我有一个Kubernetes 1.5.3集群，它使用kubeadm（详情如下）在两个节点上启动。我一直在试图启动今晚的prometheus-operator。在我删除了一种类型=普罗米修斯的资源之后，我的集群进入了一个非常奇怪的状态，当我试图重新创建普罗米修斯资源时，虽然TPR存在，但没有创建任何Pod。调查使我：kube-apiserver无法正常运行，“名称命名空间不存在索引”
[email protected]:~$ kubectl logs -n kube-system kube-apiserver-hypervisor01 | tail -10 W0226 03:57:34.641244 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:57:34.645073 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:58:04.642150 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:58:04.647953 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:58:34.642118 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:58:34.646427 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:59:04.642189 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:59:04.647978 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:59:34.646129 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist W0226 03:59:34.666355 1 listers.go:69] can not retrieve list of objects using index : Index with name namespace does not exist
解决办法在此rancher github issue是简单地重新启动API服务器，这将是很好，除了这并没有解决问题。否则，它会指向etcd上的高负载，正如您在下面看到的那样，该负载是托管为一个容器，位于当前负载小于1的主机上。etcdctl命令运行成功，导致我相信etcd本身很好。为了好的措施，我碰到了etcd docker容器，它仍然正常工作。但是，从这里开始，我迷路了，需要一些帮助让我的集群恢复正常工作状态。
群集概述：
[email protected]:~$ kubectl get no NAME STATUS AGE hypervisor01 Ready,master 2d kubes01.pss9.kickass.systems Ready 2d [email protected]:~$ kubectl get -n kube-system po NAME READY STATUS RESTARTS AGE dummy-2088944543-2p3zf 1/1 Running 1 2d etcd-hypervisor01 1/1 Running 2 2d kube-apiserver-hypervisor01 1/1 Running 3 2d kube-controller-manager-hypervisor01 1/1 Running 1 2d kube-discovery-1769846148-v8h50 1/1 Running 1 2d kube-dns-2924299975-3s26d 4/4 Running 4 2d kube-proxy-vpw73 1/1 Running 1 2d kube-proxy-zfh13 1/1 Running 0 2d kube-registry-proxy-6hhk9 1/1 Running 1 2d kube-registry-proxy-nl1s1 1/1 Running 0 2d kube-registry-v0-4d94t 1/1 Running 0 2d kube-scheduler-hypervisor01 1/1 Running 1 2d [email protected]:~$ kubectl get po NAME READY STATUS RESTARTS AGE kube-flannel-ds-589fw 2/2 Running 3 2d kube-flannel-ds-7f5sx 2/2 Running 0 2d kube-state-metrics-3229993571-20nbc 1/1 Running 0 1h node-exporter-1cdlj 1/1 Running 0 1h node-exporter-jc54s 1/1 Running 0 1h prometheus-operator-996254120-0wzg6 1/1 Running 0 2h
https://github.com/kubernetes/kubernetes/blob/v1.5.3/pkg/client/cache/listers.go#L68该代码似乎暗示事情应该继续在这里成功，所以也许这个日志是一个红鲱鱼... –