Skip to main content
Glama
yangpago-lab

mcp-http-deploy

by yangpago-lab
OverviewSchemaSourceRelated ServersDiscussionsAdmin
Python
Remote

교육용 MCP 서버

파일

mcp_server_http.py            서버 (도구 4개, transport="http")
make_data.py                  교육용 더미 데이터 생성 (이미지 빌드 때 실행됨)
Dockerfile                    python:3.12-slim 기반
docker-compose.yml            mcp-server + gateway 2개 서비스
nginx/default.conf.template   인증, 사용량 제한, 호출 기록 설정
client_test.py                접속 확인용 (langchain-mcp-adapters)

Related MCP server: Gemini Streamable HTTP MCP

실행

docker compose up -d
docker compose ps                   
python client_test.py               
docker compose down

uvicorn

fastmcp의 mcp.run(transport="http")이 내부에서 uvicorn을 띄웁니다. 서버 기동 로그에서 확인할 수 있습니다.

INFO:     Uvicorn running on http://0.0.0.0:8000

  • uvicorn은 비동기라 단일 프로세스로도 여러 클라이언트를 동시에 받습니다.

  • MCP streamable HTTP는 세션 상태를 프로세스 메모리에 둡니다. 트래픽이 정말 늘면 워커가 아니라 컨테이너 개수를 늘리고 로드밸런서를 앞에 둡니다.

Agent에서 연결

mcp_agent.py의 설정에서 transport와 url만 바뀝니다.

# stdio (4일차)
client = MultiServerMCPClient({
    "company": {"command": sys.executable, "args": ["mcp_server.py"], "transport": "stdio"},
})

# streamable HTTP (이 서버, 게이트웨이 경유)
client = MultiServerMCPClient({
    "company": {
        "url": "http://서버주소:8080/mcp",
        "transport": "streamable_http",
        "headers": {"X-API-Key": "발급받은키"},
    },
})

나머지 코드(get_tools, create_agent)는 그대로입니다. 도구를 서버로 빼고 규약으로 주고받는다는 MCP의 약속 덕분입니다.

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityMaintained
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • Gemini Streamable HTTP MCP

    luc-vocab
    F
    license
    Not graded
    quality
    Not graded
    maintenance
    A Docker-based deployment that wraps the Gemini MCP server with supergateway to expose its tool capabilities over Streamable HTTP. It enables users to perform image-editing workflows and interact with the Google Gemini API using a standardized web-accessible endpoint.
    -

  • MCPHubs

    7-e1even
    A
    license
    Not graded
    quality
    C
    maintenance
    A unified gateway and web dashboard that aggregates multiple MCP servers into a single Streamable HTTP endpoint. It supports stdio, SSE, and HTTP protocols, featuring optimized tool exposure modes to reduce token consumption for AI clients.
    5
    MIT

  • mcp_sdk_eyra_accelerator

    Vaishnavi-jagan
    F
    license
    Not graded
    quality
    D
    maintenance
    A standalone MCP server that exposes API endpoints as tools for AI assistants by proxying requests to a target API defined in an OpenAPI specification. It supports various authentication methods and utilizes Server-Sent Events (SSE) to facilitate integration with clients like Claude and ChatGPT.
    -