ImagePull Problems With Nvidia Toolkit

Harvester
1

Hello everyone, has anyone experienced issues with vGPU and the nvidia-toolkit container? I am trying to pull from a self-hosted artifactory instance and the container can’t pull the image. This host doesn’t have internet access. It still tries to reach docker.io which it can’t.

nvidia-driver-runtime-22gtq                            0/1     ImagePullBackOff   0               3d18h

Events:
  Type     Reason   Age                      From     Message
  ----     ------   ----                     ----     -------
  Warning  Failed   60m (x260 over 3d18h)    kubelet  Failed to pull image "rancher/harvester-nvidia-driver-toolkit:v1.3-20240613": rpc error: code = DeadlineExceeded desc = failed to pull and unpack image "docker.io/rancher/harvester-nvidia-driver-toolkit:v1.3-20240613": failed to resolve reference "docker.io/rancher/harvester-nvidia-driver-toolkit:v1.3-20240613": failed to do request: Head "https://registry-1.docker.io/v2/rancher/harvester-nvidia-driver-toolkit/manifests/v1.3-20240613": dial tcp 3.219.239.5:443: i/o timeout
  Warning  Failed   15m (x243 over 3d17h)    kubelet  Failed to pull image "rancher/harvester-nvidia-driver-toolkit:v1.3-20240613": rpc error: code = DeadlineExceeded desc = failed to pull and unpack image "docker.io/rancher/harvester-nvidia-driver-toolkit:v1.3-20240613": failed to resolve reference "docker.io/rancher/harvester-nvidia-driver-toolkit:v1.3-20240613": failed to do request: Head "https://registry-1.docker.io/v2/rancher/harvester-nvidia-driver-toolkit/manifests/v1.3-20240613": dial tcp 54.196.99.49:443: i/o timeout
  Normal   BackOff  37s (x21816 over 3d18h)  kubelet  Back-off pulling image "rancher/harvester-nvidia-driver-toolkit:v1.3-20240613"

image
image1230×107 5.41 KB

2

Did you update the Image Repository and Image Tag settings on the nvidia-driver-toolkit screen to point to your private registry? (see Nvidia Driver Toolkit | Harvester)