Update v1.Taint parser to accept the form key:effect and key=:effect-

[dlipovetsky]

/kind bug

What this PR does / why we need it:
Updates v1.Taint parser to accept the form key:effect, which is the string representation of a v1.Taint that has a nil Value (the empty string).

Also updates the parser to accept the form key=:effect-, since key=:effect is a valid string representation of a v1.Taint with a nil Value.

• Update code and tests in pkg/util/taints/
• Update kubectl's copy of the code and tests in pkg/kubectl/cmd/taint/
• Update kubectl taint usage

Which issue(s) this PR fixes:
Fixes #73249

Does this PR introduce a user-facing change?:

Support parsing more v1.Taint forms. `key:effect`, `key=:effect-` are now accepted.

/cc @liggitt

[dlipovetsky]

/sig scheduling

[dlipovetsky]

/cc @kubernetes/sig-scheduling-bugs
/cc @k82cn
/cc @ravisantoshgudimetla

[neolit123]

thanks @dlipovetsky
/priority important-longterm

could you please add a user facing release note instead of NONE? e.g.

Support parsing v1.Taint forms such as xx, yy

[dlipovetsky]

Thanks @neolit123! Added release note.

[dlipovetsky]

I added two new invalid spec tests and fixed a bug in the updated parser. (I also verified that the original parser passed these new tests)

[dlipovetsky]

Once the changes to pkg/util/taints/ are finalized, I'll make corresponding changes to pkg/kubectl/cmd/taint/.

[Huang-Wei]

LGTM for the pkg/util/taints/ part.

[kevin-wangzefeng]

LGTM for pkg/util/taints/, I think you can go ahead for the rest parts.

[dlipovetsky]

Thanks @kevin-wangzefeng and @Huang-Wei.

After I updated parseTaints and parseTaint in pkg/kubectl/cmd/taint/util.go, an existing test in pkg/kubectl/cmd/taint/taint_test.go failed:

--- FAIL: TestTaint (0.03s)
    --- FAIL: TestTaint/node_has_two_taints_with_the_same_key_but_different_effect,_remove_all_of_them_with_wildcard (0.00s)
        taint_test.go:313: Recovered: error: invalid taint spec: dedicated
            See 'taint -h' for help and examples
        taint_test.go:332: node has two taints with the same key but different effect, remove all of them with wildcard: node not tainted

I did not realize that kubectl taint accepts a key with no value or effect to remove a taint; the example in the kubectl usage is:

  # Remove from node 'foo' all the taints with key 'dedicated'
  kubectl taint nodes foo dedicated-

This use case was introduced in #30590. To be backward compatible, this PR should not break this use case.

There was no test in pkg/util/taints for removing a taint using the key- form. The parser prior to this PR accepts this form, but the parser in this PR rejects this form. My plan is to:

1. Add a test to pkg/util/taints for removing a taint using the key- form
2. Update the parser in this PR to accept the key- form

[dlipovetsky]

The last commit passes the tests in pkg/util/taints and pkg/kubectl/cmd/taint/.

I have changed parseTaint to accept the key form, and ensured in ParseTaints that a taint with no effect cannot be added. The alternative is for parseTaint to continue rejecting this form, and handling the form as a special case in ParseTaints. I prefer the former, because the code already passes around taints without values or effects, which implies that the key form is valid, although it must not be used to create a taint.

[kevin-wangzefeng]

I'm fine with it. Empty/null value and effect should not block parsing, though taints with only keys indicated should not be used for creating.

[bsalamat]

I reviewed and it looks fine. I have only a few minor comments. If @dlipovetsky addresses them sooner, we can get it merged.

[dlipovetsky]

@bsalamat Thanks a lot for your review. My apologies for taking a while to address your comments. I address them using separate commits for the code in pkg/util/taints and pkg/kubectl/cmd/taint, so that I can easily fixup the original commits for those packages before merging.

[dlipovetsky]

I also added a test to verify that the empty string is an invalid Taint spec.

[dlipovetsky]

Test failed due to known flake, see #74931

/retest

[bsalamat]

Thanks, @dlipovetsky! Looks good. Could you please squash commits?

/lgtm

[bsalamat]

I updated the release note. Please check it.

[dlipovetsky]

@bsalamat Squashed.

As for the release note, I'm not sure it needs to mention the key- form, since that has always been accepted--this PR just changes where its parsed internally (please see #74159 (comment)).

[bsalamat]

@dlipovetsky Thanks again. I fixed the release notes.

/lgtm
/approve

[bsalamat]

@liggitt Back to you for approval.

[dlipovetsky]

/assign @liggitt

(So that this PR shows up in your dashboard 😄)

[liggitt]

/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: bsalamat, dlipovetsky, liggitt

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• pkg/OWNERS [liggitt]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[dlipovetsky]

@ravisantoshgudimetla Please cancel the hold. Thanks!

[ravisantoshgudimetla]

/hold cancel

Done @dlipovetsky. Thanks for working on this PR.

[fejta-bot]

/retest
This bot automatically retries jobs that failed/flaked on approved PRs (send feedback to fejta).

Review the full test history for this PR.

Silence the bot with an /lgtm cancel or /hold comment for consistent failures.

[fejta-bot]

/retest
This bot automatically retries jobs that failed/flaked on approved PRs (send feedback to fejta).

Review the full test history for this PR.

Silence the bot with an /lgtm cancel or /hold comment for consistent failures.

[fejta-bot]

/retest
This bot automatically retries jobs that failed/flaked on approved PRs (send feedback to fejta).

Review the full test history for this PR.

Silence the bot with an /lgtm cancel or /hold comment for consistent failures.