gs://k8s-artifacts-cni is expensive and probably unnecessary

[BenTheElder]

We were digging through the GCP billing in #sig-k8s-infra today and it turns the egress bandwidth from re-hosting https://github.com/containernetworking/plugins to gs://k8s-artifacts-cni is the largest line item after compute, by far. It's easily like 17-20k/mo.
... we should phase this out and just pull from the official upstream releases for kube-up etc.

thread in https://kubernetes.slack.com/archives/C2C40FMNF/p1733349220696569

@upodroid noticed this when we were looking into the GCS spend not continuing to shrink following the full k8s.gcr.io redirect

@upodroid suggested in the thread we could start by changing the bucket type to reduce the cost

I think longterm we want to identify what little is using this (we took a peek in cs.k8s.io, kube-up is one of the main users, but also possibly kops) and instead just use the official upstream binary releases

[BenTheElder]

/sig k8s-infra
/priority important-longterm

[BenTheElder]

/sig release

[BenTheElder]

@upodroid added labels so we can collect data on this more easily.

There is some dwindling cost from the GCRs in this project and also cri tools which are both already being phased out AFAIK.

[BenTheElder]

This was discussed yesterday in SIG K8s Infra with @justinsb + SIG K8s Infra leads in particular.

We do NOT want to break anyone immediately, and we can afford not to, but we do want to exit this sort of thing (any lingering third party hosting, CI ...) long term.

Commented in kops at:
kubernetes/kops#17162 (comment)

[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

[BenTheElder]

/lifecycle frozen
This remains true until it is resolved.
Auto-closing the issue is unhelpful 🙄

[BenTheElder]

I think we've migrated usage within the project?

We can think about a timeline to wind down the bucket now ...