docker-compose.wave123.yml

version: "3.9"
# ─── PayGate Wave 123 — AI Model Registry, Menu Management & Portal Health ────
#
# This overlay adds the infrastructure required by Wave 123 features:
#   1. MLflow Model Registry     — tracks AI model versions, metrics, artifacts
#   2. MinIO Object Storage      — stores model artifacts and training datasets
#   3. Feast Feature Store       — serves ML features for GNN inference
#   4. Menu CDN Cache (Nginx)    — caches menu API responses at the edge
#   5. Uptime Kuma               — portal health monitoring dashboard
#   6. OpenTelemetry Collector   — aggregates traces/metrics from all services
#
# Usage:
#   docker compose -f docker-compose.yml -f docker-compose.wave123.yml up -d
#
# Required env vars (add to .env):
#   MLFLOW_TRACKING_URI, MINIO_ROOT_USER, MINIO_ROOT_PASSWORD,
#   FEAST_REDIS_URL, OTEL_EXPORTER_OTLP_ENDPOINT

services:
  # ── MLflow Model Registry ──────────────────────────────────────────────────
  mlflow:
    image: ghcr.io/mlflow/mlflow:v2.12.2
    container_name: paygate_mlflow
    restart: unless-stopped
    command: >
      mlflow server
        --backend-store-uri postgresql://paygate:${POSTGRES_PASSWORD:-paygate_secure_2026}@postgres:5432/paygate_mlflow
        --default-artifact-root s3://paygate-models/mlflow
        --host 0.0.0.0
        --port 5000
        --workers 2
    environment:
      MLFLOW_S3_ENDPOINT_URL: http://minio:9000
      AWS_ACCESS_KEY_ID: ${MINIO_ROOT_USER:-paygate_minio}
      AWS_SECRET_ACCESS_KEY: ${MINIO_ROOT_PASSWORD:-paygate_minio_secret_2026}
    ports:
      - "5000:5000"
    depends_on:
      postgres:
        condition: service_healthy
      minio:
        condition: service_healthy
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:5000/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 30s
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "mlflow"

  # ── MinIO Object Storage (model artifacts + training datasets) ─────────────
  minio:
    image: minio/minio:RELEASE.2024-04-18T19-09-19Z
    container_name: paygate_minio
    restart: unless-stopped
    command: server /data --console-address ":9001"
    environment:
      MINIO_ROOT_USER: ${MINIO_ROOT_USER:-paygate_minio}
      MINIO_ROOT_PASSWORD: ${MINIO_ROOT_PASSWORD:-paygate_minio_secret_2026}
      MINIO_SITE_NAME: paygate-model-store
    ports:
      - "9000:9000"
      - "9001:9001"
    volumes:
      - minio_data:/data
    healthcheck:
      test: ["CMD", "mc", "ready", "local"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 20s
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "minio"

  # ── MinIO Bucket Initializer ───────────────────────────────────────────────
  minio-init:
    image: minio/mc:RELEASE.2024-04-18T16-45-29Z
    container_name: paygate_minio_init
    depends_on:
      minio:
        condition: service_healthy
    entrypoint: >
      /bin/sh -c "
        mc alias set local http://minio:9000 ${MINIO_ROOT_USER:-paygate_minio} ${MINIO_ROOT_PASSWORD:-paygate_minio_secret_2026};
        mc mb --ignore-existing local/paygate-models;
        mc mb --ignore-existing local/paygate-datasets;
        mc mb --ignore-existing local/paygate-menu-assets;
        mc policy set public local/paygate-menu-assets;
        echo 'MinIO buckets initialized';
      "
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "minio-init"

  # ── Feast Feature Store (online features for GNN inference) ───────────────
  feast-online:
    image: feastdev/feature-server:0.40.0
    container_name: paygate_feast
    restart: unless-stopped
    command: serve --host 0.0.0.0 --port 6566
    environment:
      FEAST_REDIS_HOST: ${REDIS_HOST:-redis}
      FEAST_REDIS_PORT: 6379
      FEAST_REDIS_PASSWORD: ${REDIS_PASSWORD:-}
    ports:
      - "6566:6566"
    depends_on:
      - redis
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:6566/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 20s
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "feast"

