LazyRESTMapper KindsFor does not return all versions when single version is previously queried

[tenstad]

Problem

addKnownGroupAndReload is only triggered when no matches are found. If priming the mapper with a single version and then requesting all versions (not specifying version), a match is found and only a single version returned (not multiple versions).

func (m *mapper) KindsFor(resource schema.GroupVersionResource) ([]schema.GroupVersionKind, error) {
	res, err := m.getMapper().KindsFor(resource)
	if meta.IsNoMatchError(err) {
		if err := m.addKnownGroupAndReload(resource.Group, resource.Version); err != nil {

Test Case

The test passes if lazyRestMapper.KindsFor(schema.GroupVersionResource{Group: "autoscaling", Version: "v1", Resource: "horizontalpodautoscaler"}) is removed, thus making the other KindsFor trigger a addKnownGroupAndReload.

t.Run("LazyRESTMapper should be able to fetch all versions after fetching single version", func(t *testing.T) {
	g := gmg.NewWithT(t)

	httpClient, err := rest.HTTPClientFor(restCfg)
	g.Expect(err).NotTo(gmg.HaveOccurred())

	lazyRestMapper, err := apiutil.NewDynamicRESTMapper(restCfg, httpClient)
	g.Expect(err).NotTo(gmg.HaveOccurred())

	_, err = lazyRestMapper.KindsFor(schema.GroupVersionResource{Group: "autoscaling", Version: "v1", Resource: "horizontalpodautoscaler"})
	g.Expect(err).NotTo(gmg.HaveOccurred())

	kinds, err := lazyRestMapper.KindsFor(schema.GroupVersionResource{Group: "autoscaling", Resource: "horizontalpodautoscaler"})
	g.Expect(err).NotTo(gmg.HaveOccurred())
	g.Expect(len(kinds)).To(gmg.BeNumerically(">", 1))
})

Suggestion

Trigger addKnownGroupAndReload when schema.GroupVersionResource input is partial, and some fields are unspecified.

[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

[sbueringer]

/triage accepted

We fixed a somewhat related issue recently via: #2663. But I think this doesn't change the problem described here.