I have a pod, that consists of just one container. This container has the following resource section:
resources:
limits:
cpu: "4"
memory: 58Gi
nvidia.com/gpu: 1
requests:
cpu: 3500m
memory: 55GiWhen I run this pod, it starts successfully and all is good until I try to copy a file there. If I run kubectl cp foo.txt my_pod:/app, I get the following error:
tar: This does not look like a tar archive
tar: Skipping to next header
tar: Exiting with failure status due to previous errorsHowever, if I omit any resource requirements(i.e. comment the resource section out), cp works as expected.
I did some investigation and found out that the problem is only present when nvidia.com/gpu: 1 is specified in the limits. Colleagues told me it is a known problem, but no one knows why is that.
Does anyone know? What is the best workaround here?