docker-compose.yml

version: '3.8'

services:
  cratedb:
    image: crate:latest
    container_name: crate-db
    ports:
      - "4200:4200"
    networks:
      - dp_network
      - m_network
      - ls_network
      - deploy_network
    volumes:
      - ${DB_DATA_PATH}:/data
      - ${DB_DATA_PATH}/repos:/repos
    command: [ "crate", "-Cnetwork.host=_site_" , "-Cpath.repo=/repos"]
    deploy:
      replicas: 1
      restart_policy:
        condition: on-failure
    environment:
      - CRATE_HEAP_SIZE=2g

  ls_web:
    profiles:
      - label
      - dev
    restart: on-failure
    image: heartexlabs/label-studio:latest
    container_name: label-studio-web
    ports:
      - "8080:8080"
    networks:
      - ls_network
    volumes:
      - ./label-studio/config:/config
      - ${LS_DATA_PATH}:/label-studio/data
      - ${LS_TASKS_PATH}:/tasks
      - ${LS_ANNOTATIONS_PATH}:/annotations
    environment:
      - LABEL_STUDIO_LOCAL_FILES_SERVING_ENABLED=true
      - LABEL_STUDIO_LOCAL_FILES_DOCUMENT_ROOT=/label-studio/files
    command: [ "label-studio", "start" ]

  ls:
    profiles:
      - label
      - dev
    restart: on-failure
    image: nullberry/adapt-ml-ls:latest
    container_name: label-studio-dev
    networks:
      - ls_network
    volumes:
      - ./label-studio/ls:/code/ls
      - ${LS_DATA_PATH}:/label-studio/data
      - ${LS_TASKS_PATH}:/tasks
      - ${LS_ANNOTATIONS_PATH}:/annotations
      - ./test:/test
    environment:
      - DATABASE_IP=${DATABASE_IP}
    stdin_open: true
    tty: true

  dp_db:
    profiles:
      - data_programming
      - dev
    restart: on-failure
    image: mysql/mysql-server:8.0.26
    container_name: dp-mlflow-db
    expose:
      - 3306
    networks:
      - dp_network
    volumes:
      - ${DP_DATA_PATH}/dbdata:/var/lib/mysql
    environment:
      - MYSQL_DATABASE=${DP_MYSQL_DATABASE}
      - MYSQL_USER=${DP_MYSQL_USER}
      - MYSQL_PASSWORD=${DP_MYSQL_PASSWORD}
      - MYSQL_ROOT_PASSWORD=${DP_MYSQL_ROOT_PASSWORD}

  dp_web:
    profiles:
      - dev
    restart: on-failure
    image: nullberry/adapt-ml-mlflow-server:latest
    container_name: dp-mlflow-server
    ports:
      - "5000:5000"
    networks:
      - dp_network
    depends_on:
      - dp_db
    volumes:
      - ${DP_DATA_PATH}/mlruns:/mlflow/mlruns
    environment:
      - MLFLOW_TRACKING_URI=${DP_MLFLOW_TRACKING_URI}
    command: [ "mlflow",
               "server",
               "--backend-store-uri=$DP_MLFLOW_TRACKING_URI",
               "--default-artifact-root=./mlruns",
               "--host=0.0.0.0"]

  dp:
    profiles:
      - data_programming
      - dev
    image: nullberry/adapt-ml-dp:latest
    container_name: dp-mlflow
    networks:
      - dp_network
    depends_on:
      - dp_db
    volumes:
      - ${LS_ANNOTATIONS_PATH}:/annotations
      - ${DP_DATA_PATH}/mlruns:/code/mlruns
      - ${DP_DATA_PATH}/unlabeled_data:/unlabeled_data
      - ${DP_DATA_PATH}/tmp_artifacts:/tmp_artifacts
      - ./data-programming/label:/code/label
      - ./data-programming/MLproject:/code/MLproject
      - ./data-programming/conda.yaml:/code/conda.yaml
      - ./test:/test
    environment:
      - MLFLOW_TRACKING_URI=${DP_MLFLOW_TRACKING_URI}
      - DATABASE_IP=${DATABASE_IP}
    stdin_open: true
    tty: true

  m_db:
    profiles:
      - modelling
      - deploy
      - dev
    restart: on-failure
    image: mysql/mysql-server:8.0.26
    container_name: modelling-mlflow-db
    expose:
      - 3306
    networks:
      - m_network
    volumes:
      - ${MODELLING_DATA_PATH}/dbdata:/var/lib/mysql
    environment:
      - MYSQL_DATABASE=${MODELLING_MYSQL_DATABASE}
      - MYSQL_USER=${MODELLING_MYSQL_USER}
      - MYSQL_PASSWORD=${MODELLING_MYSQL_PASSWORD}
      - MYSQL_ROOT_PASSWORD=${MODELLING_MYSQL_ROOT_PASSWORD}

  m_web:
    profiles:
      - dev
    restart: on-failure
    image: nullberry/adapt-ml-mlflow-server:latest
    container_name: modelling-mlflow-server
    ports:
      - "5001:5000"
    networks:
      - m_network
    depends_on:
      - m_db
    volumes:
      - ${MODELLING_DATA_PATH}/mlruns:/mlflow/mlruns
    environment:
      - MLFLOW_TRACKING_URI=${MODELLING_MLFLOW_TRACKING_URI}
    command: [ "mlflow",
               "server",
               "--backend-store-uri=$MODELLING_MLFLOW_TRACKING_URI",
               "--default-artifact-root=./mlruns",
               "--host=0.0.0.0" ]

  m:
    profiles:
      - modelling
      - deploy
      - dev
    image: nullberry/adapt-ml-m:latest
    container_name: modelling-mlflow
    networks:
      - m_network
    depends_on:
      - m_db
    volumes:
      - ${MODELLING_DATA_PATH}/mlruns:/code/mlruns
      - ${DP_DATA_PATH}/mlruns:/dp_mlruns
      - ${MODELLING_DATA_PATH}/tmp_artifacts:/tmp_artifacts
      - ./modelling/model:/code/model
      - ./modelling/model_objs.py:/code/model_objs.py
      - ./modelling/MLproject:/code/MLproject
      - ./modelling/conda.yaml:/code/conda.yaml
      - ./test:/test
    environment:
      - MLFLOW_TRACKING_URI=${MODELLING_MLFLOW_TRACKING_URI}
      - DATABASE_IP=${DATABASE_IP}
    stdin_open: true
    tty: true

  m_deploy:
    profiles:
      - deploy
      - dev
    image: nullberry/adapt-ml-m-deploy:latest
    container_name: modelling-mlflow-deploy
    ports:
      - "80:80"
    networks:
      - deploy_network
    volumes:
      - ${MODELLING_DATA_PATH}/mlruns:/mlruns
      - ./modelling/app:/app
      - ./modelling/model_objs.py:/app/model_objs.py
    environment:
      - DATABASE_IP=${DATABASE_IP}
      - MULTICLASS_EXAMPLE_MODEL_PATH=${MULTICLASS_EXAMPLE_MODEL_PATH}
      - MULTILABEL_EXAMPLE_MODEL_PATH=${MULTILABEL_EXAMPLE_MODEL_PATH}
      - CLIMATE_FRAMES_MODEL_PATH=${CLIMATE_FRAMES_MODEL_PATH}

networks:
  ls_network:
    driver: bridge
  dp_network:
    driver: bridge
  m_network:
    driver: bridge
  deploy_network:
    driver: bridge