{
  "id": "cuda/cuda-mps-server-unavailable",
  "signature": "CUDA error: MPS server is not running (cudaErrorMpsServerNotReady)",
  "signature_zh": "CUDA 错误：MPS 服务器未运行 (cudaErrorMpsServerNotReady)",
  "regex": "cudaErrorMpsServerNotReady",
  "domain": "cuda",
  "category": "system_error",
  "subcategory": null,
  "root_cause": "The CUDA Multi-Process Service (MPS) control daemon is not active, but the application attempted to connect to it via CUDA_MPS_PIPE_DIRECTORY or similar environment variables.",
  "root_cause_type": "generic",
  "root_cause_zh": "CUDA 多进程服务 (MPS) 控制守护进程未运行，但应用程序通过 CUDA_MPS_PIPE_DIRECTORY 或类似环境变量尝试连接到它。",
  "versions": [
    {
      "version": "CUDA 12.4",
      "introduced": null,
      "deprecated": null,
      "removed": null,
      "behavior_change": null,
      "status": "active"
    },
    {
      "version": "NVIDIA Driver 550.54.14",
      "introduced": null,
      "deprecated": null,
      "removed": null,
      "behavior_change": null,
      "status": "active"
    },
    {
      "version": "Ubuntu 22.04",
      "introduced": null,
      "deprecated": null,
      "removed": null,
      "behavior_change": null,
      "status": "active"
    }
  ],
  "os_specific": {},
  "dead_ends": [
    {
      "action": "",
      "why_fails": "The daemon must be started explicitly; restarting the app alone does not launch it.",
      "fail_rate": 1.0,
      "condition": "",
      "sources": []
    },
    {
      "action": "",
      "why_fails": "The environment variable only points to the daemon's socket; if the daemon is not running, no path works.",
      "fail_rate": 0.95,
      "condition": "",
      "sources": []
    }
  ],
  "workarounds": [
    {
      "action": "export CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps; nvidia-cuda-mps-control -d",
      "success_rate": 0.9,
      "how": "export CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps; nvidia-cuda-mps-control -d",
      "condition": "",
      "sources": []
    },
    {
      "action": "unset CUDA_MPS_PIPE_DIRECTORY; unset CUDA_MPS_LOG_DIRECTORY; python your_script.py",
      "success_rate": 0.95,
      "how": "unset CUDA_MPS_PIPE_DIRECTORY; unset CUDA_MPS_LOG_DIRECTORY; python your_script.py",
      "condition": "",
      "sources": []
    }
  ],
  "workarounds_zh": [
    "export CUDA_MPS_PIPE_DIRECTORY=/tmp/nvidia-mps; nvidia-cuda-mps-control -d",
    "unset CUDA_MPS_PIPE_DIRECTORY; unset CUDA_MPS_LOG_DIRECTORY; python your_script.py"
  ],
  "transition_graph": {
    "leads_to": [],
    "preceded_by": [],
    "frequently_confused_with": []
  },
  "official_doc_url": "https://docs.nvidia.com/deploy/mps/index.html",
  "official_doc_section": null,
  "error_code": "cudaErrorMpsServerNotReady",
  "verification_tier": "ai_generated",
  "confidence": 0.82,
  "fix_success_rate": 0.88,
  "resolvable": "true",
  "first_seen": "2024-06-10",
  "last_confirmed": "2024-06-01",
  "last_updated": "2024-06-01",
  "evidence_count": 1,
  "tags": [],
  "locale": "en",
  "aliases": []
}