  # ── Menu CDN Cache (Nginx edge cache for menu API) ─────────────────────────
  menu-cdn:
    image: nginx:1.25-alpine
    container_name: paygate_menu_cdn
    restart: unless-stopped
    ports:
      - "8090:80"
    volumes:
      - ./infra/nginx/menu-cdn.conf:/etc/nginx/conf.d/default.conf:ro
      - menu_cache:/var/cache/nginx
    depends_on:
      - app
    healthcheck:
      test: ["CMD", "wget", "-qO-", "http://localhost:80/nginx-health"]
      interval: 30s
      timeout: 5s
      retries: 3
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "menu-cdn"

  # ── Uptime Kuma (Portal Health Monitoring) ─────────────────────────────────
  uptime-kuma:
    image: louislam/uptime-kuma:1.23.13
    container_name: paygate_uptime_kuma
    restart: unless-stopped
    ports:
      - "3001:3001"
    volumes:
      - uptime_kuma_data:/app/data
    healthcheck:
      test: ["CMD", "wget", "-qO-", "http://localhost:3001/api/entry-page"]
      interval: 60s
      timeout: 10s
      retries: 3
      start_period: 30s
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "uptime-kuma"

  # ── OpenTelemetry Collector (traces + metrics aggregation) ─────────────────
  otel-collector:
    image: otel/opentelemetry-collector-contrib:0.99.0
    container_name: paygate_otel_collector
    restart: unless-stopped
    command: ["--config=/etc/otel-collector-config.yaml"]
    volumes:
      - ./infra/otel/collector-config.yaml:/etc/otel-collector-config.yaml:ro
    ports:
      - "4317:4317"   # gRPC OTLP receiver
      - "4318:4318"   # HTTP OTLP receiver
      - "8888:8888"   # Prometheus metrics
      - "8889:8889"   # Prometheus exporter
    depends_on:
      - app
    healthcheck:
      test: ["CMD", "wget", "-qO-", "http://localhost:13133/"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 20s
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "otel-collector"

  # ── GNN Training Worker (Python, GPU-optional) ────────────────────────────
  gnn-worker:
    image: paygate/gnn-worker:latest
    build:
      context: ./ml
      dockerfile: Dockerfile.gnn
    container_name: paygate_gnn_worker
    restart: unless-stopped
    environment:
      MLFLOW_TRACKING_URI: http://mlflow:5000
      MLFLOW_S3_ENDPOINT_URL: http://minio:9000
      AWS_ACCESS_KEY_ID: ${MINIO_ROOT_USER:-paygate_minio}
      AWS_SECRET_ACCESS_KEY: ${MINIO_ROOT_PASSWORD:-paygate_minio_secret_2026}
      REDIS_URL: redis://redis:6379
      FEAST_ENDPOINT: http://feast-online:6566
      LOG_LEVEL: ${LOG_LEVEL:-info}
      OTEL_EXPORTER_OTLP_ENDPOINT: http://otel-collector:4318
    depends_on:
      - mlflow
      - minio
      - feast-online
      - redis
    deploy:
      resources:
        reservations:
          devices:
            - driver: nvidia
              count: all
              capabilities: [gpu]
    networks:
      - paygate_net
    labels:
      com.paygate.wave: "123"
      com.paygate.service: "gnn-worker"
    profiles:
      - ml  # Only start with: docker compose --profile ml up

# ── Volumes ───────────────────────────────────────────────────────────────────
volumes:
  minio_data:
    driver: local
    labels:
      com.paygate.wave: "123"
  menu_cache:
    driver: local
    labels:
      com.paygate.wave: "123"
  uptime_kuma_data:
    driver: local
    labels:
      com.paygate.wave: "123"

# ── Networks ──────────────────────────────────────────────────────────────────
# Reuse the existing paygate_net from docker-compose.yml
networks:
  paygate_net:
    external: true