prometheus监控k8s
使⽤prometheus监控node节点
使⽤cadvisor监控容器，最后使⽤grafana图形化显⽰数据
1、安装prometheus，需要将prometheus的配置⽂件映射进去，使⽤configMap⽅式映射
创建configMap，名字为pro
kubectl create configmap pro --from-file=./l
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  # - 'l'
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、安装prometheus
vim l
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 l
安装prometheus的node节点
vim l
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    #仅主机
15      containers:
16      - name: node-exporter  #容器名字
17        image: prom/node-exporter    #使⽤客户端的镜像名称
18        ports:
19        - containerPort: 9100
20        imagePullPolicy: IfNotPresent    #镜像策略。如果没有这个镜像就下载
21        command:    #运⾏的命令：
22        - /bin/node_exporter
23        - --path.procfs
24        - /host/proc      #对应的⽬录
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/ru
n/docker/netns|rootfs/var/lib/docker/devicemapper|rootfs/var/lib/docker/aufs)($$|/)
29        volumeMounts:            #挂载对应关系，挂载到⾥⾯的⽬录
30        - name: proc
31          mountPath: /host/proc
32        - name: sys
33          mountPath: /host/sys
34        - name: root
35          mountPath: /rootfs
36      volumes:    #外⾯的⽬录，和containers是对齐的
37        - name: proc
38          hostPath:  #主机上的⽬录
39            path: /proc
40        - name: sys
41          hostPath:
42            path: /sys
43        - name: root
44          hostPath:
45            path: /
kubectl apply -f l
3、安装cadvisor
l
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  # 不管什么问题总是重启
16      containers:
17      - name: cadvisor
18        image: google/cadvisor
19        imagePullPolicy: IfNotPresent  # 镜像策略，如果本地没有则下载
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 -l
4、安装grafana（web界⾯）
l
nodeselector
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: "grafana.server.name"
24        - name: GF_SECURITY_ADMIN_PASSWORD
25          value: "123"
kubectl apply  -l
查看pod运⾏状态
[root@master prometheus]# kubectl get pod
可以看到现在是正确的，主机36显⽰没有成功，是因为master默认有污点的，容器不会调度到master节点上，所以连接36报错。

版权声明：本站内容均来自互联网，仅供演示用，请勿用于商业和其他非法用途。如果侵犯了您的权益请与我们联系QQ:729038198，我们将在24小时内删除。