modelcontextprotocol / modelcontextprotocol/python-sdk

STDIO hangs forever when the using multiprocessing in tools

Open
#817 3 comments 0 reactions 0 assignees View on GitHub

Nobody has claimed this yet.

enhancement needs decision P3
Dominant language
Python
Stars
24.3k
Forks
4k
Avg merge
1d 1h
Merged PRs (30d)
31

Description

Describe the bug
when multiprocessing in a tool, and start mcp client by STDIO, the mcp server never response.

To Reproduce
Steps to reproduce the behavior:

  1. write mcp server:
## test_mcp.py
from mcp.server.fastmcp import FastMCP
import multiprocessing as mp


mcp = FastMCP()

@mcp.tool()
def sum() -> int:
    "count number"
    with mp.Pool(processes=2) as pool:
        res =  pool.imap(str, range(10))
        res = list(res)
    return res

if __name__ == '__main__':
    mcp.run()


  1. write mcp client usig stdio
## test_client.py
from mcp import ClientSession, StdioServerParameters, types
from mcp.client.stdio import stdio_client

# Create server parameters for stdio connection
server_params = StdioServerParameters(
    command="python",  # Executable
    args=["test_mcp.py"],  # Optional command line arguments
    env=None,  # Optional environment variables
)


async def run():
    async with stdio_client(server_params) as (read, write):
        async with ClientSession(
            read, write
        ) as session:
            # Initialize the connection
            await session.initialize()
            resources = await session.list_resources()
            # List available tools
            tools = await session.list_tools()
            print(tools)
            result = await session.call_tool(
                "sum", 
            )
            print(result)
asyncio.run(run())
  1. test running
$ python test_client.py
[05/27/25 15:42:13] INFO     Processing request of type ListResourcesRequest                             server.py:551
                    INFO     Processing request of type ListToolsRequest                                 server.py:551
meta=None nextCursor=None tools=[Tool(name='sum', description='count number', inputSchema={'properties': {}, 'title': 'sumArguments', 'type': 'object'}, annotations=None)]
                    INFO     Processing request of type CallToolRequest                                  server.py:551

The program is stuck here and won't continue running

Version

Name: mcp
Version: 1.9.0
Summary: Model Context Protocol SDK
Home-page: 
Author: Anthropic, PBC.
Author-email: 
License: MIT
Location: /data/software/micromamba/envs/gbi/lib/python3.10/site-packages
Requires: anyio, httpx, httpx-sse, pydantic, pydantic-settings, python-multipart, sse-starlette, starlette, uvicorn
Required-by: camel-ai, cellrank-mcp, fast-agent-mcp, fastmcp, liana-mcp, mcp-server-calculator, scmcp, scmcp-utils, scmcp_shared

Contributor guide

Open the contributing guide

First steps

  1. Read the whole issue, then the project's contributing guide.
  2. Comment on the issue to say you are picking it up — it saves two people doing the same work.
  3. Fork the repository and make your change on a branch.
  4. Open a pull request that references the issue number.

Research direction

Start by running test_client.py with test_mcp.py and reproduce the hang after the CallToolRequest is logged. Trace the interaction between the multiprocessing pool in the sum tool and the stdio client/server entry points. Done means the example completes and returns the tool result instead of remaining stuck.

Written by the indexing model from the issue text.

Assessment

Tech stack
python
Domain
backend
Issue type
Bug
Difficulty
4/5
Estimated time
3-5 days
Activity status
Stale
Clarity
Mostly clear
Newbie friendliness
35/100

Get new issues in your inbox

A short digest of beginner-friendly GitHub issues.