Developers Nvidia Support
Index
issue #2668
Whether in unibotics, launching robotics academy locally or even launching the backend as a developer, the vnc windows are not updated correctly when changing the exercise world.
Update
A colleague solved the problem and the vnc viewers update correctly, the problem now is that every time you try to change the world of an exercise, it tries to close processes that have already been closed giving rise to errors, in addition to a residual error message in the vnc viewer.
issue #2508
When using the -g
option in the developer options, the integrated graphics was used instead of nvidia, I didn’t really notice this until some test results didn’t add up.
Solving issue
When launching the developer script you can use the options -g
to use the integrated graphics card or -n
to use the nvidia graphics card. Before you start, make sure you have the NVIDIA Container Toolkit installed.
Now we will have to install the nvidia runtime to use it with our docker:
1
2
sudo apt-get update
sudo apt-get install -y nvidia-docker2
Now we will check if docker recognises nvidia as a new runtime (restarting the docker service to update the new configuration):
1
2
sudo systemctl restart docker
docker info | grep -i runtime
It will most likely not recognise it, so we will have to do it manually by editing or creating the /etc/docker/daemon.json
file:
1
2
3
4
5
6
7
8
{
"runtimes": {
"nvidia": {
"path": "nvidia-container-runtime",
"runtimeArgs": []
}
}
}
It is also possible that nvidia-runtime is not installed, check and install it if it is not.
1
dpkg -l | grep nvidia-container-runtime
If it is not installed:
1
sudo apt-get install -y nvidia-container-runtime
Now everything should be ready to start using nvidia with our dockers, restart the docker service to update the configuration and check that everything works correctly.
1
sudo systemctl restart docker