kube-api-endpoint relation between control-plane and worker creates SPOF in OpenStack integration

192.168.151.72 - external IP address of OpenStack LB (Octavia Amphora) as expected
10.5.5.14 - internal IP address of kubernetes-control-plane/0 as SPOF

$ juju run --unit kubernetes-worker/0 -- cat /root/.kube/config | grep server:
    server: https://192.168.151.72:443

$ juju add-relation kubernetes-worker:kube-api-endpoint kubernetes-control-plane:kube-api-endpoint
$ juju-wait -w

$ juju run --unit kubernetes-worker/0 -- cat /root/.kube/config | grep server:
    server: https://10.5.5.14:6443

$ juju remove-relation kubernetes-worker:kube-api-endpoint kubernetes-control-plane:kube-api-endpoint
$ juju-wait -w

$ juju run --unit kubernetes-worker/0 -- cat /root/.kube/config | grep server:
    server: https://192.168.151.72:443

Cross linking: https://github.com/charmed-kubernetes/kubernetes-docs/issues/687

The following comment looks like confirming the idea.

https://bugs.launchpad.net/charm-openstack-integrator/+bug/1959720/comments/13
> I can confirm that it is fixed in Charmed Kubernetes 1.22. After you
> upgrade to 1.22, make sure you remove the deprecated relation between
> kubernetes-worker:kube-api-endpoint and
> kubernetes-master:kube-api-endpoint. That will allow kubernetes-worker
> to get the API endpoint from the kube-control relation instead, which
> uses the loadbalanced API endpoint.

Subscribing ~field-high to get a confirmation that removing the relation is the right way to the issue before applying that to the production system.

Yes, since Charmed Kubernetes 1.22, we recommend removing the relation between kubernetes-worker:kube-api-endpoint and kubernetes-control-plane:kube-api-endpoint. That will allow kubernetes-worker to obtain the proper loadbalanced address from the kube-control relation instead.

We will get the documentation updated.

PR: https://github.com/charmed-kubernetes/bundle/pull/842