MultiValidatingHandler returns as soon as one handler fails, making it impossible to follow convention of reporting all validation errors

[dlipovetsky]

The prevailing convention in Kubernetes resource validation is to report all validation errors, not only the first discovered error. The built-in API types follow this convention (example). The OpenAPI and CEL rule validation follows this convention, too. Most webhooks I have seen also follow it.

We provide a utility function for webhook authors that executes multiple validators: https://github.com/kubernetes-sigs/controller-runtime/blob/6ad5c1dd4418489606d19dfb87bf38905b440561/pkg/webhook/admission/multi.go#L90-L95

It returns as soon as one validator fails. That means that subsequent validators are not called, and any errors they might discover are not reported.

I think we should provide an alternative implementation that calls all validators, even if some fail, and aggregates their errors.

Also, because the existing utility function does not follow the convention, I think we should consider deprecating it.

[dlipovetsky]

An implementation that aggregates errors from multiple handlers would have to merge handler responses., and I don't see a unambiguous way to do that.

[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

[dlipovetsky]

/remove-lifecycle rotten

It turns out that the API server itself does not return errors from all validating webhooks:https://github.com/kubernetes/kubernetes/issues/132637.

We're working on a solution to that. Once we're done, I think we can apply what've learned here.

[sbueringer]

/remove-lifecycle rotten

[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