- kube-state-metrics：kube-state-metrics 通过监听 API Server 生成有关资源对象的状态指标，比如 Deployment、Node、Pod，需要注意的是 kube-state-metrics 只是简单提供一个 metrics 数据，并不会存储这些指标数据，所以我们可以使用 Prometheus 来抓取这些数据然后存储。
kube-state-metrics 和 metrics-server 之间还是有很大不同的，二者的主要区别如下：
- kube-state-metrics 主要关注的是业务相关的一些元数据，比如 Deployment、Pod、副本状态等
- metrics-server 主要关注的是资源度量 API 的实现，比如 CPU、文件描述符、内存、请求延时等指标。
部署kube-state-metrics
cluster-role.yaml
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:labels:app.kubernetes.io/component: exporterapp.kubernetes.io/name: kube-state-metricsapp.kubernetes.io/version: 2.3.0name: kube-state-metrics
rules:
- apiGroups:- ""resources:- configmaps- secrets- nodes- pods- services- resourcequotas- replicationcontrollers- limitranges- persistentvolumeclaims- persistentvolumes- namespaces- endpointsverbs:- list- watch
- apiGroups:- appsresources:- statefulsets- daemonsets- deployments- replicasetsverbs:- list- watch
- apiGroups:- batchresources:- cronjobs- jobsverbs:- list- watch
- apiGroups:- autoscalingresources:- horizontalpodautoscalersverbs:- list- watch
- apiGroups:- authentication.k8s.ioresources:- tokenreviewsverbs:- create
- apiGroups:- authorization.k8s.ioresources:- subjectaccessreviewsverbs:- create
- apiGroups:- policyresources:- poddisruptionbudgetsverbs:- list- watch
- apiGroups:- certificates.k8s.ioresources:- certificatesigningrequestsverbs:- list- watch
- apiGroups:- storage.k8s.ioresources:- storageclasses- volumeattachmentsverbs:- list- watch
- apiGroups:- admissionregistration.k8s.ioresources:- mutatingwebhookconfigurations- validatingwebhookconfigurationsverbs:- list- watch
- apiGroups:- networking.k8s.ioresources:- networkpolicies- ingressesverbs:- list- watch
- apiGroups:- coordination.k8s.ioresources:- leasesverbs:- list- watch
cluster-role-binding.yaml
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:labels:app.kubernetes.io/component: exporterapp.kubernetes.io/name: kube-state-metricsapp.kubernetes.io/version: 2.3.0name: kube-state-metrics
roleRef:apiGroup: rbac.authorization.k8s.iokind: ClusterRolename: kube-state-metrics
subjects:
- kind: ServiceAccountname: kube-state-metricsnamespace: kube-system
service-account.yaml
apiVersion: v1
automountServiceAccountToken: false
kind: ServiceAccount
metadata:labels:app.kubernetes.io/component: exporterapp.kubernetes.io/name: kube-state-metricsapp.kubernetes.io/version: 2.3.0name: kube-state-metricsnamespace: kube-system
service.yaml
apiVersion: v1
kind: Service
metadata:labels:app.kubernetes.io/component: exporterapp.kubernetes.io/name: kube-state-metricsapp.kubernetes.io/version: 2.3.0name: kube-state-metricsnamespace: kube-system
spec:clusterIP: Noneports:- name: http-metricsport: 8080targetPort: http-metrics- name: telemetryport: 8081targetPort: telemetryselector:app.kubernetes.io/name: kube-state-metrics
deployment.yaml
apiVersion: apps/v1
kind: Deployment
metadata:labels:app.kubernetes.io/component: exporterapp.kubernetes.io/name: kube-state-metricsapp.kubernetes.io/version: 2.3.0name: kube-state-metricsnamespace: kube-system
spec:replicas: 1selector:matchLabels:app.kubernetes.io/name: kube-state-metricstemplate:metadata:labels:app.kubernetes.io/component: exporterapp.kubernetes.io/name: kube-state-metricsapp.kubernetes.io/version: 2.3.0spec:automountServiceAccountToken: truecontainers:- image: bitnami/kube-state-metrics:latestlivenessProbe:httpGet:path: /healthzport: 8080initialDelaySeconds: 5timeoutSeconds: 5name: kube-state-metricsports:- containerPort: 8080name: http-metrics- containerPort: 8081name: telemetryreadinessProbe:httpGet:path: /port: 8081initialDelaySeconds: 5timeoutSeconds: 5securityContext:allowPrivilegeEscalation: falsereadOnlyRootFilesystem: truerunAsUser: 65534nodeSelector:kubernetes.io/os: linuxserviceAccountName: kube-state-metrics
Kube-state-metrics:通过监听API Server生成有关资源对象的状态指标，比如Deployment、Node、Pod，需要注意的是kube-state-metrics只是简单的提供一个metrics数据，并不会存储这些指标数据，所以我们可以使用Prometheus来抓取这些数据然后存储，主要关注的是业务相关的一些元数据，比如Deployment、Pod、副本状态等；调度了多少个replicas？现在可用的有几个？；多少个Pod是running/stopped/terminated状态？；Pod重启了多少次？；我有多少job在运行中。
Kube-state-metrics和metrics-server对比
- metric-server是从api-server中获取cpu、内存使用率这种监控指标，并把他们发送给存储后端，如influxdb或云厂商，他当前的核心作用是：为HPA等组件提供决策指标支持。
- kube-state-metrics关注于获取k8s各种资源的最新状态，如deployment或者daemonset，之所以没有把kube-state-metrics纳入到metric-server的能力中，是因为他们的关注点本质上是不一样的。metric-server仅仅是获取、格式化现有数据，写入特定的存储，实质上是一个监控系统。而kube-state-metrics是将k8s的运行状况在内存中做了个快照，并且获取新的指标，但他没有能力导出这些指标
- 换个角度讲，kube-state-metrics本身是metric-server的一种数据来源，虽然现在没有这么做。
- 另外，像Prometheus这种监控系统，并不会去用metric-server中的数据，他都是自己做指标收集、集成的（Prometheus包含了metric-server的能力），但Prometheus可以监控metric-server本身组件的监控状态并适时报警，这里的监控就可以通过kube-state-metrics来实现，如metric-serverpod的运行状态。
)