prometheus監控k8s

使用prometheus監控node節點

使用cadvisor監控容器，最后使用grafana圖形化顯示數據

1、安裝prometheus，需要將prometheus的配置文件映射進去，使用configMap方式映射

創建configMap，名字為pro

 kubectl create configmap pro --from-file=./prometheus.yml 

prometheus.yml文件內容，修改20行內容。'192.168.55:9090','192.168.3.55:8080' 端口9090監控的是node，8080監控的容器

 1 global:
 2   scrape_interval:     15s # By default, scrape targets every 15 seconds. 3 evaluation_interval: 15s # Evaluate rules every 15 seconds. 4 5 # Attach these labels to any time series or alerts when communicating with 6  # external systems (federation, remote storage, Alertmanager). 7  external_labels: 8 monitor: 'codelab-monitor' 9 10 rule_files: 11 # - 'prometheus.rules.yml' 12 13 scrape_configs: 14 - job_name: 'prometheus' 15 16 # Override the global default and scrape targets from this job every 5 seconds. 17  # scrape_interval: 5s 18 19  static_configs: 20 - targets: ['192.168.3.55:9100','192.168.3.55:8080']

2、安裝prometheus

vim prometheus_run.yml

 1 apiVersion: apps/v1
 2 kind: DaemonSet
 3 metadata: 4  name: prometheus 5 spec: 6  selector: 7  matchLabels: 8  app: prometheus 9  template: 10  metadata: 11  labels: 12  app: prometheus 13  spec: 14 hostNetwork: true 15  containers: 16 - name: prom 17 image: prom/prometheus 18  imagePullPolicy: IfNotPresent 19  ports: 20 - containerPort: 9090 21  volumeMounts: 22 - name: pro1 23 mountPath: /etc/prometheus/ 24 readOnly: true 25  volumes: 26 - name: pro1 27  configMap: 28 name: pro

kubectl apply -f prometheus_run.yml

安裝prometheus的node節點

vim node_exporter.yml

 1 apiVersion: apps/v1
 2 kind: DaemonSet
 3 metadata: 4 name: node-exporter-daemonset 5 spec: 6  selector: 7  matchLabels: 8  name: prometheus 9  template: 10  metadata: 11  labels: 12  name: prometheus 13  spec: 14 hostNetwork: true #僅主機 15  containers: 16 - name: node-exporter #容器名字 17 image: prom/node-exporter #使用客戶端的鏡像名稱 18  ports: 19 - containerPort: 9100 20  imagePullPolicy: IfNotPresent #鏡像策略。如果沒有這個鏡像就下載 21  command: #運行的命令： 22 - /bin/node_exporter 23 - --path.procfs 24 - /host/proc #對應的目錄 25 - --path.sysfs 26 - /host/sys 27 - --collector.filesystem.ignored-mount-points 28 - ^/(sys|proc|dev|host|etc|rootfs/var/lib/docker/containers|rootfs/var/lib/docker/overlay2|rootfs/run/docker/netns|rootfs/var/lib/docker/devicemapper|rootfs/var/lib/docker/aufs)($$|/) 29  volumeMounts: #掛載對應關系，掛載到里面的目錄 30 - name: proc 31 mountPath: /host/proc 32 - name: sys 33 mountPath: /host/sys 34 - name: root 35 mountPath: /rootfs 36  volumes: #外面的目錄，和containers是對齊的 37 - name: proc 38  hostPath: #主機上的目錄 39 path: /proc 40 - name: sys 41  hostPath: 42 path: /sys 43 - name: root 44  hostPath: 45 path: / 

kubectl apply -f node_exporter.yml

3、安裝cadvisor

vim cadvisor.yml

 1 apiVersion: apps/v1
 2 kind: DaemonSet
 3 metadata: 4  name: cadvisor 5 spec: 6  selector: 7  matchLabels: 8  jk: cAdvisor 9  template: 10  metadata: 11  labels: 12  jk: cAdvisor 13  spec: 14 hostNetwork: true 15  restartPolicy: Always # 不管什么問題總是重啟 16  containers: 17 - name: cadvisor 18 image: google/cadvisor 19  imagePullPolicy: IfNotPresent # 鏡像策略，如果本地沒有則下載 20  ports: 21 - containerPort: 8080 22  volumeMounts: 23 - name: root 24 mountPath: /rootfs 25 - name: run 26 mountPath: /var/run 27 - name: sys 28 mountPath: /sys 29 - name: docker 30 mountPath: /var/lib/docker 31  volumes: 32 - name: root 33  hostPath: 34 path: / 35 - name: run 36  hostPath: 37 path: /var/run 38 - name: sys 39  hostPath: 40 path: /sys 41 - name: docker 42  hostPath: 43 path: /var/lib/docker

kubectl apply -f cadvisor.yml

4、安裝grafana（web界面）

vim grafana.yml

 1 apiVersion: apps/v1
 2 kind: DaemonSet
 3 metadata: 4  name: grafana 5 spec: 6  selector: 7  matchLabels: 8  name: grafana 9  template: 10  metadata: 11  labels: 12  name: grafana 13  spec: 14 hostNetwork: true 15  containers: 16 - name: grafana 17 image: grafana/grafana 18  imagePullPolicy: IfNotPresent 19  ports: 20 - containerPort: 3000 21  env: 22 - name: GF_SERVER_ROOT_URL 23 value: "http://grafana.server.name" 24 - name: GF_SECURITY_ADMIN_PASSWORD 25 value: "123.com"

kubectl apply  -f grafana.yml

查看pod運行狀態

[root@master prometheus]# kubectl get pod 

 

 

 我們采用的是容器端口直接暴露在主機上（hostNetwork: true），使用主機IP+端口就可以訪問到（http://192.168.3.55:9090/）

 

 

 可以看到現在是正確的，主機36顯示沒有成功，是因為master默認有污點的，容器不會調度到master節點上，所以連接36報錯。

http://192.168.3.55:8080/ 訪問cadvisor監控容器數據是否正常

 

 

 訪問grafana（http://192.168.3.55:3000/）默認賬號為admin密碼我們在配置文件寫的為123.com