docker-compose.yml

# 电网运维 RAG 智能问答系统 - 全栈编排
# S1: mysql + minio | S5: + etcd + milvus-minio + milvus | S11: + backend + frontend
# 一键启动: docker compose up -d --build

services:
  # ---------- MySQL：结构化元数据 ----------
  mysql:
    image: mysql:8.0
    container_name: grid-mysql
    restart: unless-stopped
    environment:
      MYSQL_ROOT_PASSWORD: ${MYSQL_ROOT_PASSWORD:-root123456}
      MYSQL_DATABASE: ${MYSQL_DATABASE:-grid_qa}
      MYSQL_USER: ${MYSQL_USER:-grid}
      MYSQL_PASSWORD: ${MYSQL_PASSWORD:-grid123456}
      TZ: Asia/Shanghai
    ports:
      - "3307:3306"   # 宿主机 3307，避开本机 MySQL
    volumes:
      - mysql-data:/var/lib/mysql
    command:
      --character-set-server=utf8mb4
      --collation-server=utf8mb4_unicode_ci
      --default-authentication-plugin=mysql_native_password
    healthcheck:
      test: ["CMD", "mysqladmin", "ping", "-h", "localhost", "-uroot", "-p${MYSQL_ROOT_PASSWORD:-root123456}"]
      interval: 10s
      timeout: 5s
      retries: 12
      start_period: 30s

  # ---------- MinIO：源文档对象存储 ----------
  minio:
    image: minio/minio:latest
    container_name: grid-minio
    restart: unless-stopped
    environment:
      MINIO_ROOT_USER: ${MINIO_ACCESS_KEY:-minioadmin}
      MINIO_ROOT_PASSWORD: ${MINIO_SECRET_KEY:-minioadmin}
    ports:
      - "9000:9000"
      - "9001:9001"
    volumes:
      - ./data/minio:/data
    command: server /data --console-address ":9001"

  # ---------- etcd：Milvus 元数据依赖 ----------
  etcd:
    image: quay.io/coreos/etcd:v3.5.5
    container_name: grid-etcd
    restart: unless-stopped
    environment:
      ETCD_AUTO_COMPACTION_MODE: revision
      ETCD_AUTO_COMPACTION_RETENTION: "1000"
      ETCD_QUOTA_BACKEND_BYTES: "4294967296"
      ETCD_SNAPSHOT_COUNT: "50000"
    volumes:
      - ./data/etcd:/etcd
    command: etcd -advertise-client-urls=http://etcd:2379 -listen-client-urls http://0.0.0.0:2379 --data-dir /etcd
    healthcheck:
      test: ["CMD", "etcdctl", "endpoint", "health"]
      interval: 30s
      timeout: 20s
      retries: 3

  # ---------- Milvus 专用对象存储 ----------
  milvus-minio:
    image: minio/minio:latest
    container_name: grid-milvus-minio
    restart: unless-stopped
    environment:
      MINIO_ROOT_USER: minioadmin
      MINIO_ROOT_PASSWORD: minioadmin
    volumes:
      - ./data/milvus-minio:/minio_data
    command: minio server /minio_data

  # ---------- Milvus 2.4 向量库 ----------
  milvus:
    image: milvusdb/milvus:v2.4.10
    container_name: grid-milvus
    command: ["milvus", "run", "standalone"]
    restart: unless-stopped
    depends_on:
      - etcd
      - milvus-minio
    environment:
      ETCD_ENDPOINTS: etcd:2379
      MINIO_ADDRESS: milvus-minio:9000
    ports:
      - "19530:19530"
      - "9091:9091"
    healthcheck:
      test: ["CMD", "curl", "-f", "http://localhost:9091/healthz"]
      interval: 30s
      start_period: 90s
      timeout: 20s
      retries: 3

  # ---------- Redis：热点问答缓存 ----------
  redis:
    image: redis:7-alpine
    container_name: grid-redis
    restart: unless-stopped
    ports:
      - "6379:6379"
    volumes:
      - ./data/redis:/data
    command: redis-server --appendonly yes --maxmemory 300mb --maxmemory-policy allkeys-lru --maxmemory-samples 10
    healthcheck:
      test: ["CMD", "redis-cli", "ping"]
      interval: 10s
      timeout: 3s
      retries: 5

  # ---------- Neo4j：知识图谱（设备-故障-处置 多跳推理）----------
  neo4j:
    image: neo4j:5-community
    container_name: grid-neo4j
    restart: unless-stopped
    environment:
      NEO4J_AUTH: neo4j/${NEO4J_PASSWORD:-neo4j123456}
    ports:
      - "7474:7474"   # Neo4j Browser
      - "7687:7687"   # Bolt 协议
    volumes:
      - ./data/neo4j:/data

  # ---------- Ollama：本地应急 LLM（云端 qwen/deepseek/doubao 全部不可用时的兜底，非强依赖）----------
  # LLM_FALLBACK_CHAIN 末位新增 ollama；首次启动自动拉取模型（7B q4 量化，约 4.5GB，CPU 可推理）。
  # mem_limit/cpus 防止真实触发兜底、开始推理时把 mysql/milvus/redis/neo4j 挤爆(OOM/CPU 饥饿)；
  # 按部署机器实际配置调整数值。
  ollama:
    image: ollama/ollama:latest
    container_name: grid-ollama
    restart: unless-stopped
    ports:
      - "11434:11434"
    volumes:
      - ./data/ollama:/root/.ollama
    entrypoint: ["/bin/sh", "-c"]
    command:
      - "ollama serve & sleep 3 && ollama pull ${OLLAMA_MODEL:-qwen2.5:7b-instruct-q4_K_M} && wait"
    mem_limit: 10g
    cpus: 4

  # ---------- Nacos：配置中心（可选，CONFIG_SOURCE=nacos 时启用；默认 env 不依赖）----------
  # 启动: docker compose up -d nacos ；管理台 http://localhost:8848/nacos (默认 nacos/nacos)
  nacos:
    image: nacos/nacos-server:v2.4.3
    container_name: grid-nacos
    restart: unless-stopped
    environment:
      MODE: standalone
    ports:
      - "8848:8848"
      - "9848:9848"
    volumes:
      - ./data/nacos:/home/nacos/data

  # ---------- 后端 ----------
  backend:
    build:
      context: ./backend
      args:
        - HTTP_PROXY=http://host.docker.internal:7897
        - HTTPS_PROXY=http://host.docker.internal:7897
    container_name: grid-backend
    restart: unless-stopped
    depends_on:
      - mysql
      - minio
      - milvus
      - redis
      - neo4j
      - ollama
    env_file: .env
    environment:
      # 容器内用 service name + 内部端口连接（覆盖 .env 的 localhost）
      DATABASE_URL: mysql+aiomysql://grid:grid123456@mysql:3306/grid_qa?charset=utf8mb4
      REDIS_URL: redis://redis:6379/0
      MINIO_ENDPOINT: minio:9000
      MILVUS_HOST: milvus
      NEO4J_URI: bolt://neo4j:7687
      MCP_SERVERS: '[{"name":"mock_scada","url":"http://mock-scada:9100"}]'   # N2 启动发现 mock SCADA → 注册 query_telemetry 到 agent ToolRegistry
      # 容器直连外网：云API(dashscope/deepseek/doubao 国内可达) + bge 走 hf-mirror 国内镜像。
      # 注：宿主机代理 7897 仅监听 127.0.0.1（未开 Allow LAN），容器经 host.docker.internal 连不上，
      # 故不设 HTTPS_PROXY——否则 httpx/requests 把所有出站（含 hf-mirror、云API）都发往不可达代理。
      HF_ENDPOINT: https://hf-mirror.com
      HF_HUB_OFFLINE: "1"            # 离线模式：用挂载的宿主机 HF 缓存，避免联网下载受阻
      TRANSFORMERS_OFFLINE: "1"
    extra_hosts:
      - "host.docker.internal:host-gateway"
    ports:
      - "8001:8001"
    volumes:
      - grid-backend-data:/app/data   # 持久化 data/（运行日志 + 数据备份 .sql），容器重建不丢
      - C:/Users/Administrator/.cache/huggingface/hub:/root/.cache/huggingface/hub:ro   # 宿主机 bge 缓存（离线加载，免下载）

  # ---------- Mock SCADA MCP Server（N2 MCP 总线外部 SCADA 数据源，侧车）----------
  # 复用 backend 镜像跑 mock_scada_server（真实 SCADA 接入时换连接器即可）。
  # backend 经 MCP_SERVERS=http://mock-scada:9100 发现并注册 query_telemetry 等工具到 agent ToolRegistry。
  mock-scada:
    image: rag-backend:latest
    build:
      context: ./backend
    container_name: grid-mock-scada
    restart: unless-stopped
    command: python -m app.mcp.mock_scada_server
    ports:
      - "9100:9100"

  # ---------- 前端（nginx）----------
  frontend:
    build:
      context: ./frontend
      args:
        - HTTP_PROXY=http://host.docker.internal:7897
        - HTTPS_PROXY=http://host.docker.internal:7897
    container_name: grid-frontend
    restart: unless-stopped
    depends_on:
      - backend
    ports:
      - "5173:80"

  # ---------- Prometheus 指标采集 ----------
  # 注：采集本地 uvicorn(host.docker.internal:8001)，不依赖 backend 容器
  prometheus:
    image: prom/prometheus:latest
    container_name: grid-prometheus
    restart: unless-stopped
    ports:
      - "9090:9090"
    volumes:
      - ./prometheus.yml:/etc/prometheus/prometheus.yml:ro
      - ./data/prometheus:/prometheus

  # ---------- Grafana 可视化仪表盘（好看的开箱监控界面）----------
  grafana:
    image: grafana/grafana:latest
    container_name: grid-grafana
    restart: unless-stopped
    ports:
      - "3000:3000"
    environment:
      GF_SECURITY_ADMIN_USER: admin
      GF_SECURITY_ADMIN_PASSWORD: admin
    volumes:
      - ./data/grafana:/var/lib/grafana
      - ./grafana/provisioning:/etc/grafana/provisioning:ro
    depends_on:
      - prometheus

  # ---------- Langfuse：LLM trace 可视化（N4）----------
  langfuse:
    image: langfuse/langfuse:latest
    container_name: grid-langfuse
    restart: unless-stopped
    depends_on:
      - postgres-langfuse
    environment:
      DATABASE_URL: postgresql://langfuse:langfuse@postgres-langfuse:5432/langfuse
      NEXTAUTH_SECRET: ${LANGFUSE_NEXTAUTH_SECRET:-langfuse-secret-2026}
      NEXTAUTH_URL: http://localhost:3001
      SALT: ${LANGFUSE_SALT:-langfuse-salt-2026}
      NEXT_PUBLIC_SIGNUP_DISABLED: "false"
    ports:
      - "3001:3000"   # Langfuse UI（避开 Grafana 的 3000）
    volumes:
      - ./data/langfuse:/data

  # ---------- Langfuse 专用 PostgreSQL ----------
  postgres-langfuse:
    image: postgres:15-alpine
    container_name: grid-langfuse-db
    restart: unless-stopped
    environment:
      POSTGRES_USER: langfuse
      POSTGRES_PASSWORD: langfuse
      POSTGRES_DB: langfuse
    volumes:
      - ./data/langfuse-db:/var/lib/postgresql/data

volumes:
  mysql-data:
  grid-backend-data: