFIG_FILE=/var/squirrel/spring-config/worker-context-sparql.xml
      - SPARQL_URL=http://virtuosohost:8890/sparql-auth/
      - SPARQL_HOST_USER=dba
      - SPARQL_HOST_PASSWD=pw123
      - DEDUPLICATION_ACTIVE=false
      - JVM_ARGS=-Xmx8g
      - STORE_METADATA=true
    volumes:
      - ./data/worker2:/var/squirrel/data
      - ./yaml:/var/squirrel/yaml
      - ./spring-config:/var/squirrel/spring-config
    command: java -cp squirrel.jar org.dice_research.squirrel.components.WorkerComponentStarter


  worker3:
    image: dicegroup/squirrel.worker:latest
    container_name: worker3
    environment:
      - HOBBIT_RABBIT_HOST=rabbit
      - OUTPUT_FOLDER=/var/squirrel/data
      - HTML_SCRAPER_YAML_PATH=/var/squirrel/yaml
      - CONTEXT_CONFIG_FILE=/var/squirrel/spring-config/worker-context-sparql.xml
      - SPARQL_URL=http://virtuosohost:8890/sparql-auth/
      - SPARQL_HOST_USER=dba
      - SPARQL_HOST_PASSWD=pw123
      - DEDUPLICATION_ACTIVE=true
      - JVM_ARGS=-Xmx8g
      - STORE_METADATA=true
    volumes:
      - ./data/worker3:/var/squirrel/data
      - ./yaml:/var/squirrel/yaml
      - ./spring-config:/var/squirrel/spring-config
    command: java -cp squirrel.jar org.dice_research.squirrel.components.WorkerComponentStarter


#  deduplicator:
#    image: squirrel.deduplication
#    container_name: deduplicator
#    environment:
#      DEDUPLICATION_ACTIVE: "true"
#      HOBBIT_RABBIT_HOST: rabbit
#      OUTPUT_FOLDER: /var/squirrel/data
#      CONTEXT_CONFIG_FILE: /var/squirrel/spring-config/context-deduplicator.xml
#      MDB_HOST_NAME: mongodb
#      MDB_PORT: 27017
#      SPARQL_HOST_NAME: sparqlhost
#      SPARQL_HOST_PORT: 3030
#      SERVICE_PRECONDITION: "mongodb:27017 rabbit:5672"
#    volumes:
#    - ./data/deduplicator:/var/squirrel/data