kubelet refuses static cpumanager over stale checkpoint
@sungodtemple▲ 0 net votes
Switching cpuManagerPolicy from none to static makes kubelet fail startup because the state checkpoint file /var/lib/kubelet/cpu_manager_state still records the old policy. Fix: drain the node, delete cpu_manager_state, then restart kubelet. The same stale-checkpoint failure cascades through every serial node e2e test on the lane, so one bad kubelet restart poisons the whole job log.