# ============================================================
# AgentCraft-Claw 全栈 Docker Compose 部署
# 包含：基础设施( MySQL + Redis + MinIO + Milvus ) + 后端 + 前端 + 可观测性( Prometheus + Grafana + Alertmanager )
# ============================================================
#
# 用法：
#   1. 首次启动：  docker compose up -d --build
#   2. 查看日志：  docker compose logs -f backend
#   3. 停止（保留数据）： docker compose down
#   4. 停止并清理数据（开发环境）： docker compose down -v
#   5. 仅基础设施： docker compose -f backend/docker-compose.dev.yml up -d
#
# 访问地址：
#   - 前端：    http://localhost
#   - 后端：    http://localhost/api
#   - MinIO：   http://localhost:9003 (minioadmin / minioadmin，生产必须修改)
#   - Grafana： http://localhost:3000 (admin / admin，生产必须修改)
#   - Alertmanager： http://localhost:9093
#
# 验证命令速查：
#   docker compose ps                              # 查看所有服务健康状态
#   curl http://localhost/api/actuator/health      # 后端健康检查
#   curl http://localhost:9090/api/v1/targets      # Prometheus targets
#   curl -X POST http://localhost:9093/-/reload    # 重载 Alertmanager 配置
#   curl http://localhost:9093/api/v2/status       # Alertmanager 状态
# ============================================================

# 生产环境必须设置以下环境变量：
#   BAILIAN_API_KEY    - 阿里云百炼 API Key（必填）
#   JWT_SECRET         - 至少 32 字节的随机字符串（必填）
#   MYSQL_ROOT_PASSWORD - MySQL root 密码（建议修改）
#   MINIO_ROOT_PASSWORD - MinIO root 密码（建议修改）
# 可选：
#   TAVILY_API_KEY, WEB_SEARCH_ENABLED, CORS_ALLOWED_ORIGINS
#   ALERT_EMAIL_*, DINGTALK_WEBHOOK_URL, WECHAT_WORK_WEBHOOK_URL, FEISHU_WEBHOOK_URL, SMS_WEBHOOK_URL

networks:
  agentcraft_default:
    driver: bridge
    name: agentcraft_default
  agentcraft-claw_default:
    driver: bridge

services:

  # ===== 基础设施 =====

  mysql:
    image: mysql:8.0
    container_name: agentcraft-mysql
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      MYSQL_ROOT_PASSWORD: ${MYSQL_ROOT_PASSWORD:-root}
      MYSQL_DATABASE: agentcraft
      MYSQL_CHARSET: utf8mb4
      MYSQL_COLLATION: utf8mb4_unicode_ci
    volumes:
      - agentcraft_mysql_data:/var/lib/mysql
      - ./docs/agentcraft.sql:/docker-entrypoint-initdb.d/01-init.sql:ro
    ports:
      - "3307:3306"
    healthcheck:
      test: ["CMD", "mysqladmin", "ping", "-h", "localhost", "-u", "root", "-p${MYSQL_ROOT_PASSWORD:-root}"]
      interval: 10s
      timeout: 5s
      retries: 5
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 2G

  redis:
    image: redis:7-alpine
    container_name: agentcraft-redis
    restart: unless-stopped
    networks:
      - agentcraft_default
    command: redis-server --save 60 1 --loglevel warning ${REDIS_PASSWORD:+--requirepass ${REDIS_PASSWORD}}
    volumes:
      - agentcraft_redis_data:/data
    ports:
      - "6380:6379"
    healthcheck:
      test: ["CMD-SHELL", "redis-cli ${REDIS_PASSWORD:+--pass ${REDIS_PASSWORD}} ping | grep PONG"]
      interval: 10s
      timeout: 5s
      retries: 5
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 512M

  etcd:
    image: quay.io/coreos/etcd:v3.5.5
    container_name: agentcraft-etcd
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      ETCD_AUTO_COMPACTION_MODE: revision
      ETCD_AUTO_COMPACTION_RETENTION: "1000"
      ETCD_QUOTA_BACKEND_BYTES: "4294967296"
      ETCD_SNAPSHOT_COUNT: "50000"
    command: >
      etcd
      --advertise-client-urls=http://etcd:2379
      --listen-client-urls=http://0.0.0.0:2379
      --data-dir=/etcd
    volumes:
      - agentcraft_etcd_data:/etcd
    healthcheck:
      test: ["CMD", "etcdctl", "endpoint", "health"]
      interval: 10s
      timeout: 5s
      retries: 5
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 256M

