Containers that use cuda images in k8s do not have gpu resources, but the process id can be seen using nvidia-smi

[ZYWNB666]

1、The container used is " tensorflow/tensorflow:2.16.1-gpu "
2、There is no configuration for the gpu application in the yaml file

3、You can see the pod pid in the nvidia-smi command
It seems that this pod will be displayed on all free graphics cards, I don't use any gpu virtualization tools, I don't know if it takes up this video card

[ZYWNB666]

@klueska Dude, do you have time to look at my problem?

[ZYWNB666]

@elezar Dude, do you have time to look at my problem?

[github-actions]

This issue is stale because it has been open 90 days with no activity. This issue will be closed in 30 days unless new comments are made or the stale label is removed.