Add kube-score/skip annotation

[romnn]

Thank you very much for this fantastic tool!!

I came across the use case to disable all checks for a file, more precisely the Helm test-connnection pod often found in Helm charts.

This PR proposes skipping all checks for a file ("kube-score/skip": "true") and checks ("kube-score/ignore": "*") using annotations. The root level annotation is checked on a per-file basis during parsing and tracked in the FileLocation, which allows further differentiation between ignored checks and skipped files. Additionally, nested annotations are checked on a per-check basis similar to the kube-score/ignore annotation.

I greatly appreciate any feedback if this is something you agree could be useful.
Best, Roman

Note: Unfortunately, the tests do not pass due an unrelated issue that I am not familiar with, once this is resolved I plan to merge the fix back into this PR.

RELNOTE: Add support for `"kube-score/skip": "true"` and`"kube-score/ignore": "*"` annotations.

[vercel]

@romnn is attempting to deploy a commit to the Gustav Westling's projects Team on Vercel.

A member of the Team first needs to authorize it.

[zegl]

Hey @romnn, thanks for the PR. What would be the difference between the "kube-score/skip": "true" and the "kube-score/ignore": "*" annotation? How come we need to do both? Could you please share some examples?

[romnn]

Thanks for the response and good question! They are equivalent in that both skip all checks for a resource, so supporting only one of them is also an option.

"kube-score/ignore": "*" is a very simple implementation, where "*" just matches each check in the ScoredObject.isEnabled function. Hence, it works on a per-check basis. "kube-score/skip": "true" is a root level annotation and is tracked during parsing on a per-file basis, which required more changes.

There are some subtle differences:

• "kube-score/skip": "true" will output SKPPED for the file once, while "kube-score/ignore": "*" will output SKIPPED for each check that would normally be performed for the resource.
• "kube-score/skip": "true is only checked for root-level annotations, while "kube-score/ignore": "*" can be applied to nested resource annotations

I agree that there could be some confusion around this. Both are intuitive in my opinion, but "kube-score/ignore": "*" uses an already existing annotation which could maybe be easier for adoption.

I'd gladly hear your opinion on this!