Make kubectl Scale function for RS, Deployments, Job, etc watch-based

[shyamjvs]

Offshoot of #56064 (comment)

As part of the kubectl Scale() function, we can optionally waitForReplicas to reach the desired count.
Currently, we're polling periodically to check that happens (for RS, jobs, etc). We should move them to use Watch instead - like we're doing for RCs currently:

kubernetes/pkg/kubectl/scale.go

Lines 233 to 246 in 98fb71e

	watcher, err := scaler.c.ReplicationControllers(namespace).Watch(watchOptions)
	if err != nil {
	return err
	}
	_, err = watch.Until(waitForReplicas.Timeout, watcher, func(event watch.Event) (bool, error) {
	if event.Type != watch.Added && event.Type != watch.Modified {
	return false, nil
	}
	return checkRC(event.Object.(*api.ReplicationController)), nil
	})
	if err == wait.ErrWaitTimeout {
	return fmt.Errorf("timed out waiting for %q to be synced", name)
	}
	return err

/area kubectl
/kind cleanup

[shyamjvs]

/sig apps
(can't think of a better one)

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[nikhita]

/remove-lifecycle stale

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[shyamjvs]

/remove-lifecycle stale

…

On Sat, May 19, 2018, 7:40 PM fejta-bot ***@***.***> wrote: Issues go stale after 90d of inactivity. Mark the issue as fresh with /remove-lifecycle stale. Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with /close. Send feedback to sig-testing, kubernetes/test-infra and/or fejta <https://github.com/fejta>. /lifecycle stale — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#56071 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEIhk_vL1ckiTBKs6NLjXA490So_ddwnks5t0FkVgaJpZM4Qkf6c> .

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[shyamjvs]

/remove-lifecycle stale /lifecycle frozen

…

On Fri, Aug 17, 2018, 7:58 PM fejta-bot ***@***.***> wrote: Issues go stale after 90d of inactivity. Mark the issue as fresh with /remove-lifecycle stale. Stale issues rot after an additional 30d of inactivity and eventually close. If this issue is safe to close now please do so with /close. Send feedback to sig-testing, kubernetes/test-infra and/or fejta <https://github.com/fejta>. /lifecycle stale — You are receiving this because you were mentioned. Reply to this email directly, view it on GitHub <#56071 (comment)>, or mute the thread <https://github.com/notifications/unsubscribe-auth/AEIhk2HZMGm0icJcPmEE8wXXpAPdLMEkks5uRwQtgaJpZM4Qkf6c> .

[tnozicka]

/assign
/remove-lifecycle frozen

[tnozicka]

@wojtek-t this was causing performance regressions if I recall correctly - if so I assume we want to target v1.12? I am gonna make most of the Until related fixes for v1.13 (to limit potential bugs) but I can take a second one for v1.12 (kubectl rollout status is the first one).

[shyamjvs]

This wasn't causing regression iirc. What happened was we were seeing flakes in our performance tests when RC scaling was watch-based (failing due to bug with Until). So we switched RC scaling to poll-based to avoid those flakes (and leaving open to the long-term solution of fixing Until, so we can go back to using it).

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[wojtek-t]

/remove-lifecycle stale

[tnozicka]

I've started removing the pooling but the subresource lack list/watch methods :(

kubernetes/staging/src/k8s.io/client-go/scale/interfaces.go

Lines 30 to 39 in dd4a8f9

	// ScaleInterface can fetch and update scales for
	// resources in a particular namespace which implement
	// the scale subresource.
	type ScaleInterface interface {
	// Get fetches the scale of the given scalable resource.
	Get(resource schema.GroupResource, name string) (*autoscalingapi.Scale, error)
	// Update updates the scale of the given scalable resource.
	Update(resource schema.GroupResource, scale *autoscalingapi.Scale) (*autoscalingapi.Scale, error)
	}

[tnozicka]

/cc @kubernetes/sig-autoscaling-pr-reviews ^ we are condemned to pooling with such interface

[tnozicka]

/unassign

[fejta-bot]

Issues go stale after 90d of inactivity.
Mark the issue as fresh with /remove-lifecycle stale.
Stale issues rot after an additional 30d of inactivity and eventually close.

If this issue is safe to close now please do so with /close.

Send feedback to sig-testing, kubernetes/test-infra and/or fejta.
/lifecycle stale

[wojtek-t]

/remove-lifecycle stale
/lifecycle frozen