Migrate away from google.com gcp project gke-release

[spiffxp]

Part of umbrella issue to migrate away from google.com gcp projects: #1469

Part of umbrella to migrate kubernetes e2e test images/registries to community-owned infrastructure: #1458

There are a variety of gcr.io/gke-release images referenced in kubernetes/kubernetes CI (and sundry kubernetes-sigs projects). These should be migrated to k8s.gcr.io where possible.

Suggested path forward:

• Gather the list of images from cs.k8s.io: https://cs.k8s.io/?q=gcr.io%2Fgke-release&i=nope&files=&repos=
• For each image
  • determine whether there is an equivalent in k8s.gcr.io
  • if not, work with subproject owner to setup a staging project if they need one; build / stage / promote to k8s.gcr.io
  • change reference to images to k8s.gcr.io
• announce deprecation window
• flag project for deletion post-deprecation window

/wg k8s-infra
/area artifacts
/sig testing
/sig release
/sig storage
(I see a number of csi images)
/sig windows
(I see a pause-win image)
/area release-eng

[spiffxp]

/milestone v1.21

[claudiubelu]

We've already added Windows support to the pause image, and it has already been built and promoted. Here's a PR that will replace the windows pause image references: kubernetes/kubernetes#98205

[cpanato]

@spiffxp I can help here will assign to myself to start the work

/assign

[cpanato]

Some questions:

The following images do not exist in the community registry and it is defined in this file but the tags do not exist on those files

• gcr.io/gke-release/nvidia-gpu-device-plugin@sha256:a75ec0caa9e3038bd9886b3f36641a624574ff34b064974de6ee45048de3372b
• gcr.io/gke-release/gcp-compute-persistent-disk-csi-driver:v1.0.1-gke.0

Will be easier we just copy those images to the community registry? or where this is defined so we can set up the appropriate jobs to build those images and push them to the correct place.

---

And the final one is related to some go files

those files: https://github.com/kubernetes/kubernetes/blob/master/test/utils/image/manifest_test.go and https://github.com/kubernetes/kubernetes/blob/master/test/utils/image/manifest.go
define the gcr.io/gke-release what should be the replacement for that?

cc @spiffxp if you can help me on that or point who can help

mentioning @LappleApple for visibility and maybe for future support.

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[ameukam]

/milestone clear
/remove-lifecycle stale
/priority important-longterm

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[k8s-triage-robot]

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle rotten
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle rotten

[cpanato]

/remove-lifecycle rotten

[BenTheElder]

We're not pushing to this, but we are using images published there for GCP e2e clusters, which is OK-ish in context.

[k8s-triage-robot]

The Kubernetes project currently lacks enough contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle stale
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle stale

[k8s-triage-robot]

The Kubernetes project currently lacks enough active contributors to adequately respond to all issues.

This bot triages un-triaged issues according to the following rules:

• After 90d of inactivity, lifecycle/stale is applied
• After 30d of inactivity since lifecycle/stale was applied, lifecycle/rotten is applied
• After 30d of inactivity since lifecycle/rotten was applied, the issue is closed

You can:

• Mark this issue as fresh with /remove-lifecycle rotten
• Close this issue with /close
• Offer to help out with Issue Triage

Please send feedback to sig-contributor-experience at kubernetes/community.

/lifecycle rotten

[ameukam]

/remove-lifecycle rotten
/milestone v1.33

[BenTheElder]

This is basically done:
https://cs.k8s.io/?q=gcr.io%2Fgke-release&i=nope&files=&repos=

I see two images:

1. gcr.io/gke-release/nvidia-partition-gpu (in test/e2e/testing-manifests/gpu/gce/nvidia-driver-installer.yaml and vendored copies)

• this is used to setup nvidia GPUs on GCE for testing GPU integrations. It is not released to end-users

2. gcr.io/gke-release/gcp-filestore-csi-driver:v0.2.0-gke.0 (in deploy/kubernetes/manifests/controller.yaml and copies in other repos)

• this is being released to registry.k8s.io like registry.k8s.io/cloud-provider-gcp/gcp-filestore-csi-driver:v1.7.0, we should probably cleanup the stale ref https://github.com/kubernetes-sigs/gcp-filestore-csi-driver/tree/a0dd64751b62f82d18b01ebef68b95686417d942#project-status

3. There's one other non-specific stale ref in https://github.com/kubernetes-sigs/aws-fsx-csi-driver/blob/65519d879a612994bc225df6fdd1fed4a4d72ed5/tests/e2e/vendor/k8s.io/kubernetes/test/utils/image/manifest.go#L103 but it doesn't appear to be used and has no specific image.

I would argue we should cleanup 2), not block on 3) and 🤷 on 1), it's difficult to setup nvidia GPUs on only kubernetes-project hosted images, if we wind up with an issue with this image we'll deal with it, one of the Googlers active in SIG node can help us find a current image if/when we need.