docker-compose.yml

services:
  # Background data collection service
  collector:
    build:
      context: ./collector
      dockerfile: Dockerfile
    #user: "0:0"  # Run as root
    command: ["python", "-m", "src.collector_main"]
    environment:
      - METRICS_BACKEND=${METRICS_BACKEND:-influxdb}
      - INFLUXDB_URL=http://influxdb:8086
      - INFLUXDB_TOKEN=${INFLUXDB_TOKEN}
      - INFLUXDB_ORG=${INFLUXDB_ORG:-prometheus}
      - INFLUXDB_BUCKET=${INFLUXDB_BUCKET:-gpu_metrics}
      # Performance tuning for 300+ endpoint scale
      - INFLUXDB_BATCH_SIZE=${INFLUXDB_BATCH_SIZE:-0}  # 0 = use config.yaml default (10000)
      - INFLUXDB_WRITE_TIMEOUT_MS=${INFLUXDB_WRITE_TIMEOUT_MS:-0}  # 0 = use config.yaml default (90000 = 90s)
      - INFLUXDB_MAX_CONCURRENT_WRITES=${INFLUXDB_MAX_CONCURRENT_WRITES:-0}  # 0 = use config.yaml default (5)
      - INFLUXDB_FLUSH_INTERVAL_SECONDS=${INFLUXDB_FLUSH_INTERVAL_SECONDS:-0}  # 0 = use config.yaml default (10s)
      - CONFIG_PATH=/app/config/config.yaml
      - SCHEMA_PATH=/app/config/metrics_schema.yaml
      - DATABASE_URL=sqlite:///data/targets.db
      - ENCRYPTION_KEY=${ENCRYPTION_KEY}
      # Alert/Webhook settings (for testing webhook subscriptions)
      - ALERT_FORCE_WEBHOOK_MODE=${ALERT_FORCE_WEBHOOK_MODE:-false}
    volumes:
      - shared_data:/app/data
      - ./collector/config:/app/config:ro
    depends_on:
      influxdb:
        condition: service_healthy
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8081/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 30s
    restart: unless-stopped
    # Resource limits scaled for 300+ endpoints
    # For smaller deployments (< 50 endpoints), can reduce to mem_limit: 2g, cpus: 2.
    # JSON/JSONPath parsing is GIL-bound and CPU-heavy; under-provisioned CPUs
    # starve the asyncio loop and produce silent flush-loop stalls.
    mem_limit: 8g
    cpus: 4
    networks:
      - monitoring
    # Note: Port 8081 is only exposed internally for health checks, not to host

  # Web UI and API service
  api:
    build:
      context: ./collector
      dockerfile: Dockerfile
    #user: "0:0"  # Run as root
    command: ["python", "-m", "src.api_main"]
    ports:
      - "${API_PORT:-8080}:8080"
    environment:
      - CONFIG_PATH=/app/config/config.yaml
      - SCHEMA_PATH=/app/config/metrics_schema.yaml
      - DATABASE_URL=sqlite:///data/targets.db
      - ENCRYPTION_KEY=${ENCRYPTION_KEY}
      - GRAFANA_PORT=${GRAFANA_PORT:-3000}
      - INFLUXDB_PORT=${INFLUXDB_PORT:-8086}
      - PROMETHEUS_PORT=${PROMETHEUS_PORT:-9090}
    volumes:
      - shared_data:/app/data
      - ./collector/config:/app/config:ro
    depends_on:
      - collector
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/health"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 10s
    restart: unless-stopped
    mem_limit: 2g
    cpus: 0.5
    networks:
      - monitoring

  influxdb:
    image: influxdb:2.7.11
    #user: "0:0"  # Run as root
    ports:
      - "${INFLUXDB_PORT:-8086}:8086"
    environment:
      #- DOCKER_INFLUXDB_INIT_MODE=setup
      - DOCKER_INFLUXDB_INIT_USERNAME=${INFLUXDB_ADMIN_USER:-admin}
      - DOCKER_INFLUXDB_INIT_PASSWORD=${INFLUXDB_ADMIN_PASSWORD}
      - DOCKER_INFLUXDB_INIT_ORG=${INFLUXDB_ORG:-prometheus}
      - DOCKER_INFLUXDB_INIT_BUCKET=${INFLUXDB_BUCKET:-gpu_metrics}
      - DOCKER_INFLUXDB_INIT_RETENTION=${INFLUXDB_RETENTION:-7d}
      - DOCKER_INFLUXDB_INIT_ADMIN_TOKEN=${INFLUXDB_TOKEN}
      - INFLUXDB_TOKEN=${INFLUXDB_TOKEN}
      - INFLUXDB_ORG=${INFLUXDB_ORG:-prometheus}
      - INFLUXDB_BUCKET=${INFLUXDB_BUCKET:-gpu_metrics}
      - INFLUXDB_15M_RETENTION=${INFLUXDB_15M_RETENTION:-30d}
      - INFLUXDB_HOURLY_RETENTION=${INFLUXDB_HOURLY_RETENTION:-90d}
      - INFLUXDB_HTTP_BIND_ADDRESS=:8086
      # Performance tuning for high-volume writes (300+ endpoints)
      - INFLUXDB_DATA_CACHE_MAX_MEMORY_SIZE=4294967296  # 4GB cache (increased from 2GB)
      - INFLUXDB_DATA_QUERY_CONCURRENCY=20  # Increased from 10
      - INFLUXDB_HTTP_MAX_ROW_LIMIT=100000
      - INFLUXDB_HTTP_WRITE_TIMEOUT=90s  # Match collector timeout
      - INFLUXDB_STORAGE_WAL_FSYNC_DELAY=100ms  # Batch WAL writes for better performance
    volumes:
      - influxdb_data:/var/lib/influxdb2
      - influxdb_config:/etc/influxdb2
      # Temporarily disabled init scripts - run manually after startup with:
      # ./gyanam.sh setup-all-downsampling
      - ./influxdb/scripts:/docker-entrypoint-initdb.d:ro
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8086/ping"]
      interval: 10s
      timeout: 5s
      retries: 10
      start_period: 90s
    restart: unless-stopped
    mem_limit: 16g
    # Sustained ~1.6K writes/s at 250-target scale + Grafana queries + TSM
    # compaction — InfluxDB needs real CPU to keep flush latency below ~3s.
    cpus: 4
    networks:
      - monitoring

  grafana:
    image: grafana/grafana:13.0.1
    #user: "0:0"  # Run as root
    ports:
      - "${GRAFANA_PORT:-3000}:3000"
    environment:
      - GF_SECURITY_ADMIN_USER=${GRAFANA_ADMIN_USER:-admin}
      - GF_SECURITY_ADMIN_PASSWORD=${GRAFANA_ADMIN_PASSWORD}
      - GF_USERS_ALLOW_SIGN_UP=false
      - INFLUXDB_TOKEN=${INFLUXDB_TOKEN}
      - INFLUXDB_ORG=${INFLUXDB_ORG:-prometheus}
      - INFLUXDB_BUCKET=${INFLUXDB_BUCKET:-gpu_metrics}
    volumes:
      - grafana_data:/var/lib/grafana
      - ./grafana/provisioning:/etc/grafana/provisioning:ro
    depends_on:
      influxdb:
        condition: service_healthy
    healthcheck:
      test: ["CMD-SHELL", "wget --no-verbose --tries=1 --spider http://localhost:3000/api/health || exit 1"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 30s
    restart: unless-stopped
    mem_limit: 4g
    cpus: 0.5
    networks:
      - monitoring

volumes:
  shared_data:
  influxdb_data:
  influxdb_config:
  grafana_data:

networks:
  monitoring:
    driver: bridge