  minio:
    image: minio/minio:RELEASE.2024-06-29T01-20-47Z
    container_name: agentcraft-minio
    restart: unless-stopped
    networks:
      - agentcraft_default
    command: server /data --console-address ":9001"
    environment:
      MINIO_ROOT_USER: ${MINIO_ROOT_USER:-minioadmin}
      MINIO_ROOT_PASSWORD: ${MINIO_ROOT_PASSWORD:-minioadmin}
    volumes:
      - agentcraft_minio_data:/data
    ports:
      - "9002:9000"
      - "9003:9001"
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9000/minio/health/live"]
      interval: 15s
      timeout: 5s
      retries: 5
    deploy:
      resources:
        limits:
          cpus: '1'
          memory: 1G

  milvus:
    image: milvusdb/milvus:v2.3.4
    container_name: agentcraft-milvus
    restart: unless-stopped
    networks:
      - agentcraft_default
    command: ["milvus", "run", "standalone"]
    environment:
      ETCD_ENDPOINTS: etcd:2379
      MINIO_ADDRESS: minio:9000
    volumes:
      - agentcraft_milvus_data:/var/lib/milvus
    ports:
      - "19530:19530"
    depends_on:
      etcd:
        condition: service_healthy
      minio:
        condition: service_healthy
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9091/healthz"]
      interval: 15s
      timeout: 5s
      retries: 5
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 4G

