910B机器通过docker容器部署comfyui 后(可以进行推理)，但安装ComfyUI-Manager后，从web 页面通过Manager Restart ComfyUI 会导致ComfyUI 重启失败：报错为：error code is 507033。

一、问题现象（附报错日志上下文）：
疑惑点：为什么docker run -d 启动的ComfyUI 可以发现NPU设备并正常进行推理，而通过ComfyUI-Manager Restart 则会无法发现NPU设备而导致重启失败？

描述：910B机器 部署comfyui 后(可以进行推理)，但安装ComfyUI-Manager后，从web 页面通过Manager 进行ComfyUI restart 会导致服务无法发现NPU 设备。

ComfyUI后台报错日志：

```
Restarting... [Legacy Mode]

Command: ['/usr/local/bin/python3', 'main.py', '--listen', '0.0.0.0', '--port', '8188']
[START] Security scan
[DONE] Security scan
## ComfyUI-Manager: installing dependencies done.
** ComfyUI startup time: 2025-05-15 03:30:53.965
** Platform: Linux
** Python version: 3.10.17 (main, Apr 29 2025, 23:47:08) [GCC 12.2.0]
** Python executable: /usr/local/bin/python3
** ComfyUI Path: /app
** ComfyUI Base Folder Path: /app
** User directory: /app/user
** ComfyUI-Manager config path: /app/user/default/ComfyUI-Manager/config.ini
** Log path: /app/user/comfyui.log

[notice] A new release of pip is available: 23.0.1 -> 25.1.1
[notice] To update, run: pip install --upgrade pip

Prestartup times for custom nodes:
   3.3 seconds: /app/custom_nodes/ComfyUI-Manager

Checkpoint files will always be loaded safely.
Traceback (most recent call last):
  File "/app/main.py", line 137, in <module>
    import execution
  File "/app/execution.py", line 13, in <module>
    import nodes
  File "/app/nodes.py", line 22, in <module>
    import comfy.diffusers_load
  File "/app/comfy/diffusers_load.py", line 3, in <module>
    import comfy.sd
  File "/app/comfy/sd.py", line 7, in <module>
    from comfy import model_management
  File "/app/comfy/model_management.py", line 221, in <module>
    total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)
  File "/app/comfy/model_management.py", line 168, in get_torch_device
    return torch.device("npu", torch.npu.current_device())
  File "/usr/local/lib/python3.10/site-packages/torch_npu/npu/utils.py", line 62, in current_device
    torch_npu.npu._lazy_init()
  File "/usr/local/lib/python3.10/site-packages/torch_npu/npu/__init__.py", line 215, in _lazy_init
    torch_npu._C._npu_init()
RuntimeError: Initialize:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:222 NPU function error: c10_npu::SetDevice(device_id_), error code is 507033
[ERROR] 2025-05-15-03:30:59 (PID:9, Device:-1, RankID:-1) ERR00100 PTA call acl api failed
[Error]: Failed to start the device. 
        Rectify the fault based on the error information in the ascend log.
EE1001: [PID: 9] 2025-05-15-03:30:59.713.495 The argument is invalid.Reason: rtGetDevMsg execute failed, reason=[context pointer null]
        Solution: 1.Check the input parameter range of the function. 2.Check the function invocation relationship.
        TraceBack (most recent call last):
        TsdOpen failed. devId=0, tdt error=35[FUNC:PrintfTsdError][FILE:runtime.cc][LINE:2692]
        Start aicpu executor failed, retCode=0x7020009 devId=0[FUNC:DeviceRetain][FILE:runtime.cc][LINE:3475]
        Check param failed, dev can not be NULL![FUNC:PrimaryContextRetain][FILE:runtime.cc][LINE:3224]
        Check param failed, ctx can not be NULL![FUNC:PrimaryContextRetain][FILE:runtime.cc][LINE:3251]
        Check param failed, context can not be null.[FUNC:NewDevice][FILE:api_impl.cc][LINE:2616]
        New device failed, retCode=0x7010006[FUNC:SetDevice][FILE:api_impl.cc][LINE:2640]
        rtSetDevice execute failed, reason=[device retain error][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
        open device 0 failed, runtime result = 507033.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
        Check param failed, dev can not be NULL![FUNC:DeviceRetain][FILE:runtime.cc][LINE:3527]
        ctx is NULL![FUNC:GetDevErrMsg][FILE:api_impl.cc][LINE:5533]
        The argument is invalid.Reason: rtGetDevMsg execute failed, reason=[context pointer null]
```

二、软件版本:
-- CANN 版本: (Ascend-cann-kernels-910b_8.1.RC1.alpha001_linux-aarch64.run, Ascend-cann-toolkit_8.1.RC1.alpha001_linux-aarch64.run) 
--Tensorflow/Pytorch/MindSpore 版本: NO
![输入图片说明](https://foruda.gitee.com/images/1746696786353819790/479f7137_5131031.png "屏幕截图")
--Python 版本 :Python: 3.10.16
-- MindStudio版本 (e.g., MindStudio 2.0.0 (beta3)): NO
--操作系统版本 (e.g., Ubuntu 18.04): Ubuntu 22.04.4 LTS
![输入图片说明](https://foruda.gitee.com/images/1746696860760489154/32a6f459_5131031.png "屏幕截图")

三、测试步骤：
1、参照上述软件版本，制作并打包的Comfyui 910B arm 版本的docker镜像，在通过docker run 启动后，可以正常进行文生图推理（sd3.5 and flux1.***）。
-- docker run -d 启动comfyui容器与初次推理日志信息（成功的日志，只要不通过Manager Restart， 一直可以进行推理）：

```
LD_LIBRARY_PATH: /usr/local/Ascend/driver/lib64/common:/usr/local/Ascend/driver/lib64/driver:/usr/local/Ascend/ascend-toolkit/latest/tools/aml/lib64:/usr/local/Ascend/ascend-toolkit/latest/tools/aml/lib64/plugin:/usr/local/Ascend/ascend-toolkit/latest/lib64:/usr/local/Ascend/ascend-toolkit/latest/lib64/plugin/opskernel:/usr/local/Ascend/ascend-toolkit/latest/lib64/plugin/nnengine:/usr/local/Ascend/ascend-toolkit/latest/opp/built-in/op_impl/ai_core/tbe/op_tiling/lib/linux/aarch64:
[START] Security scan
[DONE] Security scan
## ComfyUI-Manager: installing dependencies done.
** ComfyUI startup time: 2025-05-14 11:32:02.726
** Platform: Linux
** Python version: 3.10.17 (main, Apr 29 2025, 23:47:08) [GCC 12.2.0]
** Python executable: /usr/local/bin/python3
** ComfyUI Path: /app
** ComfyUI Base Folder Path: /app
** User directory: /app/user
** ComfyUI-Manager config path: /app/user/default/ComfyUI-Manager/config.ini
** Log path: /app/user/comfyui.log

[notice] A new release of pip is available: 23.0.1 -> 25.1.1
[notice] To update, run: pip install --upgrade pip

Prestartup times for custom nodes:
   3.3 seconds: /app/custom_nodes/ComfyUI-Manager

Checkpoint files will always be loaded safely.
Total VRAM 62420 MB, total RAM 2055652 MB
pytorch version: 2.4.0
Set vram state to: NORMAL_VRAM
Device: npu
Using pytorch attention
Python version: 3.10.17 (main, Apr 29 2025, 23:47:08) [GCC 12.2.0]
ComfyUI version: 0.3.30
ComfyUI frontend version: 1.17.11
[Prompt Server] web root: /usr/local/lib/python3.10/site-packages/comfyui_frontend_package/static
### Loading: ComfyUI-Manager (V3.32.2)
[ComfyUI-Manager] network_mode: public
### ComfyUI Version: v0.3.30-23-gdbc726f8 | Released on '2025-04-29'
### Loading: ComfyUI-Impact-Pack (V8.14.2)
Import times for custom nodes:
   0.0 seconds: /app/custom_nodes/websocket_image_save.py
   0.0 seconds: /app/custom_nodes/AIGODLIKE-ComfyUI-Translation
   0.1 seconds: /app/custom_nodes/ComfyUI-Manager
   0.3 seconds: /app/custom_nodes/ComfyUI-Impact-Pack
To see the GUI go to: http://0.0.0.0:8188
got prompt
model weight dtype torch.float16, manual cast: None
model_type FLOW
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: npu:0, offload device: cpu, dtype: torch.bfloat16
no CLIP/text encoder weights in checkpoint, the text encoder model will not be loaded.
CLIP/text encoder model load device: npu:0, offload device: cpu, current: cpu, dtype: torch.float16
clip missing: ['text_projection.weight']
Requested to load SD3ClipModel_
loaded completely 60872.46015625 10644.189453125 True
Requested to load SD3
loaded completely 50195.91279296875 15366.797485351562 True
100%|██████████| 20/20 [00:05<00:00,  3.62it/s]
Requested to load AutoencodingEngine
loaded completely 34066.40234375 159.87335777282715 True
Prompt executed in 50.51 seconds
got prompt
100%|██████████| 20/20 [00:02<00:00,  7.32it/s]
Prompt executed in 2.86 seconds
```

