kubectl drain node_name is an idempotent command which allows us to easily retry it During the upgrade, since a new node has just joined the cluster, the etcd leader might change and the control plane nodes are not stable. That can easily disrupt the drain command but we will keep retrying to drain the node Since a 1hour deadline is set on the drain node, we know we will get the page and we won't restart indefinitely.
2.3 KiB
2.3 KiB