Getting error in loading shared libraries

[razapk]

Hi,
I am getting the following error:

darknet: error while loading shared libraries: libcuda.so.1: cannot open shared object file: No such file or directory

My Dockerfile looks like:

FROM daisukekobayashi/darknet:gpu-cv-cc86
# I think the lines here are not relevant to the problem
...
darknet detect cfg/yolov3.cfg yolov3.weights data/dog.jpg

Thanks in advance,
Raza

[maininformer]

@razapk
try:

export LD_LIBRARY_PATH="/usr/local/cuda/compat:${LD_LIBRARY_PATH}"

inside the container

Alternatively, you may add the following line to the runtime container; put it in the end

 ENV LD_LIBRARY_PATH="/usr/local/cuda/compat:${LD_LIBRARY_PATH}"

note you will need to build your own container with this method:

• Copy the Dockerfile
• Copy the configure.sh
• run $ chmod +x configure.sh
• run $ docker build --build-arg SOURCE_BRANCH=master --build-arg SOURCE_COMMIT=8bbdb --build-arg CONFIG=gpu-cv-cc75 --tag darknet:latest .

you may replace the --build-args with your desired values.

I am hesitant to PR this and will play with it a bit more first.

[dannje]

Hello,

I'm getting the same error as above:
darknet: error while loading shared libraries: libcuda.so.1: cannot open shared object file: No such file or directory

After running:
docker run --runtime=nvidia --rm -v C:\Users\Dan\Desktop\ocv\dockershit\darknet\build\darknet\x64:/workspace -w /workspace daisukekobayashi/darknet:gpu darknet detector test data/coco.data yolov3.cfg yolov3.weights -i 0 -thresh 0.25 dog.jpg -ext_output

Thanks for your help,
Dan

[daisukekobayashi]

@razapk @dannje
Hi. What OS are you using?
Using —runtime=nvidia option, you need the NVIDIA Container Toolkit. And I think that the NVIDIA Container Toolkit only support Linux distributions. You can see supported platforms here.

In dannje's command, it seems that you are using Windows.

[razapk]

@daisukekobayashi
I am using Ubuntu 20.04.2 LTS and have Nvidia Container Toolkit installed. Here is my docker compose file I am using:

version: "3.9"
services:
  backend:
    build: .
    ports:
      - "3333:3333"
    deploy:
      resources:
        reservations:
          devices:
          - driver: nvidia
            count: 1
            capabilities: [gpu, utility]

[dancronintix]

I got it working @daisukekobayashi
Thank you!

I was on Windows 10 attempting to use WSL which doesn't have CUDA. I needed to install WSL2 and that worked. 👍