监控集群
概述
在 Kubernetes 中，可以使用 Metrics Server 来完成系统资源的采集工作，通过 Metrics 采集节点和 Pod 的内存、磁盘、CPU 和网络的使用率。通过这些指标，可以结合 Horizontal Pod Autoscaler(HPA) 来完成自动伸缩[链接]。
操作步骤
部署 metrics-server
bash
# 将 crt 证书复制到所有的节点上
$ scp /etc/kubernetes/pki/front-proxy-ca.crt root@master2.cluster.k8s:/etc/kubernetes/pki/front-proxy-ca.crt
$ scp /etc/kubernetes/pki/front-proxy-ca.crt root@node1.cluster.k8s:/etc/kubernetes/pki/front-proxy-ca.crt
# 在主节点（master1.cluster.k8s）上，通过 helm 安装 metrics-server
$ helm install kube-metrics-server mirror/kube-metrics-server -n kube-system
NAME: kube-metrics-server
LAST DEPLOYED: Thu Jul 27 04:47:23 2023
NAMESPACE: kube-system
STATUS: deployed
REVISION: 1
TEST SUITE: None
# 获取 helm 部署信息
$ helm list -n kube-system
NAME NAMESPACE REVISION UPDATED STATUS CHART APP VERSION
kube-flannel kube-system 1 2023-07-27 04:20:53.353204291 +0800 CST deployed kube-flannel-v0.22.0 v0.22.0
kube-metrics-server kube-system 1 2023-07-27 04:47:23.939085564 +0800 CST deployed kube-metrics-server-0.6.3 0.6.3
nfs-permanent kube-system 1 2023-07-27 04:39:13.846105718 +0800 CST deployed nfs-permanent-4.0.18 4.0.2
nfs-temporary kube-system 1 2023-07-27 04:39:36.230024145 +0800 CST deployed nfs-temporary-4.0.18 4.0.2
# 查看 metrics-server Pod 的状态
$ kubectl get po -n kube-system -l k8s-app=metrics-server
NAME READY STATUS RESTARTS AGE
metrics-server-fb5d74cd4-gxpfl 1/1 Running 0 31s获取节点和 Pod 的资源使用率
等待 metrics-server 的 Pod 的状态转变为 Running 后，等待几分钟，就可以查看节点和 Pod 的资源使用率了。
bash
# 获取节点的资源使用率
$ kubectl top node
NAME CPU(cores) CPU% MEMORY(bytes) MEMORY%
master1.cluster.k8s 296m 7% 953Mi 24%
master2.cluster.k8s 343m 8% 777Mi 20%
master3.cluster.k8s 258m 6% 759Mi 19%
node1.cluster.k8s 64m 1% 482Mi 12%
node2.cluster.k8s 77m 1% 434Mi 11%
node3.cluster.k8s 60m 1% 301Mi 7%
# 获取所有 Pod 的资源使用率
$ kubectl top po -A
NAMESPACE NAME CPU(cores) MEMORY(bytes)
ingress-nginx ingress-nginx-controller-5bvq5 2m 91Mi
ingress-nginx ingress-nginx-controller-jw59l 2m 92Mi
kube-system coredns-5d78c9869d-56smn 3m 20Mi
kube-system coredns-5d78c9869d-fmrv5 3m 18Mi
kube-system etcd-master1.cluster.k8s 121m 79Mi
kube-system etcd-master2.cluster.k8s 114m 72Mi
kube-system etcd-master3.cluster.k8s 102m 73Mi
kube-system kube-apiserver-master1.cluster.k8s 95m 306Mi
kube-system kube-apiserver-master2.cluster.k8s 111m 260Mi
kube-system kube-apiserver-master3.cluster.k8s 85m 286Mi
kube-system kube-controller-manager-master1.cluster.k8s 6m 27Mi
kube-system kube-controller-manager-master2.cluster.k8s 52m 59Mi
kube-system kube-controller-manager-master3.cluster.k8s 5m 25Mi
kube-system kube-flannel-ds-4xblf 14m 24Mi
kube-system kube-flannel-ds-c77gm 13m 20Mi
kube-system kube-flannel-ds-fhjtl 11m 17Mi
kube-system kube-flannel-ds-hm46x 13m 22Mi
kube-system kube-flannel-ds-ngrcr 13m 25Mi
kube-system kube-flannel-ds-v4qwj 13m 21Mi
kube-system kube-proxy-5pj6c 9m 23Mi
kube-system kube-proxy-67v7k 1m 24Mi
kube-system kube-proxy-d7ffg 1m 24Mi
kube-system kube-proxy-hcm5r 1m 23Mi
kube-system kube-proxy-jjt2t 1m 23Mi
kube-system kube-proxy-sgpmx 9m 23Mi
kube-system kube-scheduler-master1.cluster.k8s 7m 23Mi
kube-system kube-scheduler-master2.cluster.k8s 8m 26Mi
kube-system kube-scheduler-master3.cluster.k8s 6m 21Mi
kube-system metrics-server-fb5d74cd4-gxpfl 9m 25Mi
kube-system nfs-permanent-6885f4b8f9-zf5rq 4m 9Mi
kube-system nfs-temporary-76c5c9bf57-h5pn4 4m 13Mi