docker-compose.yml

---
# The services of MCloud: STF, the grid and the nginx in front of them (README.md).
# The image tags and the ports come from .env, the settings of the services from variables.env:
# ./zebrunner.sh setup writes both from their *.original files.
networks:
  default:
    name: infra
    external: true

# the logs of every service are rotated: the default json-file driver of docker keeps them all (LOG_MAX_SIZE, LOG_MAX_FILE of .env)
x-logging: &logging
  driver: json-file
  options:
    max-size: "${LOG_MAX_SIZE:-50m}"
    max-file: "${LOG_MAX_FILE:-5}"

# the STF units: one image, the settings of variables.env, started after the database migration
x-stf: &stf
  image: "public.ecr.aws/zebrunner/stf:${TAG_STF}"
  env_file:
    - variables.env
  depends_on:
    stf-migrate:
      condition: service_completed_successfully
  logging: *logging
  restart: always

# the units with an HTTP port: it accepts connections (docker ps shows an unhealthy unit)
x-http-healthcheck: &http-healthcheck
  test: ["CMD", "node", "-e", "require('net').connect(3000, '127.0.0.1').on('connect', () => process.exit(0)).on('error', () => process.exit(1))"]
  interval: 10s
  timeout: 5s
  retries: 3
  start_period: 60s

# the units without a port: healthy while they are connected (ESTABLISHED) to the ports of their arguments, their
# ZeroMQ brokers and RethinkDB; ZeroMQ reconnects by itself, a unit without its connections is cut off from STF
x-connections-script: &connections-script >-
  const fs = require('fs'); const up = new Set();
  for (const f of ['/proc/net/tcp', '/proc/net/tcp6']) { try {
  for (const l of fs.readFileSync(f, 'utf8').split('\n').slice(1)) {
  const c = l.trim().split(/\s+/); if (c[3] === '01') up.add(parseInt(c[2].split(':').pop(), 16)); } } catch (e) {} }
  process.exit(process.argv.slice(1).every((p) => up.has(Number(p))) ? 0 : 1)
x-connections-healthcheck: &connections-healthcheck
  interval: 10s
  timeout: 5s
  retries: 3
  start_period: 60s

services:
  stf-proxy:
    image: "nginx:${TAG_NGINX}"
    container_name: stf-proxy
    volumes:
      - ./configuration/stf-proxy/nginx.conf:/etc/nginx/nginx.conf:ro
      - ./configuration/stf-proxy/ssl:/etc/nginx/ssl:ro
      - ./configuration/stf-proxy/htpasswd:/etc/nginx/htpasswd:ro
      - ./configuration/stf-proxy/grid:/usr/share/nginx/html/grid:ro
    ports:
      - "${STF_PORT}:80" # primary HTTP port
    depends_on:
      stf-app:
        condition: service_healthy
    healthcheck:
      test: ["CMD", "bash", "-c", "echo > /dev/tcp/127.0.0.1/80"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 30s
    logging: *logging
    restart: always
  # the TCP relay of the remote debug of STF (adb connect) for the nodes in a private network: ./zebrunner.sh starts it
  # with the profile adb-relay when ADB_RELAY of .env is true (README.md#remote-debug-through-the-adb-relay)
  adb-relay:
    image: "nginx:${TAG_NGINX}"
    container_name: adb-relay
    profiles: [adb-relay]
    # it listens on the remote debug ports of the devices on every address of the server, connects to the nodes and
    # reads the devices of the nodes from stf-proxy on the port of the host: the network of the host instead of published
    # ports
    network_mode: host
    # nginx and the sync of the devices of STF
    entrypoint: ["bash", "/opt/adb-relay/adb-relay.sh"]
    environment:
      - ADB_RELAY=${ADB_RELAY:-false}
      - ADB_RELAY_PORTS=${ADB_RELAY_PORTS:-}
      - ADB_RELAY_SYNC_SECONDS=${ADB_RELAY_SYNC_SECONDS:-300}
      - STF_URL=${STF_URL}
      - STF_PORT=${STF_PORT}
    volumes:
      - ./configuration/adb-relay/nginx.conf:/etc/nginx/nginx.conf:ro
      # adb-relay.sh, its awk programs and stf-token (STF_TOKEN of variables.env only, written by ./zebrunner.sh)
      - ./configuration/adb-relay:/opt/adb-relay:ro
    healthcheck:
      # nginx writes its pid once it listens on every port of the ranges
      test: ["CMD", "sh", "-c", "kill -0 $$(cat /run/nginx.pid)"]
      interval: 10s
      timeout: 5s
      retries: 3
      start_period: 30s
    logging: *logging
    restart: always
  mcloud-grid:
    image: "public.ecr.aws/zebrunner/mcloud-grid:${TAG_SELENIUM_HUB}"
    container_name: mcloud-grid
    env_file:
      - variables.env
    environment:
      - STF_URL=http://stf-api:3000
      # 10 min session wait timeout as 20 min is configured on NGiNX
      - GRID_NEW_SESSION_WAIT_TIMEOUT=600000
      - CHECK_APPIUM_STATUS=true
      # the details of the grid in its JSON logs
      # - MCLOUD_LOG_LEVEL=FINE
    # the nodes register on AGENTS_ADDRESS:4446, the grid of the join string of ./zebrunner.sh node-invite
    ports:
      - "${AGENTS_ADDRESS}:4446:4444"
    # the healthcheck is the one of the image (/wd/hub/status)
    logging: *logging
    restart: always
  rethinkdb:
    image: "rethinkdb:${TAG_RETHINKDB}"
    container_name: rethinkdb
    volumes:
      - "stf-db-volume:/data"
    command: rethinkdb --bind all --cache-size 8192 --no-update-check --no-http-admin
    healthcheck:
      test: ["CMD", "bash", "-c", "echo > /dev/tcp/127.0.0.1/28015"]
      interval: 2s
      timeout: 5s
      retries: 30
    logging: *logging
    restart: always
  # creates and updates the tables (e.g. the sessions of the sign-ins) before the other units start
  stf-migrate:
    image: "public.ecr.aws/zebrunner/stf:${TAG_STF}"
    container_name: stf-migrate
    env_file:
      - variables.env
    command: stf migrate
    depends_on:
      rethinkdb:
        condition: service_healthy
    logging: *logging
  stf-app:
    <<: *stf
    container_name: stf-app
    healthcheck: *http-healthcheck
    command: stf app --port 3000 --auth-url ${STF_URL}/${AUTH_URL} --websocket-url ${STF_URL}/
  stf-auth:
    <<: *stf
    container_name: stf-auth
    healthcheck: *http-healthcheck
    # ${STF_URL}/stf/ with slash at the end produce invalid redirect! /stf/?jwt=e...
    command: stf ${AUTH_SYSTEM} --port 3000 --app-url ${STF_URL}/stf
