kube-prometheus安装部署(一)

2021-07-31 16:08 约 351 字预计阅读 2 分钟 - 次阅读

警告

本文最后更新于 2021-11-05 23:30，文中内容可能已过时。

一、下载编排文件

1
2
git clone https://github.com.cnpmjs.org/prometheus-operator/kube-prometheus.git
cd kube-prometheus/manifests/

二、部署operator crd

1
k apply -f setup/

三、部署prometheus

1
k apply -f .

如果需要数据持久化,可以配置pvc,前提是集群有storage-class,修改prometheus-prometheus.yaml,在最后追加如下内容
1
2
3
4
5
6
7
 storage:
   volumeClaimTemplate:
     spec:
       storageClassName: nfs-storage
       resources:
         requests:
           storage: 50Gi

四、解决报错

kube-state-metrics镜像拉取错误，k describe pod -n monitoring kube-state-metrics-76f6cb7996-2nqpf查看详细信息

需要访问google，em。。。我从本地拉了一个传到阿里云镜像仓库了,镜像地址: registry.cn-shanghai.aliyuncs.com/soulchild/kube-state-metrics:v2.0.0

修改镜像: k set image -n monitoring deployment kube-state-metrics kube-state-metrics=registry.cn-shanghai.aliyuncs.com/soulchild/kube-state-metrics:v2.0.0

部署完成

五、配置ingress提供外部访问

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
apiVersion: networking.k8s.io/v1
kind: Ingress
metadata:
  name: prometheus
  namespace: monitoring
spec:
  rules:
  - host: prom.ops.cn
    http:
      paths:
      - pathType: Prefix
        path: "/"
        backend:
          service:
            name: prometheus-k8s
            port:
              name: web
  - host: alert.ops.cn
    http:
      paths:
      - pathType: Prefix
        path: "/"
        backend:
          service:
            name: alertmanager-main
            port:
              name: web
  - host: grafana.ops.cn
    http:
      paths:
      - pathType: Prefix
        path: "/"
        backend:
          service:
            name: grafana
            port:
              name: http

六、解决controllerManager和scheduler不能自动发现

kube-prometheus默认会在kube-system命名空间下查找controllerManager和scheduler的service对应的endpoint作为target地址,由于我本地是二进制部署的集群,并没有这两个service和ep,所以需要手动创建一下

controller-manager:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
apiVersion: v1
kind: Service
metadata:
  name: kube-controller-manager
  namespace: kube-system
  labels:
    app.kubernetes.io/name: kube-controller-manager
spec:
  type: ClusterIP
  clusterIP: None
  ports:
  - name: https-metrics
    port: 10257
---
apiVersion: v1
kind: Endpoints
metadata:
  name: kube-controller-manager
  namespace: kube-system
  labels:
    app.kubernetes.io/name: kube-controller-manager
subsets:
- addresses:
  - ip: 172.17.20.201
  - ip: 172.17.20.202
  - ip: 172.17.20.203
  ports:
  - name: https-metrics
    port: 10257
    protocol: TCP

scheduler:

 1
 2
 3
 4
 5
 6
 7
 8
 9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
apiVersion: v1
kind: Service
metadata:
  name: kube-scheduler
  namespace: kube-system
  labels:
    app.kubernetes.io/name: kube-scheduler
spec:
  type: ClusterIP
  clusterIP: None
  ports:
  - name: https-metrics
    port: 10259
---
apiVersion: v1
kind: Endpoints
metadata:
  name: kube-scheduler
  namespace: kube-system
  labels:
    app.kubernetes.io/name: kube-scheduler
subsets:
- addresses:
  - ip: 172.17.20.201
  - ip: 172.17.20.202
  - ip: 172.17.20.203
  ports:
  - name: https-metrics
    port: 10259
    protocol: TCP

七、解决CoreDns不能自动发现

kube-prometheus同样通过service查找coredns的ep作为target,kube-prometheus在选择service的时候使用的是app.kubernetes.io/name: kube-dns这样的标签,但是我部署的core-dns没有这个标签，所以需要添加这个标签

k label service -n kube-system kube-dns app.kubernetes.io/name=kube-dns

可以看到刚才的几个组件都可以正常显示了

八、访问测试

grafana的默认账号密码是admin

请我喝杯水

赞赏

微信号

微信打赏