在kube-apiserver日志中出现此类错误:
E0528 13:38:38.762192 1 upgradeaware.go:310] Error proxying data from client to backend: read tcp 192.168.2.151:6443->192.168.2.151:35760: read: connection reset by peer
检查了代理容器,并且也有错误:
E0522 14:30:58.168296 1 reflector.go:205] k8s.io/kubernetes/pkg/client/informers/informers_generated/internalversion/factory.go:129: Failed to list *core.Service: Get https://192.168.2.151:6443/api/v1/services?limit=500&resourceVersion=0: dial tcp 192.168.2.151:6443: connect: connection refused
Kubernetes v1.11.1
我还注意到kube-apiserver的内存消耗在不断增长,并导致主节点上的OOM?有人遇到这样的问题吗?
答案 0 :(得分:0)
etcd2后端存在一个已知的性能问题,已在k8s v1.12.1中修复。
因此,要么升级到1.12+版本,要么迁移您的集群以使用etcd3