compose.production.yaml

# Production compose — the recipe published for self-hosters (M20.2 §3.2),
# and the one our own box runs. Sail's compose.yaml is dev-only.
#
#   docker compose -f compose.production.yaml up -d
#
# Expects a host .env alongside this file (never baked into the image),
# TLS certs under /etc/letsencrypt, and an ACME webroot at ./certbot-webroot.
# See docs/ops/get-live/ for the full walkthrough.

x-logging: &logging
  driver: json-file
  options:
    max-size: "10m"
    max-file: "3"

x-app: &app
  image: ghcr.io/n8bar/cryptozing-app:${CZ_TAG:-latest}
  restart: unless-stopped
  # A real init as PID 1. schedule:work and queue:work are not inits: children
  # that finish get reparented to them and are never reaped, and the zombies
  # count against the container user's process limit on the host.
  init: true
  env_file: .env
  logging: *logging
  volumes:
    - ./storage:/var/www/html/storage
  depends_on:
    db:
      condition: service_healthy

services:
  app:
    <<: *app
    volumes:
      - ./storage:/var/www/html/storage
      - app-public:/shared/public

  web:
    image: nginx:1.27-alpine
    restart: unless-stopped
    # Only the template variable — no app secrets in this container.
    environment:
      CZ_SERVER_NAME: ${CZ_SERVER_NAME}
    logging: *logging
    ports:
      - "80:80"
      - "443:443"
    volumes:
      - ./docker/production/nginx/templates:/etc/nginx/templates:ro
      - app-public:/var/www/html/public:ro
      - /etc/letsencrypt:/etc/letsencrypt:ro
      - ./certbot-webroot:/var/www/certbot:ro
    depends_on:
      - app

  db:
    image: mysql:8.0
    restart: unless-stopped
    logging: *logging
    # Internal network only — no published ports.
    environment:
      MYSQL_DATABASE: ${DB_DATABASE}
      MYSQL_USER: ${DB_USERNAME}
      MYSQL_PASSWORD: ${DB_PASSWORD}
      MYSQL_ROOT_PASSWORD: ${DB_ROOT_PASSWORD}
    volumes:
      - dbdata:/var/lib/mysql
      - ./docker/production/mysql/prod.cnf:/etc/mysql/conf.d/prod.cnf:ro
    healthcheck:
      test: ["CMD-SHELL", "mysqladmin ping -h127.0.0.1 -u\"$$MYSQL_USER\" -p\"$$MYSQL_PASSWORD\" --silent"]
      interval: 15s
      timeout: 5s
      retries: 5
      start_period: 60s

  # Both workers run the app image without PHP-FPM, so the image's FPM ping
  # cannot speak for them (#153). Each refreshes a heartbeat as its loop turns
  # and answers on the age of it — a wedged worker goes quiet while its process
  # stays up, which is the failure the restart policy alone never catches.
  queue:
    <<: *app
    command: php artisan queue:work --tries=3 --timeout=90 --max-time=3600
    healthcheck:
      # The loop turns every few seconds, idle or not.
      test: ["CMD", "php", "/var/www/html/docker/production/worker-healthcheck.php", "queue", "60"]
      interval: 30s
      timeout: 5s
      retries: 3
      start_period: 30s

  # Also covers the payment watcher: wallet:watch-payments runs as a
  # per-minute scheduled command, not an always-on process (§2.3.1.2).
  scheduler:
    <<: *app
    command: php artisan schedule:work
    healthcheck:
      # One tick a minute, and the first waits for the top of the next one.
      test: ["CMD", "php", "/var/www/html/docker/production/worker-healthcheck.php", "scheduler", "180"]
      interval: 60s
      timeout: 10s
      retries: 3
      start_period: 120s

volumes:
  dbdata:
  app-public: