chore: upgrade kubernetes dependencies

[sagikazarmark]

Signed-off-by: Mark Sagi-Kazar [email protected]

What this PR does / why we need it:

Upgrades the Kubernetes libraries to 1.19.2
Keeping libraries up to date is generally a good idea. I'm not aware of any major changes that might break backwards compatibility.

Which issue(s) this PR fixes (optional, in fixes #<issue number>(, fixes #<issue_number>, ...) format, will close the issue(s) when PR gets merged):
Fixes #

Special notes for your reviewer:

The logr package needed a manual update because of container-runtime. Once 0.7.0 is out, we don't need to manually pin the logr library version.

Some other libraries have also been updated because they are also transitive dependencies.

[k8s-ci-robot]

Welcome @sagikazarmark!

It looks like this is your first PR to kubernetes-sigs/kubefed 🎉. Please refer to our pull request process documentation to help your PR have a smooth ride to approval.

You will be prompted by a bot to use commands during the review process. Do not be afraid to follow the prompts! It is okay to experiment. Here is the bot commands documentation.

You can also check if kubernetes-sigs/kubefed has its own contribution guidelines.

You may want to refer to our testing guide if you run into trouble with your tests not passing.

If you are having difficulty getting your pull request seen, please follow the recommended escalation practices. Also, for tips and tricks in the contribution process you may want to read the Kubernetes contributor cheat sheet. We want to make sure your contribution gets all the attention it needs!

Thank you, and welcome to Kubernetes. 😃

[sagikazarmark]

Hm, can't see any obvious reason why the tests are breaking...

[jimmidyson]

Thank you for this. Please fix the kind image and, if you don't mind, update to v0.19.3 dependencies, as well as running go mod tidy please to try to keep go.sum in a reasonable state.

[sagikazarmark]

@jimmidyson thanks for the review. I addressed/replied to your comments.

I actually ran go mod tidy, can you see anything unusual?

[jimmidyson]

@sagikazarmark I was surprised there weren't a lot more removals from go.sum. It's not a blocker though.

[jimmidyson]

/approve

[sagikazarmark]

I'll take a look at it. My guess is there is a transitive dependency on an earlier Kubernetes version.

[sagikazarmark]

Controller runtime for instance still relies on 1.18.6

[jimmidyson]

@sagikazarmark Ah ok that makes sense

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: jimmidyson, makkes, sagikazarmark

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:

• OWNERS [jimmidyson]

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

[jimmidyson]

@sagikazarmark Sorry for the bother but would you mind rebasing this against master and dropping the last two commits in this branch (only need the updated kubernetes dependencies)? Hopefully that should sort out the CI issues.

[sagikazarmark]

@jimmidyson I've just rebased it, looks like at the same time you did. 😄

[jimmidyson]

Oh I'm sorry @sagikazarmark, I was just trying to help out to get it in, hope that's ok. I'm going to push another commit just to bump the dependencies to latest now they're published, then we can get this merged. Thank you for kicking this upgrade off.

[sagikazarmark]

Sure, no worries. 😄 Let me know if I can help. Thanks!

[sagikazarmark]

Looks like it fails with the same error 😕

[jimmidyson]

@sagikazarmark My dodgy rebase - I knew I should have waited for you 😉

[sagikazarmark]

Hm, I'm not sure, the rebase looks fine to me. Only go.mod and go.sum have changes.

[jimmidyson]

Yeah it seems to be a consistent test failure.

[jimmidyson]

To test the one failing test you can run (after spinning up some clusters and deploying kubefed - scripts/create-clusters.sh and make deploy.kind should help I think):

$ make bin/e2e
$ bin/e2e -kubeconfig=${HOME}/.kube/config -ginkgo.v -single-call-timeout=1m -ginkgo.trace -ginkgo.randomizeAllSpecs -ginkgo.focus="deployments.apps.* should be created, read, updated and deleted successfully"

I'm also looking at this btw so hope we can get this fixed quickly and merged.

[sagikazarmark]

I can see a bunch of ERROR: Cluster object is not as expected errors caused by an extra "deployment.kubernetes.io/revision": "1" annotation in the object.

[jimmidyson]

Yeah it looks like a bug in the jsonpatch upgrade (test below).

package a

import (
	"encoding/json"
	"fmt"
	"testing"

	jsonpatch "github.com/evanphx/json-patch"
	"k8s.io/apimachinery/pkg/apis/meta/v1/unstructured"

	"sigs.k8s.io/kubefed/pkg/controller/util"
)

func TestPatch(t *testing.T) {
	orig := &unstructured.Unstructured{
		Object: map[string]interface{}{
			"metadata": map[string]interface{}{
				"annotations": map[string]interface{}{
					"deployment.kubernetes.io/revision": "1",
				},
			},
		},
	}
	origJSON, _ := orig.MarshalJSON()

	overrides := util.ClusterOverrides{{
		Op:   "add",
		Path: "/metadata/annotations",
		Value: map[string]string{
			"foo": "bar",
		},
	}, {
		Path: "/metadata/annotations/foo",
		Op:   "remove",
	}}
	jsonPatchBytes, _ := json.Marshal(overrides)
	patch, _ := jsonpatch.DecodePatch(jsonPatchBytes)

	patchedObjectJSONBytes, _ := patch.Apply(origJSON)
	fmt.Println(string(patchedObjectJSONBytes))
}

[jimmidyson]

Or potentially a fix in the JSONPatch library... I think this is it tbh.

[jimmidyson]

Ah yes this is because the override is wrong... fixing now.

[jimmidyson]

OK fixed. The override was actually invalid...

[jimmidyson]

/lgtm

[jimmidyson]

Thanks @sagikazarmark!

[sagikazarmark]

Thanks for the help @jimmidyson