KEP-5073: Runtime verification testing #2
Conversation
6543525 to
8778716
Compare
8778716 to
0d03172
Compare
|
Thanks. This would give me a lot more confidence in deciding that generated validation is equivalent to handwritten validation with a release of "oops, let's put that back" being possible. |
There was a problem hiding this comment.
+1 on adding runtime verification and the associated feature gate. LGTM! I'm fine with merging this as is I believe that there are a couple of other places to touch to include this new feature gate:
- kep.yaml
- a # of places where the feature gates for this KEP is mentioned in the KEP itself
I'm happy to merge this as is and update those areas as I go through and make additional changes from feedback on the KEP.
| such that the presence of mismatches between declarative validation and hand written validation can | ||
| be safely checked against production workloads. | ||
|
|
||
| When a `DeclarativeValidationShadowMetrics` feature gate is enabled, the following will be collected for each validation operation: |
There was a problem hiding this comment.
Talked offline, might make sense to update the name suggested there: DeclarativeValidationMismatchMetrics
|
Thanks @aaron-prindle. Feel free to merge into your PR and continue refinement there |
3f616dd
into
aaron-prindle:declarative-validation-with-validation-gen
KEP-5073: Runtime verification testing
…dation-gen (kubernetes#5074) * KEP-5073: Declarative Validation of Kubernetes Native Types With validation-gen * resolving KEP comments * resolving KEP comments round #2 * resolving KEP comments round kubernetes#3 * update KEP-5073 to clarify migration details * fix prototype link in a # of places * Propose a runtime verification check * Update TOC * move runtime check section * plumb DeclarativeValidationMismatchMetrics feature gate through KEP * resolving KEP comments round kubernetes#4 * update reviewers and approvers * resolving KEP comments round kubernetes#5 * resolving KEP comments round kubernetes#6 * remove recommended and unresovled from zero values + defaulting section * fix metric name --------- Co-authored-by: yongruilin <yongrlin@outlook.com> Co-authored-by: Joe Betz <jpbetz@google.com>
This was proposed to me by @deads2k.
I think it offers a good way to manage risk when Declarative Validation is rolled out to production clusters/fleets.