0x00

记录一次安装pve-exporter并且配置prometheus-operator的serviceMonitor。

prometheus-pve-exporter的作用是收集proxmox VE宿主机的状态并且提供一个接口供prometheus监控系统使用。

0x01 exporter安装

安装过程（python安装）

进入proxmox VE后台，进入节点的web shell
安装pip3
```
apt install python3-pip
```
安装prometheus-pve-exporter
```
pip3 install prometheus-pve-exporter
```

编写配置文件

mkdir /etc/prometheus-pve-exporter
nano /etc/prometheus-pve-exporter/pve.yml

写入以下内容

default:
user: your-username
password: your-password

编写启动脚本

nano /etc/init.d/prometheus-pve-exporter

写入以下内容

#!/bin/sh
pve_exporter /etc/prometheus-pve-exporter/pve.yml 9221 0.0.0.0
exit 0

测试并且重启
```
sh /etc/init.d/prometheus-pve-exporter &
curl http://localhost:9221/
```
如果获取到返回值说明启动成功。

0x02 prometheus配置
prometheus需要发现这个exporter。要让prometheus自动发现服务，需要添加一个ServiceMonitor，prometheus可以自动获取到ServiceMonitor，相当于添加了一个中转站。
创建文件pve-monitor.yaml

apiVersion: v1
kind: Endpoints
metadata:
  name: pve-monitoring
  namespace: monitoring
subsets:
  - addresses:
      - ip: xxx.xxx.xxx.xxx # 一台运行pve_exporter的服务器IP
    ports:
      - name: metrics
        port: 9221
        protocol: TCP
---
apiVersion: v1
kind: Service
metadata:
  labels:
    metrics: pve
  name: pve-monitoring
  namespace: monitoring
spec:
  ports:
    - port: 9221
      name: metrics
      protocol: TCP
  type: ClusterIP
---
apiVersion: monitoring.coreos.com/v1
kind: ServiceMonitor
metadata:
  name: pve-servicemonitor
  labels:
    team: data-provider # 用于让prometheus发现这个服务，需要手动设置
  namespace: monitoring
spec:
  endpoints:
    - interval: 30s
      port: metrics
      path: "/pve"
      params:
        - xxx.xxx.xxx.xxx # 从哪些机器获取性能指标
        - xxx.xxx.xxx.xxx 
  namespaceSelector:
    matchNames:
      - monitoring
  selector:
    matchLabels:
      metrics: pve

应用

kaf pve-monitor.yaml

0x00

0x01 exporter安装

0x02 prometheus配置