Kublet PLEG不健康问题排障

环境：Rancher管控的K8S集群。
现象：某个Node频繁出现“PLEG is not healthy: pleg was last seen active 3m46.752815514s ago; threshold is 3m0s”错误，频率在5-10分钟就会出现一次。
排查：

kubectl get pods –all-namespaces 发现有一个istio-ingressgateway-6bbdd58f8c-nlgnd一直处于Terminating状态，也就是说杀不死。

到Node上docker logs –tail 100 kubelet也看到这个Pod的状态异常：
I0218 01:21:17.383650 10311 kubelet.go:1775] skipping pod synchronization – [PLEG is not healthy: pleg was last seen active 3m46.752815514s ago; threshold is 3m0s]
…
E0218 01:21:30.654433 10311 generic.go:271] PLEG: pod istio-ingressgateway-6bbdd58f8c-nlgnd/istio-system failed reinspection: rpc error: code = DeadlineExceeded desc = context deadline exceeded

用kubelet delete pod尝试删除，命令挂住。
用kubectl delete pod –force –grace-period=0，强制删除Pod。
再到Node上检查这个容器是否真的被停止，docker ps -a| grep ingressgateway-6bbdd58f8c-nlgnd，看到容器处于Exited状态。
观察Node状态，问题依旧。
把Pod关联的Deployment删除，把一只处于Terminating的Pod用kubectl delete pod –force –grace-period=0的方式删除。
重新部署Deployment。
问题解决。

相关issue