Kind e2e tests seem to flake during install with webhook error.

[vaikas]

Describe the bug

Last couple of days seen a few failures to install the eventing component (e2e tests are not failing, but install of eventing is):

Couple of examples are here:
https://github.com/knative/eventing/runs/1763499662?check_suite_focus=true
https://github.com/knative/eventing/runs/1763499632?check_suite_focus=true

customresourcedefinition.apiextensions.k8s.io/inmemorychannels.messaging.knative.dev created
deployment.apps/imc-controller created
Error from server (InternalError): error when creating "STDIN": Internal error occurred: failed calling webhook "config.webhook.eventing.knative.dev": Post https://eventing-webhook.knative-eventing.svc:443/config-validation?timeout=2s: dial tcp 10.96.251.146:443: connect: connection refused
deployment.apps/imc-dispatcher created
2021/01/25 17:03:45 error executing 'kubectl apply': exit status 1
Error: Process completed with exit code 1.

A clear and concise description of what the bug is.

Expected behavior
Tests to not flake.

To Reproduce
Steps to reproduce the behavior.

Knative release version
HEAD

Additional context
Add any other context about the problem here such as proposed priority

[grantr]

@pierDipi does #4811 fix this? @vaikas thought it might.

[pierDipi]

No, it doesn't, KinD E2E tests don't use chaosduck:

eventing/.github/workflows/kind-e2e.yaml

Lines 161 to 162 in d58bd38

	# Be KinD to these tests.
	kubectl scale -n${SYSTEM_NAMESPACE} deployment/chaosduck --replicas=0

[vaikas]

zomg, that's right I already asked that earlier :)

Thanks!

[vaikas]

I went through the failed actions for the last 2 weeks, and I only see install failures related to the etcdserver failures:
#4975

I'm wondering if this got fixed by:
#4741

I'm closing, we can reopen if it comes back.