Serving webhook connection failure despite readiness check passing #610
Description
Hi,
I was working with latest version by using the main branch to get around the limitations fixed in #602 (a release would be greatly appreciated as quickstart is broken for every user currently 😅 ) and likely found a bug in the current webhook readiness check implementation from #607 .
It seems that the webhook check only checks for ready Pods to exist and not if Endpoints of the webhook Service are available.
It's only a short time between the pod being ready and the service being available by having endpoints but it was enough to fail for me occasionally with the following logs:
❯ ./kn-quickstart kind --registry -k 1.34.0
Running Knative Quickstart using Kind
✅ Checking dependencies...
Kind version is: 0.30.0
💽 Installing local registry...
...
enabling experimental podman provider
Creating cluster "knative" ...
✓ Ensuring node image (kindest/node:v1.34.0) 🖼
✓ Preparing nodes 📦
✓ Writing configuration 📜
✓ Starting control-plane 🕹️
✓ Installing CNI 🔌
✓ Installing StorageClass 💾
✓ Waiting ≤ 2m0s for control-plane = Ready ⏳
• Ready after 16s 💚
Set kubectl context to "kind-knative"
You can now use your cluster with:
kubectl cluster-info --context kind-knative
Have a question, bug, or feature request? Let us know! https://kind.sigs.k8s.io/#community 🙂
🔗 Patching node: knative-control-plane
🍿 Installing Knative Serving v1.19.6 ...
CRDs installed...
Core installed...
Waiting for webhook to be ready...
Webhook is ready...
Error from server (InternalError): Internal error occurred: failed calling webhook "config.webhook.serving.knative.dev": failed to call webhook: Post "https://webhook.knative-serving.svc:443/config-validation?timeout=10s": dial tcp 10.96.107.170:443: connect: connection refused
Error: failed to install serving to kind cluster knative: tag resolving configuration: exit status 1I'll create a PR to check for endpoints but as it is a bit hard to reproduce the problem might not be 100% fixed.
Thanks, Vincent
/kind bug