An error inside mmcv/runner/dist_utils.py

[xiangze]

I am trying to run tools/test.py using single GPU on docker with recommended datasets .
At first, I tried

export PYTHONPATH=$(pwd):$PYTHONPATH
python3 ./tools/test.py ./projects/configs/stage1_track_map/base_track_map.py ./ckpts/uniad_base_track_map.pth 

But an error occured.

RuntimeError: context has already been set

to solve this warning, I changed at the end of test.py as in https://stackoverflow.com/questions/61939952/mp-set-start-methodspawn-triggered-an-error-saying-the-context-is-already-be

if __name__ == '__main__':
    # NOTE: To fix the serialization issue in nuScenes-dev-kit, we adopt this method to skip the pickle steps
    torch.multiprocessing.set_start_method('spawn', force=True)
    #torch.multiprocessing.set_start_method('fork')

Then following error message occured.

/usr/local/lib/python3.8/dist-packages/torch/cuda/__init__.py:52: UserWarning: CUDA initialization: CUDA unknown error - this may be due to an incorrectly set up environment, e.g. changing env variable CUDA_VISIBLE_DEVICES after program start. Setting the available devices to be zero. (Triggered internally at  ../c10/cuda/CUDAFunctions.cpp:115.)
  return torch._C._cuda_getDeviceCount() > 0
No CUDA runtime is found, using CUDA_HOME='/usr/local/cuda'
projects.mmdet3d_plugin
Traceback (most recent call last):
  File "./tools/test.py", line 265, in <module>
    main()
  File "./tools/test.py", line 184, in main
    init_dist(args.launcher, **cfg.dist_params)
  File "/usr/local/lib/python3.8/dist-packages/mmcv/runner/dist_utils.py", line 18, in init_dist
    _init_dist_pytorch(backend, **kwargs)
  File "/usr/local/lib/python3.8/dist-packages/mmcv/runner/dist_utils.py", line 29, in _init_dist_pytorch
    rank = int(os.environ['RANK'])
  File "/usr/lib/python3.8/os.py", line 675, in __getitem__
    raise KeyError(key) from None
KeyError: 'RANK'

Is this related to usage of mmcv or other libraries?

working forked repository is https://github.com/xiangze/UniAD/tree/test_with_container
Dockerfile is https://github.com/xiangze/UniAD/tree/test_with_container/docker/Dockerfile