clarify support of new features across version-skewed clusters

[liggitt]

Follow up from #11060, tracked in #12329

How users are expected to manage use of new features in version-skewed clusters is not clear in user-facing documentation:

• what statement should be made about new features in version 1.x when running with a 1.x+ control plane and kubelets older than 1.x? is the user responsible for labeling nodes and workloads to only steer pods using those features to nodes with those capabilities? @kubernetes/sig-node-pr-reviews @kubernetes/sig-architecture-pr-reviews

Page to Update:
https://kubernetes.io/docs/setup/version-skew-policy/

[roberthbailey]

This is difficult because it depends on the feature. Some features only require an updated control plane (e.g. things that are wholly implemented in the apiserver such as support for CRD versions or in the schedule like multiple scheduler support) while others require kubelets to pick up a new version.

It seems like each feature should describe the minimum required version for the feature to be used, and also indicate whether that version applies only to the control plane or also to the nodes. For features that require nodes to be updated, labeling can be helpful, but it would be better if the feature owner implemented the feature to degrade gracefully in the scenario where the control plane version is ahead of the node version.

[imkin]

/wg lts

[sftim]

/language en

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[sftim]

/kind feature
/priority backlog

[fejta-bot]

Stale issues rot after 30d of inactivity.
Mark the issue as fresh with /remove-lifecycle rotten.
Rotten issues close after an additional 30d of inactivity.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle rotten