Auto-oncall-system

# Tool List > Enterprise-level intelligent conversation and operation and maintenance assistant, supporting RAG knowledge base Q&A and AIOps intelligent diagnosis [](https://www.python.org/) [](https://fastapi.tiangolo.com/) [](https://www.langchain.com/) ## ✨ Core Features - 🤖 **Intelligent Conversation** - LangChain multi-turn conversation + streaming output - 📚 **RAG Q&A** - Vector retrieval enhancement, supporting document upload, automatic vector index establishment, and automatic knowledge base update - 🔧 **AIOps Diagnosis** - Plan-Execute-Replan automatic fault diagnosis and root cause analysis - 🌐 **Web Interface** - Modern UI, supporting multiple conversation modes: quick Q&A/streaming conversation - 🔌 **MCP Integration** - Log query and monitoring data tool access ## 🛠️ Tech Stack - **Framework**: FastAPI + LangChain + LangGraph - **LLM**: Alibaba Cloud DashScope (Tongyi Qianwen) - **Vector Database**: Milvus - **Tool Protocol**: MCP (Model Context Protocol) ## 🚀 Quick Start ### Environment Requirements - Python 3.10+ - Alibaba Cloud DashScope API Key ([Get Address](https://dashscope.aliyun.com/)) ### Installation and Launch #### Linux/macOS Environment ```bash # 1. Clone the project git clone <repository_url> cd ~/auto-oncall-system # 2. Install dependencies (recommended to use uv) # Method 1: Use uv (recommended, faster) pip install uv uv venv source .venv/bin/activate uv pip install -e . # Method 2: Use pip pip install -e . # 3. Edit configuration file # First use needs to edit .env file, fill in your DASHSCOPE_API_KEY vim .env # or use other editors # 4. One-click initialization (start Docker + service + document upload) make init # 5. One-click launch make start ``` #### Windows Environment (PowerShell/CMD) If Windows does not support the `make` command, you can manually execute the following steps to start the service: ```powershell # 1. Clone the project git clone <repository_url> cd ~/auto-oncall-system # 2. Create a virtual environment and install dependencies # Method 1: Use uv (recommended, faster) pip install uv # Create a virtual environment uv venv # Activate the virtual environment .venv\Scripts\activate # Install all dependencies uv pip install -e . # Method 2: Use pip python -m venv .venv .venv\Scripts\activate pip install -e . # 3. Edit configuration file # Open .env file with Notepad or other editors, fill in your DASHSCOPE_API_KEY notepad .env # 4. Start Docker Desktop # Ensure Docker Desktop is installed and running # 5. Start Milvus vector database (Docker Compose) docker compose -f vector-database.yml up -d # 6. Wait for Milvus to start (about 5-10 seconds) timeout /t 10 # 7. Start MCP service # Start CLS log query service (open a new PowerShell window) python mcp_servers/cls_server.py # Start Monitor monitoring service (open a new PowerShell window) python mcp_servers/monitor_server.py # 8. Start FastAPI main service (open a new PowerShell window) # Note: Logs will be output to logs\app_YYYY-MM-DD.log python -m uvicorn app.main:app --host 0.0.0.0 --port 9900 # 9. Upload documents to vector database (open a new PowerShell window) # Wait for the service to start and then execute timeout /t 5 python -c "import requests, os, time; [requests.post('http://localhost:9900/api/upload', files={'file': open(f'aiops-docs/{f}', 'rb')}) or time.sleep(1) for f in os.listdir('aiops-docs') if f.endswith('.md')]" ``` **Windows One-Click Launch Script** (recommended) Use the launch script: ```powershell # Launch all services .\start-windows.bat # Stop all services .\stop-windows.bat ``` ### Access Service - **Web Interface**: http://localhost:9900 - **API Documentation**: http://localhost:9900/docs ## 📡 API Interface ### Core Interface | Function | Method | Path | Description | |------|------|------|------| | Ordinary Conversation | POST | `/api/chat` | One-time return | | Streaming Conversation | POST | `/api/chat_stream` | SSE streaming output | | AIOps Diagnosis | POST | `/api/aiops` | Automatic fault diagnosis (streaming) | | File Upload | POST | `/api/upload` | Upload and index documents | | Health Check | GET | `/api/health` | Service status check | ### Example Usage ```bash # Ordinary conversation curl -X POST "http://localhost:9900/api/chat" \ -H "Content-Type: application/json" \ -d '{"Id":"session-123","Question":"Hello"}' # Streaming conversation curl -X POST "http://localhost:9900/api/chat_stream" \ -H "Content-Type: application/json" \ -d '{"Id":"session-123","Question":"Hello"}' \ --no-buffer # AIOps diagnosis curl -X POST "http://localhost:9900/api/aiops" \ -H "Content-Type: application/json" \ -d '{"session_id":"session-123"}' \ --no-buffer ``` ## 📁 Project Structure ``` auto-oncall-system/ ├── app/ # Application core │ ├── __init__.py # Package initialization (automatic loading of log configuration) │ ├── main.py # FastAPI application entry │ ├── config.py # Configuration management (environment variables, MCP server configuration) │ ├── api/ # API routing layer │ │ ├── __init__.py │ │ ├── chat.py # Conversation interface (RAG chat) │ │ ├── aiops.py # AIOps interface (fault diagnosis) │ │ ├── file.py # File management (document upload) │ │ └── health.py # Health check (service status) │ ├── services/ # Business service layer │ │ ├── __init__.py │ │ ├── rag_agent_service.py # RAG Agent (LangGraph state diagram) │ │ ├── aiops_service.py # AIOps service (plan-execute-replan) │ │ ├── vector_store_manager.py # Vector storage manager │ │ ├── vector_embedding_service.py # Vector embedding service │ │ ├── vector_index_service.py # Vector index service │ │ ├── vector_search_service.py # Vector search service │ │ └── document_splitter_service.py # Document splitter service │ ├── agent/ # Agent module │ │ ├── __init__.py │ │ ├── mcp_client.py # MCP client (tool call) │ │ └── aiops/ # AIOps core logic │ │ ├── __init__.py │ │ ├── planner.py # Planner │ │ ├── executor.py # Executor │ │ ├── replanner.py # Replanner │ │ ├── state.py # State definition │ │ └── utils.py # Tool functions │ ├── models/ # Data model layer │ │ ├── __init__.py │ │ ├── aiops.py # AIOps model │ │ ├── document.py # Document model │ │ ├── request.py # Request model │ │ └── response.py # Response model │ ├── tools/ # Agent toolset │ │ ├── __init__.py │ │ ├── knowledge_tool.py # Knowledge base query tool │ │ └── time_tool.py # Time tool │ ├── core/ # Core components │ │ ├── __init__.py │ │ ├── llm_factory.py # LLM factory (model management) │ │ └── milvus_client.py # Milvus client │ └── utils/ # Tool classes │ ├── __init__.py │ └── logger.py # Log configuration (Loguru) ├── static/ # Web frontend (pure static) │ ├── index.html # Main page │ ├── app.js # Frontend logic │ └── styles.css # Style sheet ├── mcp_servers/ # MCP servers │ ├── cls_server.py # CLS log query service │ ├── monitor_server.py # Monitoring data service │ └── README.md # MCP service description ├── aiops-docs/ # Operation and maintenance knowledge base (Markdown documents) ├── logs/ # Log directory (Loguru automatic creation) │ └── app_YYYY-MM-DD.log # Log file rotated by day ├── uploads/ # Temporary directory for uploaded files ├── volumes/ # Milvus data persistence directory ├── .env # Environment variable configuration (need to create manually) ├── Makefile # Project management commands (Linux/macOS) ├── start-windows.bat # Windows launch script ├── stop-windows.bat # Windows stop script ├── vector-database.yml # Milvus Docker Compose configuration ├── pyproject.toml # Project configuration (dependencies, metadata) ├── uv.lock # uv dependency lock file ├── pyrightconfig.json # Pyright type checking configuration └── README.md # Project description ``` ## ⚙️ Configuration Description Configured through `.env` file: ```bash # Alibaba Cloud LLM DashScope configuration (required) # Key management: https://bailian.console.aliyun.com/cn-beijing/?spm=5176.29597918.J_SEsSjsNv72yRuRFS2VknO.2.61ac133ccTVQLw&tab=demohouse#/api-key DASHSCOPE_API_KEY=your-api-key (configure your own key) DASHSCOPE_API_BASE=https://dashscope.aliyuncs.com/compatible-mode/v1 # Default to Singapore site if not configured DASHSCOPE_MODEL=qwen-max # Milvus configuration MILVUS_HOST=localhost MILVUS_PORT=19530 # RAG configuration RAG_TOP_K=3 CHUNK_MAX_SIZE=800 CHUNK_OVERLAP=100 ``` ## 🎯 AIOps Intelligent Operation and Maintenance Based on **Plan-Execute-Replan** mode to achieve automatic fault diagnosis. ### Core Features - ✅ Automatic diagnosis plan formulation (Planner) - ✅ Intelligent tool calls (Executor) - ✅ Dynamic step adjustment (Replanner) - ✅ Streaming output of diagnosis process - ✅ Generate structured report ### Quick Test ```bash # Service has been started automatically through make init # If you need to restart the service: make restart # Access the Web interface, click "Intelligent Operation and Maintenance and Diagnosis Tool" # Or use API curl -X POST "http://localhost:9900/api/aiops" \ -H "Content-Type: application/json" \ -d '{"session_id":"test"}' \ --no-buffer ``` ### Diagnosis Process ``` 1. Planner formulates plan → Generate 4-6 diagnosis steps 2. Executor executes steps → Call MCP tools (log query, monitoring data) 3. Replanner evaluates results → Decide to continue/adjust/generate report 4. Output diagnosis report → Root cause analysis + operation and maintenance suggestions ``` ## 📝 Development Guide ### Common Commands ```bash # Project management make init # One-click initialization (Docker + service + document) make start # Launch all services make stop # Stop all services make restart # Restart all services # Dependency management make install-dev # Install development dependencies make sync # Synchronize dependencies # Docker management make up # Launch Docker container make down # Stop Docker container # Code quality make format # Format code make lint # Code check ``` ## 🐛 Frequently Asked Questions ### Windows Environment Issues #### 1. `make` command not available Windows does not support the `make` command, please use the provided batch script: ```powershell # Launch service .\start-windows.bat # Stop service .\stop-windows.bat ``` #### 2. PowerShell execution policy restriction If you encounter "script cannot be loaded because script execution is disabled on this system" error: ```powershell # Temporarily allow script execution (administrator privileges) Set-ExecutionPolicy -ExecutionPolicy RemoteSigned -Scope Process # Or use CMD instead of PowerShell cmd .\start-windows.bat ``` #### 3. Port occupied (Windows) ```powershell # View occupied port process netstat -ano | findstr :9900 # End process (replace PID with actual process ID) taskkill /F /PID <PID> ``` ### General Issues ### API Key Error ```bash # Check Environment Variables cat .env | grep DASHSCOPE_API_KEY # Linux/macOS type .env | findstr DASHSCOPE_API_KEY # Windows ``` ### Milvus Connection Failure ```bash # Ensure Docker service is available and running on your local machine (you can use Docker Desktop) # Check Milvus status docker ps | grep milvus # Restart Milvus (using docker compose) docker compose -f vector-database.yml restart # Or restart a single service docker compose -f vector-database.yml restart standalone ``` ### Service Failure to Start **Linux/macOS:** ```bash # View service logs tail -f logs/app_$(date +%Y-%m-%d).log # FastAPI main service (Loguru logs) tail -f mcp_cls.log # CLS MCP service tail -f mcp_monitor.log # Monitor MCP service # Check port occupancy lsof -i :9900 # FastAPI lsof -i :8003 # CLS MCP lsof -i :8004 # Monitor MCP ``` **Windows:** ```powershell # View service logs (get today's date) $today = Get-Date -Format "yyyy-MM-dd" type logs\app_$today.log # FastAPI main service (Loguru logs) type mcp_cls.log # CLS MCP service type mcp_monitor.log # Monitor MCP service # Or view the latest log file Get-ChildItem logs\*.log | Sort-Object LastWriteTime -Descending | Select-Object -First 1 | Get-Content -Tail 50 # Check port occupancy netstat -ano | findstr :9900 # FastAPI netstat -ano | findstr :8003 # CLS MCP netstat -ano | findstr :8004 # Monitor MCP ``` ## 📚 Reference Resources - [FastAPI Documentation](https://fastapi.tiangolo.com/) - [LangChain Documentation](https://python.langchain.com/) - [LangGraph Plan-Execute](https://langchain-ai.github.io/langgraph/tutorials/plan-and-execute/) - [Alibaba Cloud DashScope](https://dashscope.aliyun.com/) - [MCP Protocol](https://modelcontextprotocol.io/) ## 📄 License MIT License