  # ===== 后端 =====
  backend:
    build:
      context: ./backend
      dockerfile: Dockerfile
    container_name: agentcraft-backend
    restart: unless-stopped
    environment:
      # 数据库
      MYSQL_HOST: mysql
      MYSQL_PORT: 3306
      MYSQL_DATABASE: agentcraft
      MYSQL_USERNAME: root
      MYSQL_PASSWORD: ${MYSQL_ROOT_PASSWORD:-root}
      MYSQL_USE_SSL: "false"
      MYSQL_ALLOW_PUBLIC_KEY_RETRIEVAL: "true"
      # Redis
      REDIS_HOST: redis
      REDIS_PORT: 6379
      REDIS_PASSWORD: ${REDIS_PASSWORD:-}
      # Milvus
      MILVUS_HOST: milvus
      MILVUS_PORT: 19530
      # MinIO
      MINIO_ENDPOINT: http://minio:9000
      MINIO_ACCESS_KEY: ${MINIO_ROOT_USER:-minioadmin}
      MINIO_SECRET_KEY: ${MINIO_ROOT_PASSWORD:-minioadmin}
      # 安全
      JWT_SECRET: ${JWT_SECRET}
      CORS_ALLOWED_ORIGINS: ${CORS_ALLOWED_ORIGINS:-http://localhost}
      # LLM (必须替换为实际 API Key)
      BAILIAN_API_KEY: ${BAILIAN_API_KEY}
      OPENCLAW_API_KEY: ${OPENCLAW_API_KEY:-}
      # Tavily (可选，用于联网搜索)
      TAVILY_API_KEY: ${TAVILY_API_KEY:-}
      WEB_SEARCH_ENABLED: ${WEB_SEARCH_ENABLED:-false}
      # 日志级别
      APP_LOG_LEVEL: ${APP_LOG_LEVEL:-INFO}
      # Loki 日志推送地址
      LOKI_URL: http://loki:3100
      # 性能优化配置
      LLM_COST_MODELS: ${LLM_COST_MODELS:-qwen-plus=0.0005:0.0015,qwen-max=0.002:0.006,gpt-4o=0.018:0.054}
      LLM_BUDGET_ALERT_THRESHOLD: ${LLM_BUDGET_ALERT_THRESHOLD:-0.80}
      MASKING_ENABLED: ${MASKING_ENABLED:-true}
      MASKING_DEFAULT_LEVEL: ${MASKING_DEFAULT_LEVEL:-L2_API}
      HIKARI_MAX_POOL_SIZE: ${HIKARI_MAX_POOL_SIZE:-50}
      HIKARI_MIN_IDLE: ${HIKARI_MIN_IDLE:-10}
      # 链路追踪配置
      JAEGER_ENDPOINT: ${JAEGER_ENDPOINT:-http://jaeger:4317}
      # Neo4j 图数据库
      NEO4J_URI: ${NEO4J_URI:-bolt://neo4j:7687}
      NEO4J_USERNAME: ${NEO4J_USERNAME:-neo4j}
      NEO4J_PASSWORD: ${NEO4J_PASSWORD:-agentcraft}
      # Nacos 配置中心
      NACOS_SERVER_ADDR: ${NACOS_SERVER_ADDR:-nacos:8848}
      NACOS_NAMESPACE: ${NACOS_NAMESPACE:-}
      NACOS_GROUP: ${NACOS_GROUP:-DEFAULT_GROUP}
      # RabbitMQ
      RABBITMQ_HOST: ${RABBITMQ_HOST:-rabbitmq}
      RABBITMQ_PORT: ${RABBITMQ_PORT:-5672}
      RABBITMQ_USERNAME: ${RABBITMQ_USERNAME:-agentcraft}
      RABBITMQ_PASSWORD: ${RABBITMQ_PASSWORD:-agentcraft}
      # Python AI 子服务 gRPC
      AI_SERVICE_GRPC_HOST: ${AI_SERVICE_GRPC_HOST:-python-ai-service}
      AI_SERVICE_GRPC_PORT: ${AI_SERVICE_GRPC_PORT:-50051}
      # Graph RAG 开关
      GRAPHRAG_ENABLED: ${GRAPHRAG_ENABLED:-false}
      # Phase 3 智能化升级配置
      HIERARCHICAL_MODEL_ENABLED: ${HIERARCHICAL_MODEL_ENABLED:-true}
      VLLM_PLANNER_URL: ${VLLM_PLANNER_URL:-http://vllm-planner:8000}
      VLLM_REVIEWER_URL: ${VLLM_REVIEWER_URL:-http://vllm-reviewer:8001}
      MODEL_TRAINER_URL: ${MODEL_TRAINER_URL:-http://model-training-service:8080}
      AGENTCRAFT_RL_ENABLED: ${AGENTCRAFT_RL_ENABLED:-false}
      NEGOTIATION_ENABLED: ${NEGOTIATION_ENABLED:-false}
      # Phase 4 平台化与商业化配置
      TENANT_ENABLED: ${TENANT_ENABLED:-false}
      BILLING_ENABLED: ${BILLING_ENABLED:-false}
      API_KEY_ENABLED: ${API_KEY_ENABLED:-false}
      COLLABORATION_ENABLED: ${COLLABORATION_ENABLED:-false}
      TTS_ENABLED: ${TTS_ENABLED:-false}
      COMPLIANCE_ENABLED: ${COMPLIANCE_ENABLED:-false}
      WATERMARK_ENABLED: ${WATERMARK_ENABLED:-false}
      OPA_URL: ${OPA_URL:-http://opa:8181}
      # 高级/预留配置（需在 application.yml 中显式接入后生效）
      LOCAL_MODEL_BASE_URL: ${LOCAL_MODEL_BASE_URL:-http://localhost:8000/v1}
      TEI_BASE_URL: ${TEI_BASE_URL:-http://localhost:8080}
      MULTIMODAL_OCR_ENGINE: ${MULTIMODAL_OCR_ENGINE:-tesseract}
      MULTIMODAL_EMBEDDING_URL: ${MULTIMODAL_EMBEDDING_URL:-http://localhost:8080}
      SSO_ENABLED: ${SSO_ENABLED:-false}
      SSO_PROVIDER_TYPE: ${SSO_PROVIDER_TYPE:-oauth2}
      AGENT_INTEROP_CALLBACK_SECRET: ${AGENT_INTEROP_CALLBACK_SECRET:-}
    # 告警通知通道配置（Alertmanager 也会使用，但后端作为通用转发层同样需要）
      ALERT_EMAIL_SMTP_HOST: ${ALERT_EMAIL_SMTP_HOST:-}
      ALERT_EMAIL_SMTP_PORT: ${ALERT_EMAIL_SMTP_PORT:-}
      ALERT_EMAIL_TLS: ${ALERT_EMAIL_TLS:-true}
      ALERT_EMAIL_FROM: ${ALERT_EMAIL_FROM:-}
      ALERT_EMAIL_USERNAME: ${ALERT_EMAIL_USERNAME:-}
      ALERT_EMAIL_PASSWORD: ${ALERT_EMAIL_PASSWORD:-}
      ALERT_EMAIL_TO: ${ALERT_EMAIL_TO:-}
      DINGTALK_WEBHOOK_URL: ${DINGTALK_WEBHOOK_URL:-}
      DINGTALK_AT_MOBILES: ${DINGTALK_AT_MOBILES:-}
      DINGTALK_KEYWORD: ${DINGTALK_KEYWORD:-告警}
      WECHAT_WORK_WEBHOOK_URL: ${WECHAT_WORK_WEBHOOK_URL:-}
      FEISHU_WEBHOOK_URL: ${FEISHU_WEBHOOK_URL:-}
      SMS_WEBHOOK_URL: ${SMS_WEBHOOK_URL:-}
      SMS_API_KEY: ${SMS_API_KEY:-}
      SMS_TEMPLATE_CODE: ${SMS_TEMPLATE_CODE:-}
      SMS_TO_LIST: ${SMS_TO_LIST:-}
    ports:
      - "8080:8080"
    depends_on:
      mysql:
        condition: service_healthy
      redis:
        condition: service_healthy
      minio:
        condition: service_healthy
      milvus:
        condition: service_healthy
    networks:
      - agentcraft_default
      - agentcraft-claw_default
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:8080/actuator/health"]
      interval: 15s
      timeout: 5s
      retries: 5
      start_period: 60s
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 2G

  # ===== 前端 =====
  frontend:
    build:
      context: ./frontend
      dockerfile: Dockerfile
    container_name: agentcraft-frontend
    restart: unless-stopped
    ports:
      - "80:80"
    depends_on:
      backend:
        condition: service_healthy
    networks:
      - agentcraft-claw_default
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost/"]
      interval: 15s
      timeout: 5s
      retries: 5
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 256M

  # ===== 可观测性 =====
  loki:
    image: grafana/loki:2.9.10
    container_name: agentcraft-loki
    restart: unless-stopped
    networks:
      - agentcraft_default
    command: -config.file=/etc/loki/loki-config.yaml
    volumes:
      - agentcraft_loki_data:/loki
      - ./infra/loki/loki-config.yaml:/etc/loki/loki-config.yaml:ro
    ports:
      - "3100:3100"
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 512M

  prometheus:
    image: prom/prometheus:v2.53.0
    container_name: agentcraft-prometheus
    restart: unless-stopped
    networks:
      - agentcraft_default
    command:
      - '--config.file=/etc/prometheus/prometheus.yml'
      - '--storage.tsdb.path=/prometheus'
      - '--storage.tsdb.retention.time=15d'
      - '--web.console.libraries=/etc/prometheus/console_libraries'
      - '--web.console.templates=/etc/prometheus/consoles'
      - '--web.enable-lifecycle'
    volumes:
      - ./infra/prometheus/prometheus.yml:/etc/prometheus/prometheus.yml:ro
      - ./infra/prometheus/alerts.yml:/etc/prometheus/alerts.yml:ro
      - agentcraft_prometheus_data:/prometheus
    ports:
      - "9090:9090"
    depends_on:
      backend:
        condition: service_healthy
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 512M

  grafana:
    image: grafana/grafana:10.4.4
    container_name: agentcraft-grafana
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      GF_SECURITY_ADMIN_USER: ${GRAFANA_ADMIN_USER:-admin}
      GF_SECURITY_ADMIN_PASSWORD: ${GRAFANA_ADMIN_PASSWORD:-admin}
      GF_USERS_ALLOW_SIGN_UP: "false"
    volumes:
      - ./infra/grafana/provisioning:/etc/grafana/provisioning:ro
      - ./infra/grafana/dashboards:/var/lib/grafana/dashboards:ro
      - agentcraft_grafana_data:/var/lib/grafana
    ports:
      - "3000:3000"
    depends_on:
      - prometheus
      - loki
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 256M

  alertmanager:
    image: prom/alertmanager:v0.27.0
    container_name: agentcraft-alertmanager
    restart: unless-stopped
    networks:
      - agentcraft_default
    env_file:
      - .env
    command:
      - '--config.file=/etc/alertmanager/alertmanager.yml'
      - '--storage.path=/alertmanager'
      - '--web.external-url=http://localhost:9093'
      - '--cluster.advertise-address=127.0.0.1:9093'
    volumes:
      - ./infra/alertmanager/alertmanager.yml:/etc/alertmanager/alertmanager.yml:ro
      - agentcraft_alertmanager_data:/alertmanager
    ports:
      - "9093:9093"
    depends_on:
      backend:
        condition: service_healthy
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 256M

