Docker cant find libcusparse.so.11

[Boehm92]

Hi All,

i have problems to deploy my GNN to docker. My code runs perfectly in my virtual env.

I created this docker file:

FROM pytorch/pytorch:1.10.0-cuda11.3-cudnn8-runtime

WORKDIR /app
COPY ./requirements.txt /app/requirements.txt

RUN pip3 install --no-cache-dir --upgrade pip setuptools && \
    ln -s /usr/include/locale.h /usr/include/xlocale.h && \
    pip install torch==1.10.0+cu113 -f https://download.pytorch.org/whl/cu113/torch_stable.html && \
    pip install torch-scatter==2.0.9 torch-sparse==0.6.12 torch-cluster==1.5.9 torch-spline-conv==1.2.1 torch-geometric==2.0.3 -f https://data.pyg.org/whl/torch-1.10.0+cu113.html && \
    pip3 install --no-cache-dir -r requirements.txt && \
    rm /usr/include/xlocale.h

COPY . .
ENV PYTHONPATH `pwd`/..

CMD [ "python3", "Example" ]

The container is build but when running it this error message appears:

File "__main__.py", line 4, in <module>
from gnnProject import DataImporter
from torch_sparse import SparseTensor
File "/opt/conda/lib/python3.7/site-packages/torch_sparse/__init__.py", line 19, in <module>
torch.ops.load_library(spec.origin)
File "/opt/conda/lib/python3.7/site-packages/torch/_ops.py", line 110, in load_library
ctypes.CDLL(path)
File "/opt/conda/lib/python3.7/ctypes/__init__.py", line 364, in __init__
self._handle = _dlopen(self._name, mode)
OSError: libcusparse.so.11: cannot open shared object file: No such file or directory

This error is also discussed in "#2322" but there is no answer for docker in it.
I would be greatfull for any advice

Kind Regards
Stefan

[rusty1s]

You might be able to properly link the installed CUDA 11.3 to LD_LIBRARY_PATH.