Skip to content

deployment.md

Latest commit

 

History

History
135 lines (100 loc) · 2.89 KB

deployment.md

File metadata and controls

135 lines (100 loc) · 2.89 KB

Deployment Guide

This guide explains how to deploy the TechWord Translator MCP Server as a public service.

Deployment Options

1. Docker (Self-hosted)

Deploy on any server with Docker installed.

Steps:

  1. Build the image:

    docker compose build

  2. Run with environment variables:

    docker compose up -d

Using .env file:

  1. Create a .env file:

    TECHWORD_TRANSLATOR_API_URL=your-api-url

  2. Start the service:

    docker compose up -d

Connecting Clients to Public Server

For MCP Clients using stdio

Since MCP servers typically communicate via stdio (standard input/output), deploying as a "public server" requires a different approach:

Option A: MCP-over-HTTP Proxy You'll need to create an HTTP wrapper that:

  1. Accepts MCP protocol messages over HTTP/WebSocket
  2. Forwards them to the stdio-based MCP server
  3. Returns responses back to the client

Option B: Direct Distribution Instead of hosting centrally, distribute the server as a package:

  1. Publish to PyPI: pip install techword-translator-mcp
  2. Users run locally: python -m techword_translator
  3. Configuration via environment variables

Option C: Claude Desktop Integration Users install locally and configure in Claude Desktop:

{
  "mcpServers": {
    "techword-translator": {
      "command": "python",
      "args": ["-m", "techword_translator"]
    }
  }
}

Publishing to PyPI

To make your server publicly available via PyPI:

  1. Create account at https://pypi.org

  2. Install build tools:

    pip install build twine

  3. Build the package:

    python -m build

  4. Upload to PyPI:

    twine upload dist/*

  5. Users can then install:

    pip install techword-translator-mcp

Security Considerations

When deploying publicly:

  1. Environment Variables: Use secure secret management
  2. Rate Limiting: Implement rate limiting to prevent abuse
  3. Monitoring: Set up logging and monitoring
  4. Updates: Keep dependencies updated for security patches

Monitoring and Logs

Docker

docker logs -f techword-mcp

Local Installation

Check logs in your terminal or use log files if configured

Scaling Considerations

  • MCP servers are typically lightweight
  • Most deployments work well with a single instance
  • For high traffic, consider:
    • Load balancing multiple instances
    • Caching API responses
    • Connection pooling for HTTP requests

Troubleshooting

Server won't start

  • Check environment variables are set correctly
  • Verify API credentials are valid
  • Check logs for error messages

Connection issues

  • Ensure the API base URL is accessible
  • Check firewall rules
  • Verify network connectivity

Performance issues

  • Monitor API response times
  • Check for rate limiting from the upstream API
  • Consider implementing caching