02/09/2023, 5:25 AM
another question is about the practice for maintaining and upgrading the rke2 cluster. More specifically my question is about how to deal with the workload above the nodes when planing a upgrade (which needs to restart rke2/kubelet/containerd?). I cannot make it clear the side-effect and bad things for shutting down the rke2/kubelet/containerd. For example, rook is a ceph storage orchestrator on k8s, however, ceph will automatically replicate data to live disks if some disks get out of the cluster. I wonder to know if I want to upgrade the underlying rke2 cluster, what should I do with workloads like rook? should I drain a node or what to do to keep my workload healthy. I’m really fresh to k8s and rke2, great thanks for you all. Any suggested documentation for me to learn how to operate and maintain a production k8s/rke2 cluster?