Sequential Questioning MCP Server

# Deployment Guide for Sequential Questioning MCP Server This guide outlines the steps to deploy the Sequential Questioning MCP Server in various environments. ## Prerequisites - Docker and Docker Compose installed - Python 3.10 or higher (if not using Docker) - Access to a PostgreSQL database - Access to a Qdrant vector database instance (or use the Docker Compose configuration) - API key for the LLM service (OpenAI or equivalent) ## Environment Variables Create a `.env` file with the following variables: ``` # Database Configuration DATABASE_URL=postgresql://user:password@db:5432/sequential_questioning DATABASE_USER=user DATABASE_PASSWORD=password DATABASE_NAME=sequential_questioning DATABASE_HOST=db DATABASE_PORT=5432 # Application Configuration API_KEY=your_api_key_here DEBUG=false ENVIRONMENT=production LOG_LEVEL=info # Vector Database Configuration VECTOR_DB_URL=http://qdrant:6333 VECTOR_COLLECTION_NAME=questions # LLM Configuration OPENAI_API_KEY=your_openai_api_key_here MODEL_NAME=gpt-4-turbo ``` Adjust these values according to your deployment environment. ## Deployment Options ### Option 1: Docker Compose (Recommended) 1. Clone the repository: ```bash git clone https://github.com/yourusername/sequential-questioning.git cd sequential-questioning ``` 2. Create the `.env` file as described above. 3. Build and run the services: ```bash docker-compose up -d ``` 4. The API will be available at `http://localhost:8000`. ### Option 2: Manual Deployment 1. Clone the repository: ```bash git clone https://github.com/yourusername/sequential-questioning.git cd sequential-questioning ``` 2. Create and activate a virtual environment: ```bash python -m venv venv source venv/bin/activate # On Windows: venv\Scripts\activate ``` 3. Install dependencies: ```bash pip install -e . ``` 4. Create the `.env` file as described above, adjusting the database URL to point to your PostgreSQL instance. 5. Run the application: ```bash uvicorn app.main:app --host 0.0.0.0 --port 8000 ``` ### Option 3: Kubernetes Deployment 1. Create Kubernetes configuration files: **deployment.yaml**: ```yaml apiVersion: apps/v1 kind: Deployment metadata: name: sequential-questioning spec: replicas: 3 selector: matchLabels: app: sequential-questioning template: metadata: labels: app: sequential-questioning spec: containers: - name: sequential-questioning image: yourdockerregistry/sequential-questioning:latest ports: - containerPort: 8000 envFrom: - secretRef: name: sequential-questioning-secrets - configMapRef: name: sequential-questioning-config ``` **service.yaml**: ```yaml apiVersion: v1 kind: Service metadata: name: sequential-questioning spec: selector: app: sequential-questioning ports: - port: 80 targetPort: 8000 type: ClusterIP ``` 2. Create Kubernetes secrets and config maps for environment variables. 3. Apply the configuration: ```bash kubectl apply -f deployment.yaml kubectl apply -f service.yaml ``` ## Health Check Once deployed, verify the application is running correctly by accessing the health check endpoint: ``` GET /health ``` Expected response: ```json { "status": "ok", "version": "1.0.0" } ``` ## Monitoring Monitor the application using the internal monitoring endpoints: ``` GET /mcp-internal/monitoring/metrics ``` This endpoint requires the API key for authentication. ## Database Migrations The application handles database migrations automatically on startup. However, if you need to run migrations manually: ```bash alembic upgrade head ``` ## Scaling Considerations - The application is stateless and can be horizontally scaled by increasing the number of replicas. - For high-traffic environments, consider implementing a load balancer. - Monitor and adjust the database connection pool size based on load. - Consider using a managed vector database service for production deployments. ## Backup and Recovery - Regularly backup the PostgreSQL database. - Backup the vector database collections. - Store environment variables and configuration securely. ## Security Considerations - Use HTTPS in production. - Rotate API keys regularly. - Implement rate limiting for API endpoints. - Consider using a managed identity service for authentication in production.