BACKBEAT/docker-compose.swarm.yml

version: '3.8'

services:
  # BACKBEAT Pulse Service - Leader-elected tempo broadcaster
  # REQ: BACKBEAT-REQ-001 - Single BeatFrame publisher per cluster
  # REQ: BACKBEAT-OPS-001 - One replica prefers leadership
  backbeat-pulse:
    image: anthonyrawlins/backbeat-pulse:v1.0.4
    command: >
      ./pulse
      -cluster=chorus-production
      -admin-port=8080
      -raft-bind=0.0.0.0:9000
      -data-dir=/data
      -nats=nats://nats:4222
      -tempo=2
      -bar-length=8
      -log-level=info
    environment:
      # REQ: BACKBEAT-OPS-003 - Configuration via environment variables
      - BACKBEAT_CLUSTER_ID=chorus-production
      - BACKBEAT_TEMPO_BPM=2     # 30-second beats for production
      - BACKBEAT_BAR_LENGTH=8    # 4-minute windows
      - BACKBEAT_PHASE_PLAN=plan,work,review
      - BACKBEAT_NATS_URL=nats://nats:4222
      - BACKBEAT_MIN_BPM=1       # 60-second beats minimum
      - BACKBEAT_MAX_BPM=60      # 1-second beats maximum
      - BACKBEAT_LOG_LEVEL=info
    
    # REQ: BACKBEAT-OPS-002 - Health probes for liveness/readiness
    healthcheck:
      test: ["CMD", "wget", "--no-verbose", "--tries=1", "--spider", "http://localhost:8080/healthz"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 60s
    
    deploy:
      replicas: 1              # Single leader with automatic failover
      restart_policy:
        condition: on-failure
        delay: 30s             # Wait longer for NATS to be ready
        max_attempts: 5
        window: 120s
      update_config:
        parallelism: 1
        delay: 30s             # Wait for leader election
        failure_action: rollback
        monitor: 60s
        order: start-first
      placement:
        preferences:
          - spread: node.hostname
        constraints:
          - node.hostname != rosewood  # Avoid intermittent gaming PC
      resources:
        limits:
          memory: 256M
          cpus: '0.5'
        reservations:
          memory: 128M
          cpus: '0.25'
      # Traefik routing for admin API
      labels:
        - traefik.enable=true
        - traefik.http.routers.backbeat-pulse.rule=Host(`backbeat-pulse.chorus.services`)
        - traefik.http.routers.backbeat-pulse.tls=true
        - traefik.http.routers.backbeat-pulse.tls.certresolver=letsencryptresolver
        - traefik.http.services.backbeat-pulse.loadbalancer.server.port=8080
    
    networks:
      - backbeat-net
      - tengig              # External network for Traefik
    
    # Container logging
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"
        tag: "backbeat-pulse/{{.Name}}/{{.ID}}"

  # BACKBEAT Reverb Service - StatusClaim aggregator
  # REQ: BACKBEAT-REQ-020 - Subscribe to INT-B and group by window_id
  # REQ: BACKBEAT-OPS-001 - Reverb can scale stateless
  backbeat-reverb:
    image: anthonyrawlins/backbeat-reverb:v1.0.1
    command: >
      ./reverb
      -cluster=chorus-production
      -nats=nats://nats:4222
      -bar-length=8
      -log-level=info
    environment:
      # REQ: BACKBEAT-OPS-003 - Configuration matching pulse service
      - BACKBEAT_CLUSTER_ID=chorus-production
      - BACKBEAT_NATS_URL=nats://nats:4222
      - BACKBEAT_LOG_LEVEL=info
      - BACKBEAT_WINDOW_TTL=300s    # 5-minute cleanup
      - BACKBEAT_MAX_WINDOWS=100    # Memory limit
    
    # REQ: BACKBEAT-OPS-002 - Health probes for orchestration
    healthcheck:
      test: ["CMD", "wget", "--no-verbose", "--tries=1", "--spider", "http://localhost:8080/healthz"]
      interval: 30s
      timeout: 10s
      retries: 3
      start_period: 60s
    
    deploy:
      replicas: 2              # Stateless, can scale horizontally
      restart_policy:
        condition: on-failure
        delay: 10s
        max_attempts: 3
        window: 120s
      update_config:
        parallelism: 1
        delay: 15s
        failure_action: rollback
        monitor: 45s
        order: start-first
      placement:
        preferences:
          - spread: node.hostname
        constraints:
          - node.hostname != rosewood
      resources:
        limits:
          memory: 512M         # Larger for window aggregation
          cpus: '1.0'
        reservations:
          memory: 256M
          cpus: '0.5'
      # Traefik routing for admin API  
      labels:
        - traefik.enable=true
        - traefik.http.routers.backbeat-reverb.rule=Host(`backbeat-reverb.chorus.services`)
        - traefik.http.routers.backbeat-reverb.tls=true
        - traefik.http.routers.backbeat-reverb.tls.certresolver=letsencryptresolver
        - traefik.http.services.backbeat-reverb.loadbalancer.server.port=8080
    
    networks:
      - backbeat-net
      - tengig              # External network for Traefik
    
    # Container logging
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"
        tag: "backbeat-reverb/{{.Name}}/{{.ID}}"

  # NATS Message Broker - Use existing or deploy dedicated instance
  # REQ: BACKBEAT-INT-001 - Topics via NATS for at-least-once delivery
  nats:
    image: nats:2.9-alpine
    command: ["--jetstream"]
    
    deploy:
      replicas: 1
      restart_policy:
        condition: on-failure
        delay: 10s
        max_attempts: 3
        window: 120s
      placement:
        preferences:
          - spread: node.hostname
        constraints:
          - node.hostname != rosewood
      resources:
        limits:
          memory: 256M
          cpus: '0.5'
        reservations:
          memory: 128M
          cpus: '0.25'
    
    networks:
      - backbeat-net
    
    # Container logging
    logging:
      driver: "json-file"
      options:
        max-size: "10m"
        max-file: "3"
        tag: "nats/{{.Name}}/{{.ID}}"

# Network configuration
networks:
  tengig:
    external: true          # External network for Traefik
  
  backbeat-net:
    driver: overlay
    attachable: true        # Allow external containers to connect
    ipam:
      config:
        - subnet: 10.202.0.0/24

# Persistent storage
# volumes: