failed to build docker images

i am trying to build a docker image in a VM with A40 pass-through.

os version: Ubuntu 22.04 cuda version: 12.1, V12.1.105, gpu： A40

commnd: sudo docker build -t comfy3d . --build-arg "HTTP_PROXY=http://127.0.0.1:7890 " --build-arg "HTTPS_PROXY=http://127.0.0.1:7890" --build-arg "NO_PROXY=localhost,127.0.0.1,.example.com" --network=host --no-cache

i got error as follows.

=> ERROR [13/19] RUN pip install --no-cache -r requirements.txt     -r requirements_post.txt     ninja scikit-learn rembg[gpu] open_clip_torch                                2.7s
------
 > [13/19] RUN pip install --no-cache -r requirements.txt     -r requirements_post.txt     ninja scikit-learn rembg[gpu] open_clip_torch:
0.411 Defaulting to user installation because normal site-packages is not writeable
0.480 Processing ./tgs/models/snowflake/pointnet2_ops_lib
0.482   Preparing metadata (setup.py): started
2.319   Preparing metadata (setup.py): finished with status 'error'
2.327   error: subprocess-exited-with-error
2.327
2.327   × python setup.py egg_info did not run successfully.
2.327   │ exit code: 1
2.327   ╰─> [16 lines of output]
2.327       No CUDA runtime is found, using CUDA_HOME='/usr/local/cuda'
2.327       Traceback (most recent call last):
2.327         File "<string>", line 2, in <module>
2.327         File "<pip-setuptools-caller>", line 34, in <module>
2.327         File "/app/custom_nodes/ComfyUI-3D-Pack/tgs/models/snowflake/pointnet2_ops_lib/setup.py", line 20, in <module>
2.327           os.environ["TORCH_CUDA_ARCH_LIST"] = ".".join(map(str, torch.cuda.get_device_capability()))
2.327                                                                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2.327         File "/home/user/.local/lib/python3.11/site-packages/torch/cuda/__init__.py", line 439, in get_device_capability
2.327           prop = get_device_properties(device)
2.327                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
2.327         File "/home/user/.local/lib/python3.11/site-packages/torch/cuda/__init__.py", line 453, in get_device_properties
2.327           _lazy_init()  # will define _get_device_properties
2.327           ^^^^^^^^^^^^
2.327         File "/home/user/.local/lib/python3.11/site-packages/torch/cuda/__init__.py", line 302, in _lazy_init
2.327           torch._C._cuda_init()
2.327       RuntimeError: Found no NVIDIA driver on your system. Please check that you have an NVIDIA GPU and installed a driver from http://www.nvidia.com/Download/index.aspx
2.327       [end of output]
2.327
2.327   note: This error originates from a subprocess, and is likely not a problem with pip.
2.329 error: metadata-generation-failed
2.329
2.329 × Encountered error while generating package metadata.
2.329 ╰─> See above for output.
2.329
2.329 note: This is an issue with the package mentioned above, not pip.
2.329 hint: See above for details.
------
Dockerfile:67
--------------------
  66 |     COPY --chown=user:user tgs/ tgs/
  67 | >>> RUN pip install --no-cache -r requirements.txt \
  68 | >>>     # post requirements installation require gpu, setup
  69 | >>>     # `nvidia-container-runtime`, for docker, see
  70 | >>>     # https://stackoverflow.com/a/61737404
  71 | >>>     -r requirements_post.txt \
  72 | >>>     # those seem to be missed
  73 | >>>     ninja scikit-learn rembg[gpu] open_clip_torch
  74 |
--------------------
ERROR: failed to solve: process "/bin/sh -c pip install --no-cache -r requirements.txt     -r requirements_post.txt     ninja scikit-learn rembg[gpu] open_clip_torch" did not complete successfully: exit code: 1

thanks a lot for your help.

MrForExample / ComfyUI-3D-Pack

failed to build docker images #141