12-24-2025, 07:54 AM
I'm tasked with setting up a new on-premises Kubernetes cluster for our development and staging environments, and I'm evaluating whether to use a managed distribution like RKE2 or k3s versus a more manual kubeadm installation. Our team is relatively new to Kubernetes, so operational simplicity and good documentation are key, but we also need a setup that mirrors our future production architecture. For those who have stood up self-managed clusters, what were the most time-consuming or error-prone parts of the initial Kubernetes cluster setup, particularly around networking with Calico or Cilium and persistent storage integration? How did you handle ongoing maintenance tasks like certificate rotation and version upgrades without causing significant downtime?