kube-prometheus 部署指南

kube-prometheus 部署指南

1、介绍：

略

2、版本：

Kubernetes 版本：1.16.2

Docker 版本：18.09.9

Kube-prometheus 版本：0.2.0

3、下载 kube-prometheus 镜像部署文件

前往 github 上下载 0.2.0 版本的压缩包

4、进入源码的 manifests 文件夹

cd kube-prometheus/manifests/

5、修改源码文件，将 service 以 NodePort 的方式提供服务

vim prometheus-service.yaml

apiVersion: v1

kind: Service

metadata:

  labels:

    prometheus: k8s

  name: prometheus-k8s

  namespace: monitoring

spec:

  type: NodePort

  ports:

  - name: http

    port: 9090

    targetPort: web

    nodePort: 9090

  selector:

    app: prometheus

    prometheus: k8s

  sessionAffinity: ClientIP

vim alertmanager-service.yaml

apiVersion: v1

kind: Service

metadata:

  labels:

    app: grafana

  name: grafana

  namespace: monitoring

spec:

  type: NodePort

  ports:

  - name: http

    port: 3000

    targetPort: http

    nodePort: 3000

  selector:

    app: grafana

vim prometheus-service.yaml

apiVersion: v1

kind: Service

metadata:

  labels:

    alertmanager: main

  name: alertmanager-main

  namespace: monitoring

spec:

  type: NodePort

  ports:

  - name: web

    port: 9093

    targetPort: web

    nodePort: 9093

  selector:

    alertmanager: main

    app: alertmanager

  sessionAffinity: ClientIP

6、修改配置文件设置 prometheus 历史数据保存 15 天，并且使用 ceph 设置数据持久化

vim 00StorageClass.yaml

kind: StorageClass

apiVersion: storage.k8s.io/v1

metadata:

  name: cephfs

provisioner: ceph.com/cephfs

parameters:

    monitors: 192.168.10.101:6789,192.168.10.102:6789,192.168.10.103:6789

    adminId: admin

    adminSecretName: ceph-admin-secret

    adminSecretNamespace: "kube-system"

    claimRoot: /volumes/kubernetes

修改prometheus部署文件

vim prometheus-prometheus.yaml

apiVersion: monitoring.coreos.com/v1

kind: Prometheus

metadata:

  labels:

    prometheus: k8s

  name: k8s

  namespace: monitoring

spec:

  alerting:

    alertmanagers:

    - name: alertmanager-main

      namespace: monitoring

      port: web

  baseImage: quay.io/prometheus/prometheus

  nodeSelector:

    kubernetes.io/os: linux

  podMonitorSelector: {}

  replicas: 2

  retention: 15d #设置历史数据保存15天

  resources:

    requests:

      memory: 400Mi

  ruleSelector:

    matchLabels:

      prometheus: k8s

      role: alert-rules

  securityContext:

    fsGroup: 0 #一定修改不然会有权限错误

    runAsNonRoot: false#一定修改不然会有权限错误

    runAsUser: 0#一定修改不然会有权限错误

  serviceAccountName: prometheus-k8s

  serviceMonitorNamespaceSelector: {}

  serviceMonitorSelector: {}

  version: v2.11.0

  storage:

    volumeClaimTemplate:

      spec:

        storageClassName: cephfs

        resources:

          requests:

            storage: 20Gi

7、设置 grafana 持久化

先创建pvc文件

vim 00grafana-pvc.yaml

kind: PersistentVolumeClaim

apiVersion: v1

metadata:

  name: grafana

namespace: monitoring  #指定monitoring

spec:

storageClassName: cephfs #指定StorageClass

  accessModes:

    - ReadWriteOnce

  resources:

    requests:

      storage: 10Gi

8、监控 kube-controller-manager 和 kube-scheduler 这两个系统组件

创建两个 service 对应上述系统组件

vim 00kube-scheduler-service.yaml

apiVersion: v1

kind: Service

metadata:

  namespace: kube-system

  name: kube-scheduler

  labels:

    k8s-app: kube-scheduler

spec:

  selector:

    component: kube-scheduler

  ports:

  - name: http-metrics

    port: 10251

    targetPort: 10251

    protocol: TCP

vim 00kube-controller-manager-service.yaml

apiVersion: v1

kind: Service

metadata:

  namespace: kube-system

  name: kube-controller-manager

  labels:

    k8s-app: kube-controller-manager

spec:

  selector:

    component: kube-controller-manager

  ports:

  - name: http-metrics

    port: 10252

    targetPort: 10252

    protocol: TCP

修改grafana部署文件

- name: grafana-storage

        persistentVolumeClaim:

          claimName: grafana

9、现在就可以 apply 文件了

kubectl apply

-

f .