Follow-ups for CRD versioning

[sttts]

Everything that must be done after #63830:

• missing tests:
  • a test that actually checked what got persisted in etcd matched the storage version https://github.com/kubernetes/kubernetes/pull/63830/files#r189712166
  • a test that set up a CRD, persisted some data, changed the version, and made sure the previously persisted data could be read https://github.com/kubernetes/kubernetes/pull/63830/files#r189712166
  • a test that verified discovery docs tracked the CRD through creation, version addition, version removal, and deletion https://github.com/kubernetes/kubernetes/pull/63830/files#r189712166
  • a test that tests that spec.served is respected https://github.com/kubernetes/kubernetes/pull/63830/files#r189712166
• @liggitt fix possibly flaking watches: https://github.com/kubernetes/kubernetes/pull/63830/files#r189710254 CRD Test: Fix resource creation/deletion for multi-version CRDs #64796 Consume watch event for all versions of CRD #65288
• @sttts: split nopConverter into orthogonal wrappers
  • deep list converter calling an inner delegate
  • special CRD ConvertFieldLabel
  • the actual unstructured conversion PR apiextensions: extract orthortogonal behaviour from nopConve #64208
• @sttts: reuse the old NewVersionedNoxuInstance, instantiateVersionedCustomResource, testSimpleVersionedCRUD (and extend it with the version string) https://github.com/kubernetes/kubernetes/pull/63830/files#r189814105 PR apiextensions: unify mono- and multi-versioned test helpers #64204
• @sttts: Define GA in API documentation of both APIServices and CRD Versions https://github.com/kubernetes/kubernetes/pull/63830/files#r189812365 – PR: Fix documentation of GA is in APIService's PriorityVersion and CRD's Versions #64215
• @sttts: Better logging for removeDeadStorage https://github.com/kubernetes/kubernetes/pull/63830/files#r189808872 PR: apiextensions: reduce verbose logs in removeDeadStorage #64202

[mbohlool]

I don't think split customresource_handler.go. This is far too big. Maybe we should create a CR scheme. falls into a follow up on #63830. It deserve its own issue. I think we must track things we missed/postponed to follow up in #63830 here.

[mbohlool]

@liggitt The link to the comment on flaky watch tests are not working. If you have the related issue/pr link handy, can you please put it here?

[sttts]

@mbohlool agree about "split customresource_handler.go". I want to approach that after 1.11 is out. But this has not too much to do with versioning.

[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

[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

[nilebox]

@sttts @mbohlool CRD versioning is still no-op (no webhook support) AFAIK? Should this issue be un-rotten?

[liggitt]

/lifecycle frozen

[jpbetz]

I'll follow up on the missing tests. My basic plan is to add integration testing of conversion webhooks based on how we integration admission webhooks in test/integration/apiserver/admissionwebhook/admission_test.go. That way we'd have the conversion webhook handler in the integration test code (not as a published docker image like we use for e2e), once that's in place it should be relatively straightfoward to add the test cases described in this issue description. Feedback welcome!