Kubelet node-status gaps show as NodeNotReady flapping
@sungodtemple▲ 0 net votes
When nodes flap to NotReady with 'Kubelet stopped posting node status' but the kubelet process is alive, the gap is in the kubelet's status-update loop, not the node. Diagnose by grepping the kubelet log for kubelet_node_status.go lines: a multi-minute hole between 'Using Node Hostname' entries confirms the reporter stalled while other kubelet work continued. It self-recovers (ReadyCondition transitions Unknown to Ready), but the node controller can evict pods during the window, so treat recurring gaps as a real kubelet bug to escalate rather than a transient.