四、测试Manager Restart， 导致容器退出的日志（接了一部分推理成功日志，请从Restarting... [Legacy Mode]开始看起，为Manager Restart 失败日志）：

```
got prompt
model weight dtype torch.float16, manual cast: None
model_type FLOW
Using pytorch attention in VAE
Using pytorch attention in VAE
VAE load device: npu:0, offload device: cpu, dtype: torch.bfloat16
no CLIP/text encoder weights in checkpoint, the text encoder model will not be loaded.
CLIP/text encoder model load device: npu:0, offload device: cpu, current: cpu, dtype: torch.float16
clip missing: ['text_projection.weight']
Requested to load SD3ClipModel_
loaded completely 60872.46015625 10644.189453125 True
Requested to load SD3
loaded completely 50195.91279296875 15366.797485351562 True
100%|██████████| 20/20 [00:05<00:00,  3.62it/s]
Requested to load AutoencodingEngine
loaded completely 34066.40234375 159.87335777282715 True
Prompt executed in 50.51 seconds
got prompt
100%|██████████| 20/20 [00:02<00:00,  7.32it/s]
Prompt executed in 2.86 seconds

Restarting... [Legacy Mode]

Command: ['/usr/local/bin/python3', 'main.py', '--listen', '0.0.0.0', '--port', '8188']
[START] Security scan
[DONE] Security scan
## ComfyUI-Manager: installing dependencies done.
** ComfyUI startup time: 2025-05-15 03:30:53.965
** Platform: Linux
** Python version: 3.10.17 (main, Apr 29 2025, 23:47:08) [GCC 12.2.0]
** Python executable: /usr/local/bin/python3
** ComfyUI Path: /app
** ComfyUI Base Folder Path: /app
** User directory: /app/user
** ComfyUI-Manager config path: /app/user/default/ComfyUI-Manager/config.ini
** Log path: /app/user/comfyui.log

[notice] A new release of pip is available: 23.0.1 -> 25.1.1
[notice] To update, run: pip install --upgrade pip

Prestartup times for custom nodes:
   3.3 seconds: /app/custom_nodes/ComfyUI-Manager

Checkpoint files will always be loaded safely.
Traceback (most recent call last):
  File "/app/main.py", line 137, in <module>
    import execution
  File "/app/execution.py", line 13, in <module>
    import nodes
  File "/app/nodes.py", line 22, in <module>
    import comfy.diffusers_load
  File "/app/comfy/diffusers_load.py", line 3, in <module>
    import comfy.sd
  File "/app/comfy/sd.py", line 7, in <module>
    from comfy import model_management
  File "/app/comfy/model_management.py", line 221, in <module>
    total_vram = get_total_memory(get_torch_device()) / (1024 * 1024)
  File "/app/comfy/model_management.py", line 168, in get_torch_device
    return torch.device("npu", torch.npu.current_device())
  File "/usr/local/lib/python3.10/site-packages/torch_npu/npu/utils.py", line 62, in current_device
    torch_npu.npu._lazy_init()
  File "/usr/local/lib/python3.10/site-packages/torch_npu/npu/__init__.py", line 215, in _lazy_init
    torch_npu._C._npu_init()
RuntimeError: Initialize:build/CMakeFiles/torch_npu.dir/compiler_depend.ts:222 NPU function error: c10_npu::SetDevice(device_id_), error code is 507033
[ERROR] 2025-05-15-03:30:59 (PID:9, Device:-1, RankID:-1) ERR00100 PTA call acl api failed
[Error]: Failed to start the device. 
        Rectify the fault based on the error information in the ascend log.
EE1001: [PID: 9] 2025-05-15-03:30:59.713.495 The argument is invalid.Reason: rtGetDevMsg execute failed, reason=[context pointer null]
        Solution: 1.Check the input parameter range of the function. 2.Check the function invocation relationship.
        TraceBack (most recent call last):
        TsdOpen failed. devId=0, tdt error=35[FUNC:PrintfTsdError][FILE:runtime.cc][LINE:2692]
        Start aicpu executor failed, retCode=0x7020009 devId=0[FUNC:DeviceRetain][FILE:runtime.cc][LINE:3475]
        Check param failed, dev can not be NULL![FUNC:PrimaryContextRetain][FILE:runtime.cc][LINE:3224]
        Check param failed, ctx can not be NULL![FUNC:PrimaryContextRetain][FILE:runtime.cc][LINE:3251]
        Check param failed, context can not be null.[FUNC:NewDevice][FILE:api_impl.cc][LINE:2616]
        New device failed, retCode=0x7010006[FUNC:SetDevice][FILE:api_impl.cc][LINE:2640]
        rtSetDevice execute failed, reason=[device retain error][FUNC:FuncErrorReason][FILE:error_message_manage.cc][LINE:53]
        open device 0 failed, runtime result = 507033.[FUNC:ReportCallError][FILE:log_inner.cpp][LINE:161]
        Check param failed, dev can not be NULL![FUNC:DeviceRetain][FILE:runtime.cc][LINE:3527]
        ctx is NULL![FUNC:GetDevErrMsg][FILE:api_impl.cc][LINE:5533]
        The argument is invalid.Reason: rtGetDevMsg execute failed, reason=[context pointer null]
```

五：plog 日志压缩包（百度网盘）：

通过网盘分享的文件：log.tar.gz
链接: https://pan.baidu.com/s/1NlHghozSpoVcNrDbAak7dg?pwd=6ks8 提取码: 6ks8 复制这段内容后打开百度网盘手机App，操作更方便哦

Ascend/pytorch
暂停

内容风险标识

评论 (3)

Ascend/pytorch暂停 .gitee-modal { width: 500px !important; }

内容风险标识