Why am I getting this error while running Docker Container?

Docker Desktop
1

I am following this repository to setup the environment for radar semantic segmentation. GitHub - valeoai/MVRSS: [ICCV 2021] Multi-View Radar Semantic Segmentation

I am using Docker to install all the dependencies. Particularly at step 2, when it comes to running a container, i am facing the following error.

Command:
docker run -d --ipc=host -it -v /home/shahidnawaz/datasets:/home/datasets_local -v /home/shahidnawaz/logs:/home/logs --name mvrss --gpus all mvrss:Dockerfile sleep infinity

Error:
docker: Error response from daemon: failed to create task for container: failed to create shim task: OCI runtime create failed: runc create failed: unable to start container process: error during container init: error running hook #0: error running hook: exit status 1, stdout: , stderr: Auto-detected mode as ‘legacy’ nvidia-container-cli: initialization error: load library failed: libnvidia-ml.so.1: cannot open shared object file: no such file or directory: unknown.

I am new to using these tools, I have already tried different solutions available online. Any help regarding solution would be much appreciated. Thanks.

2

You are trying to run a 3 year old piece of software and it seems to be missing a library.

Did you build the image using the Dockerfile yourself? Did you upgrade it from Ubuntu 18?

3

I built the image myself using Dockerfile. I am using Ubuntu 22.04.

4

Did you update the Dockerfile? Because that uses Ubuntu 18.

5

yes, i updated the Dockerfile. Additionally, i modified some versions of dependencies as well since the versions mentioned in Dockerfile were not available.

6

What did you find when you searched for the error message?

Also is it possible to use a GPU with Docker Desktop for Linux as well? I know it is supported on Windows because WSL2 supports it, but I haven’t really used that either. Since Docker Desktop runs a virtual machine, your GPU will probably not exist in the virtual machine, only on the host.