versions.mk: change REGISTRY, add comment

[jgehrcke]

This is a required follow-up after https://github.com/NVIDIA/k8s-dra-driver-gpu/pull/281/files.

Another correction on this PR is a rename which was done in the README, but not in actuality: 1632380#diff-b335630551682c19a781afebcf4d07bf978fb1f8ac04c6bf87428ed5106870f5R64-R76

[copy-pr-bot]

This pull request requires additional validation before any workflows can run on NVIDIA's runners.

Pull request vetters can view their responsibilities here.

Contributors can view more details about this message here.

[Copilot]

Pull Request Overview

This PR updates the registry URL value referenced in the configuration documentation, following up on previous changes to align all references.

• Change REGISTRY value in README from "nvcr.io/nvidia/cloud-native" to "nvcr.io/nvidia".
• Align documentation with the intended registry naming convention.

Comments suppressed due to low confidence (1)

README.md:150

• The updated registry value no longer includes the '-cloud-native' suffix; please confirm that this change is intentional and update any necessary documentation or comments to reflect this new configuration.

REGISTRY=nvcr.io/nvidia

[elezar]

Question: do the demo scripts depend on this?

[jgehrcke]

Yes. REGISTRY is used in IMAGE_NAME which is scattered throughout Makefile targets.

I did veeery basic validation as part of this patch:

Did grep -nR cloud-native to see if there are other, obvious places that need changing. Answer is no.

Ran ./demo/clusters/kind/build-dra-driver-gpu.sh (renamed in this PR) and it worked. Part of its output was:

...
 => => writing image sha256:25c33684877a5fdd5d360138e0080d03f281b55e6e43a09baa9b3bb8ebeec471                                                                       0.0s
 => => naming to nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1-ubi9                                                                                               0.0s
docker image inspect nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1-ubi9 > /dev/null || docker pull nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1-ubi9
docker tag nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1-ubi9 nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1
+ cd /home/jgehrcke/dev/k8s-dra-driver-gpu/demo/clusters/kind/scripts
...
++ : kindest/node:v1.32.0
+ kind load docker-image --name k8s-dra-driver-gpu-cluster nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1
Image: "nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1" with ID "sha256:25c33684877a5fdd5d360138e0080d03f281b55e6e43a09baa9b3bb8ebeec471" not yet present on node "k8s-dra-driver-gpu-cluster-worker", loading...
Image: "nvcr.io/nvidia/k8s-dra-driver-gpu:v25.3.0-rc.1" with ID "sha256:25c33684877a5fdd5d360138e0080d03f281b55e6e43a09baa9b3bb8ebeec471" not yet present on node "k8s-dra-driver-gpu-cluster-control-plane", loading...
+ set +x

which I think confirms that the image was built and then 'loaded' in the kind environment, using the new location / REGISTRY.