  # ===== Phase 2 基础设施 =====

  # RabbitMQ 消息队列
  rabbitmq:
    image: rabbitmq:3.12-management-alpine
    container_name: agentcraft-rabbitmq
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      RABBITMQ_DEFAULT_USER: ${RABBITMQ_USERNAME:-agentcraft}
      RABBITMQ_DEFAULT_PASS: ${RABBITMQ_PASSWORD:-agentcraft}
    volumes:
      - agentcraft_rabbitmq_data:/var/lib/rabbitmq
    ports:
      - "5672:5672"
      - "15672:15672"
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 512M

  # Nacos 配置中心
  nacos:
    image: nacos/nacos-server:v2.3.0
    container_name: agentcraft-nacos
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      MODE: standalone
      SPRING_DATASOURCE_PLATFORM: mysql
      MYSQL_SERVICE_HOST: mysql
      MYSQL_SERVICE_DB_NAME: nacos
      MYSQL_SERVICE_PORT: 3306
      MYSQL_SERVICE_USER: root
      MYSQL_SERVICE_PASSWORD: ${MYSQL_ROOT_PASSWORD:-root}
      NACOS_AUTH_ENABLE: "true"
    ports:
      - "8848:8848"
      - "9848:9848"
    depends_on:
      mysql:
        condition: service_healthy
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 1G

  # Jaeger 链路追踪
  jaeger:
    image: jaegertracing/all-in-one:1.59
    container_name: agentcraft-jaeger
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      COLLECTOR_OTLP_ENABLED: "true"
    ports:
      - "16686:16686"
      - "4317:4317"
      - "4318:4318"
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 1G

  # Neo4j 图数据库（Graph RAG）
  neo4j:
    image: neo4j:5.20-community
    container_name: agentcraft-neo4j
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      NEO4J_AUTH: ${NEO4J_USERNAME:-neo4j}/${NEO4J_PASSWORD:-agentcraft}
      NEO4J_PLUGINS: '["apoc", "gds"]'
      NEO4J_dbms_memory_heap_max__size: 1G
    volumes:
      - agentcraft_neo4j_data:/data
      - agentcraft_neo4j_logs:/logs
    ports:
      - "7474:7474"
      - "7687:7687"
    deploy:
      resources:
        limits:
          cpus: '1'
          memory: 2G

  # Python AI 子服务（Embedding / Rerank / OCR / Safety）
  python-ai-service:
    build:
      context: ./python-ai-service
      dockerfile: Dockerfile
    container_name: agentcraft-python-ai
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      - APP_PORT=8080
      - GRPC_PORT=50051
      - MODEL_CACHE_DIR=/models
      - LOG_LEVEL=INFO
    volumes:
      - agentcraft_models_cache:/models
    ports:
      - "8081:8080"
      - "50051:50051"
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 4G

  # Worker 进程（异步消费者）
  worker:
    build:
      context: ./backend
      dockerfile: Dockerfile.worker
    container_name: agentcraft-worker
    restart: unless-stopped
    environment:
      # 数据库
      MYSQL_HOST: mysql
      MYSQL_PORT: 3306
      MYSQL_DATABASE: agentcraft
      MYSQL_USERNAME: root
      MYSQL_PASSWORD: ${MYSQL_ROOT_PASSWORD:-root}
      # Redis
      REDIS_HOST: redis
      REDIS_PORT: 6379
      REDIS_PASSWORD: ${REDIS_PASSWORD:-}
      # Milvus
      MILVUS_HOST: milvus
      MILVUS_PORT: 19530
      # RabbitMQ
      RABBITMQ_HOST: rabbitmq
      RABBITMQ_PORT: 5672
      RABBITMQ_USERNAME: ${RABBITMQ_USERNAME:-agentcraft}
      RABBITMQ_PASSWORD: ${RABBITMQ_PASSWORD:-agentcraft}
      # Neo4j
      NEO4J_URI: ${NEO4J_URI:-bolt://neo4j:7687}
      NEO4J_USERNAME: ${NEO4J_USERNAME:-neo4j}
      NEO4J_PASSWORD: ${NEO4J_PASSWORD:-agentcraft}
      # Python AI gRPC
      AI_SERVICE_GRPC_HOST: ${AI_SERVICE_GRPC_HOST:-python-ai-service}
      AI_SERVICE_GRPC_PORT: ${AI_SERVICE_GRPC_PORT:-50051}
      # 日志
      APP_LOG_LEVEL: ${APP_LOG_LEVEL:-INFO}
    depends_on:
      mysql:
        condition: service_healthy
      redis:
        condition: service_healthy
      milvus:
        condition: service_healthy
      rabbitmq:
        condition: service_healthy
    networks:
      - agentcraft_default
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 2G

  # ===== Phase 3: 模型训练与 vLLM 推理 =====

  vllm-planner:
    image: vllm/vllm-openai:v0.5.0
    container_name: agentcraft-vllm-planner
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      - CUDA_VISIBLE_DEVICES=0
    volumes:
      - agentcraft_models:/models
      - agentcraft_lora:/lora-adapters
    command: >
      --model /models/Qwen2.5-7B
      --enable-lora
      --lora-modules planner=/lora-adapters/planner
      --port 8000
      --tensor-parallel-size 1
    ports:
      - "8000:8000"
    deploy:
      resources:
        limits:
          cpus: '4'
          memory: 16G
    # 生产环境启用 GPU 时取消注释：
    # runtime: nvidia

  vllm-reviewer:
    image: vllm/vllm-openai:v0.5.0
    container_name: agentcraft-vllm-reviewer
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      - CUDA_VISIBLE_DEVICES=0
    volumes:
      - agentcraft_models:/models
      - agentcraft_lora:/lora-adapters
    command: >
      --model /models/Qwen2.5-3B
      --enable-lora
      --lora-modules reviewer=/lora-adapters/reviewer,safety=/lora-adapters/safety
      --port 8001
    ports:
      - "8001:8001"
    deploy:
      resources:
        limits:
          cpus: '2'
          memory: 8G
    # 生产环境启用 GPU 时取消注释：
    # runtime: nvidia

  model-training-service:
    build:
      context: ./model-training-service
      dockerfile: Dockerfile
    container_name: agentcraft-model-trainer
    restart: unless-stopped
    networks:
      - agentcraft_default
    environment:
      - VLLM_PLANNER_URL=http://vllm-planner:8000
      - VLLM_REVIEWER_URL=http://vllm-reviewer:8001
      - MODELS_DIR=/models
      - LORA_DIR=/lora-adapters
    volumes:
      - agentcraft_models:/models
      - agentcraft_lora:/lora-adapters
    ports:
      - "8082:8080"
    deploy:
      resources:
        limits:
          cpus: '4'
          memory: 16G
    # 生产环境启用 GPU 时取消注释：
    # runtime: nvidia

  # ===== Phase 4: OPA 策略引擎 =====

  opa:
    image: openpolicyagent/opa:0.65.0
    container_name: agentcraft-opa
    restart: unless-stopped
    networks:
      - agentcraft_default
    command: run --server --addr :8181 /policies
    volumes:
      - ./infra/opa/policies:/policies:ro
    ports:
      - "8181:8181"
    healthcheck:
      test: ["CMD", "wget", "-qO-", "http://localhost:8181/health"]
      interval: 15s
      timeout: 5s
      retries: 5
    deploy:
      resources:
        limits:
          cpus: '0.5'
          memory: 256M

volumes:
  agentcraft_mysql_data:
  agentcraft_redis_data:
  agentcraft_etcd_data:
  agentcraft_minio_data:
  agentcraft_milvus_data:
  agentcraft_prometheus_data:
  agentcraft_grafana_data:
  agentcraft_loki_data:
  agentcraft_alertmanager_data:
  # Phase 2 新增
  agentcraft_rabbitmq_data:
  agentcraft_neo4j_data:
  agentcraft_neo4j_logs:
  agentcraft_models_cache:
  # Phase 3 新增
  agentcraft_models:
  agentcraft_lora: