improve safety for end-user during control plane deletion

[boluisa]

User Story

As an operator, I would like to delete control plane nodes in a way that does not leave behind etcd members or in a way that does not require running Kubeadm reset prior to machine deletion.

Detailed Description

A delete of a control plane object should remove it corresponding etcd member from the etcd cluster.

/kind feature

[vincepri]

The current code in v1alpha3 takes care of this when scaling down https://github.com/kubernetes-sigs/cluster-api/blob/master/controlplane/kubeadm/controllers/kubeadm_control_plane_controller.go#L541-L550, is this what you were asking?

[vincepri]

/priority awaiting-more-evidence

[boluisa]

@vincepri no, not via scaling down. I meant by deleting the control plane object itself. for example: kubectl delete machine controlplane-1

[detiber]

One way to achieve this:

• Add a KCP owned finalizer to the Machine when creating or adopting
• Update etcd/kubeadm config entries when deletion of Machine is detected before removing finalizer

[vincepri]

/milestone Next

gonna punt to next for now given the pending testing and robustness in KCP #2753

[vincepri]

Somewhat related: #2818

[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

[vincepri]

Closing this given that we reconcile the etcd membership with what Machines are available (and without a deletion timestamp).

/close

[k8s-ci-robot]

@vincepri: Closing this issue.

In response to this:

Closing this given that we reconcile the etcd membership with what Machines are available (and without a deletion timestamp).

/close

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.