Jesse Hu

Thanks. But this won't work if delete CAPI provider (at the end) fails in the middle.

This sounds similiar to [Validating Helm Chart Values with JSON Schemas](https://www.arthurkoziel.com/validate-helm-chart-values-with-json-schemas/)

I hit the same issue when using kapp-controller and PackageInstall to deploy calico tigera-operator helm chart which creates a CR and deploys a controller reconciling this CR. kapp-controller and tigera-operator...

In this case the KubeadmControlPlane API Server is **not able** to receive requests, so setting KCP.status.ready to false is reasonable ? https://github.com/kubernetes-sigs/cluster-api/blob/355348c26e6128c37ec35e3cd2ff1b24892ad96c/controlplane/kubeadm/api/v1beta1/kubeadm_control_plane_types.go#L197C1-L201C27 ``` // Ready denotes that the KubeadmControlPlane API...

I see. By searching for kcp.Status.Ready (find by reference gives same result), kcp.Status.Ready is ready only in 1 place. kcp.Status.Ready is set to true when kcp.Status.ReadyReplicas > 0. So if...

Thanks for explaining the workflow! Is there an API contract description to mention KCP.status.ready only means KCP was ready once but do not guarantee it's currently ready? CAPI 1.4.x code...

Thank you both very much 👍 According to the above code `cluster.Status.ControlPlaneReady = ready`, cluster.Status.ControlPlaneReady will never be set to false even if KCP becomes unavailable. I agree we can...

/remove-lifecycle rotten

We have verified this harbor arm64 build patch by @morlay works https://github.com/goharbor/harbor/compare/v2.8.3...morlay:harbor:patch-v2.8.3.

I'm pulling an image. It's not cached because kapp-controller reports not able to pull the image when the image registry server goes down.