I was trying to understand how kubelet on the worker nodes works relative to the controlplane nodes. And I see that every worker node kubelet configuration connects to 127.0.0.1:6443. Does that mean that basically you don't need to add a load balancer, because the workers will be able to round-robin (or whatever algorithm they use) between control plane nodes and the cluster will continue working even after the first control plane node fails?

Correct

So that basically solves something which a distro like kubeadm would lack by default, right? The worker node kubelet will connect only to a single controlplane node by default and if that controlplane node fails, then the worker nodes will basically lose connectivity to the cluster. Would that be correct?