Skip to content

Why it is smoothly to execute the task in the openspace lonely, however, when it comes to the openclaw with openspace, the mcp server always exceeds the limited time #73

@xzlee-beep

Description

@xzlee-beep

We have completed the configuration task of MCP Server in the openclaw.json, however, when we delegate the task to the openspace in the openclaw control web, it always exceeds the limited time using the mcp server. The related log always stays in the process of 2026-04-09 14:09:19.124 [INFO ] utils.py:3995 - LiteLLM completion() model= qwen3.5-plus; provider = dashscope (It can be observerd in the following picture).

Image

When we execute the task directly in the openspcae it shows the success, and we present the following picture.

Image.

We use openclaw to analysis the problem, it suggets the following:
Error Type: MCP error -32001: Request timed out
Timeout Setting: 300 seconds (5 minutes)
Observed Phenomenon: Multiple consecutive calls to execute_task all timeout, but the OpenSpace process continues running normally
Possible Root Causes:

  1. LLM Call Timeout (OpenSpace internal default: 120 seconds)
    The LLM API call within OpenSpace times out before the MCP layer timeout
  2. Skill Selection Stage Blocking
    The initial skill discovery/selection phase hangs during the first request
  3. Subprocess Pipe Handling Issues
    Communication channel between processes gets blocked or fails to transmit responses
  4. Missing Real-time Output Feedback
    No progress logs or streaming output during execution, making it impossible to diagnose where the process is stuck

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions