nodeStatusUpdate

--node-status-update-frequency为kubelet上传自身状态

--node-monitor-period为kube-controller多久查看一次node状态

--node-monitor-grace-period为kube-controller认为node多久无响应后是NotReady，一般是--node-status-update-frequency的N倍

--node-startup-grace-period node在多久无响应后标记为unhealthy

--pod-eviction-timeout为kube-controller认为node无响应后多久驱逐该node上的pod

• Fast update and Fast Reaction

这个不一定生效，详情见 https://www.cnblogs.com/gaorong/p/12312590.html

参数	值	默认值
--node-status-update-frequency	4s	10s
--node-monitor-period	2s	5s
--node-monitor-grace-period	20s	40s
--pod-eviction-timeout	30s	5m

• Medium Update and Average Reaction

参数	值	默认值
--node-status-update-frequency	20s	10s
--node-monitor-period	5s	5s
--node-monitor-grace-period	2m	40s
--pod-eviction-timeout	1m	5m

• Low Update and Show reaction

参数	值	默认值
--node-status-update-frequency	1m	10s
--node-monitor-period	5s	5s
--node-monitor-grace-period	5m	40s
--pod-eviction-timeout	1m	5m

参考 https://github.com/kubernetes-sigs/kubespray/blob/master/docs/kubernetes-reliability.md

pod 状态更新: https://github.com/kubernetes/kubernetes/issues/55713#issuecomment-922168158