  stf-processor001:
    <<: *stf
    container_name: stf-processor001
    healthcheck:
      <<: *connections-healthcheck
      test: ["CMD", "node", "-e", *connections-script, "7160", "7260", "28015"]
    command: stf processor proc001 --connect-app-dealer tcp://stf-triproxy-app:7160 --connect-dev-dealer tcp://stf-triproxy-dev:7260
  stf-reaper:
    <<: *stf
    container_name: stf-reaper
    healthcheck:
      <<: *connections-healthcheck
      test: ["CMD", "node", "-e", *connections-script, "7150", "7270", "28015"]
    command: stf reaper reaper001 --connect-push tcp://stf-triproxy-dev:7270 --connect-sub tcp://stf-triproxy-app:7150 --heartbeat-timeout 30000
  stf-storage-plugin-apk:
    <<: *stf
    container_name: stf-storage-plugin-apk
    healthcheck: *http-healthcheck
    command: stf storage-plugin-apk --port 3000 --storage-url ${STF_URL}/
  stf-storage-plugin-image:
    <<: *stf
    container_name: stf-storage-plugin-image
    healthcheck: *http-healthcheck
    command: stf storage-plugin-image --port 3000 --storage-url ${STF_URL}/
  # SECRET and STF_STORAGE_TOKEN of variables.env protect the storage routes
  stf-storage-temp:
    <<: *stf
    container_name: stf-storage-temp
    healthcheck: *http-healthcheck
    volumes:
      - "stf-temp-volume:/data"
    command: stf storage-temp --port 3000 --save-dir /data
  stf-triproxy-app:
    <<: *stf
    container_name: stf-triproxy-app
    healthcheck:
      <<: *http-healthcheck
      test: ["CMD", "node", "-e", "require('net').connect(7160, '127.0.0.1').on('connect', () => process.exit(0)).on('error', () => process.exit(1))"]
    command: stf triproxy app001 --bind-pub "tcp://*:7150" --bind-dealer "tcp://*:7160" --bind-pull "tcp://*:7170"
    # the nodes connect to the dealer only; their ZeroMQ ports are on AGENTS_ADDRESS, the host of the join string of
    # ./zebrunner.sh node-invite
    ports:
      - "${AGENTS_ADDRESS}:7160:7160"
  stf-triproxy-dev:
    <<: *stf
    container_name: stf-triproxy-dev
    healthcheck:
      <<: *http-healthcheck
      test: ["CMD", "node", "-e", "require('net').connect(7260, '127.0.0.1').on('connect', () => process.exit(0)).on('error', () => process.exit(1))"]
    command: stf triproxy dev001 --bind-pub "tcp://*:7250" --bind-dealer "tcp://*:7260" --bind-pull "tcp://*:7270"
    ports:
      - "${AGENTS_ADDRESS}:7250:7250"
      - "${AGENTS_ADDRESS}:7260:7260"
      - "${AGENTS_ADDRESS}:7270:7270"
  stf-websocket:
    <<: *stf
    container_name: stf-websocket
    healthcheck: *http-healthcheck
    command: >-
      stf websocket --port 3000 --storage-url ${STF_URL}/
      --connect-sub tcp://stf-triproxy-app:7150 --connect-push tcp://stf-triproxy-app:7170
  stf-api:
    <<: *stf
    container_name: stf-api
    healthcheck: *http-healthcheck
    # --nodes-connect-*: the ZeroMQ endpoints of the answer of the enrollment, the published ports of the triproxies at
    # NODES_HOST (the host of the join string); --nodes-stf-url is STF_URL of variables.env, the public URL
    command: >-
      stf api --port 3000
      --connect-sub tcp://stf-triproxy-app:7150 --connect-push tcp://stf-triproxy-app:7170
      --connect-sub-dev tcp://stf-triproxy-dev:7250 --connect-push-dev tcp://stf-triproxy-dev:7270
      --nodes-connect-sub tcp://${NODES_HOST}:7250 --nodes-connect-push tcp://${NODES_HOST}:7270
      --nodes-connect-app-dealer tcp://${NODES_HOST}:7160 --nodes-connect-dev-dealer tcp://${NODES_HOST}:7260
  stf-api-groups-engine:
    <<: *stf
    container_name: stf-api-groups-engine
    healthcheck:
      <<: *connections-healthcheck
      test: ["CMD", "node", "-e", *connections-script, "7150", "7170", "7250", "7270", "28015"]
    command: >-
      stf groups-engine
      --connect-sub tcp://stf-triproxy-app:7150 --connect-push tcp://stf-triproxy-app:7170
      --connect-sub-dev tcp://stf-triproxy-dev:7250 --connect-push-dev tcp://stf-triproxy-dev:7270
volumes:
  stf-db-volume:
  stf-temp-volume: