【K3S】06 - Rancher部署Prometheus + Grafana

如遇图片无法加载请使用代理访问

Cluster monitoring with Prometheus and Rancher

前言

本文主要讲述 2023 K3s Rancher 部署 Prometheus + Grafana

如果你还没有部署 K3s 和 Rancher ，你可以浏览这篇文章：【K3S】01 - 异地集群初始化

环境声明

hostname	系统	配置	节点	角色	部署
m1	Ubuntu-Server(20.04)	2c4g	192.168.0.67/32	control-plane,etcd,master	k3s(v1.24.6+k3n1) server nginx rancher(2.7.1) Helm(3.10.3)
n1	Ubuntu-Server(20.04)	1c2g	192.168.0.102/32	control-plane,etcd,master	k3s(v1.24.6+k3n1) server
m2	Ubuntu-Server(20.04)	2c4g	172.25.4.244/32	control-plane,etcd,master	k3s(v1.24.6+k3n1) server
harbor	Ubuntu-Server(20.04)	2c4g	192.168.0.88	Docker-Hub Jenkins CI/CD	Harbor(2.7.1) Jenkins(2.3) Docker-Compose

节点均用 WireGuard 打通内网，后续所有节点路由均用内网ip访问

安装

首先添加Prometheus的仓库地址

1	helm repo add prometheus-community https://prometheus-community.github.io/helm-charts

拉取离线包

我们拉取了指定版本的离线包

1	helm pull prometheus-community/kube-prometheus-stack --version 45.1.0

解压离线包

1	tar xvf kube-prometheus-stack-*

修改镜像地址为国内源

cd kube-prometheus-stack/
sed -i "s#registry.k8s.io/kube-state-metrics/kube-state-metrics#registry.cn-hangzhou.aliyuncs.com/chenby/kube-state-metrics#g" charts/kube-state-metrics/values.yaml
nano value.yaml

# ctrl + w 开启搜索
# 输入 repository: ingress-nginx/kube-webhook-certgen 回车

# 将 registry: registry.k8s.io 改为
# registry: registry.cn-hangzhou.aliyuncs.com

# 将 repository: ingress-nginx/kube-webhook-certgen 改为
# repository: chenby/kube-webhook-certgen

# 最终效果如下

image:
  registry: registry.cn-hangzhou.aliyuncs.com
  repository: chenby/kube-webhook-certgen

安装

1 2	kubectl create ns monitor helm install op . -n monitor

修改 op-grafana 的类型为 NodePort

1
2
3

kubectl edit svc op-grafana -n monitor
# ClusterIP 改为 NodePort
type: NodePort

修改 op-kube-prometheus-stack-prometheus 的类型为 NodePort

1
2
3

kubectl edit svc op-kube-prometheus-stack-prometheus -n monitor
# ClusterIP 改为 NodePort
type: NodePort

查看安装结果

1
2
3

kubectl --namespace monitor get pods -l "release=op"
kubectl --namespace monitor get svc | grep op
kubectl --namespace monitor get pod | grep op

访问 任意节点ip:NodePort即可

Grafana 的默认用户名为 admin，密码为prom-operator

root@m1:/yaml/kube-prometheus-stack# kubectl --namespace monitor get svc | grep op
alertmanager-operated                   ClusterIP   None            <none>        9093/TCP,9094/TCP,9094/UDP   103m
op-grafana                              NodePort    10.43.40.251    <none>        80:31844/TCP                 121m
op-kube-prometheus-stack-alertmanager   ClusterIP   10.43.152.63    <none>        9093/TCP                     121m
op-kube-prometheus-stack-operator       ClusterIP   10.43.206.216   <none>        443/TCP                      121m
op-kube-prometheus-stack-prometheus     NodePort    10.43.113.100   <none>        9090:32088/TCP               121m
op-kube-state-metrics                   ClusterIP   10.43.99.132    <none>        8080/TCP                     121m
op-prometheus-node-exporter             ClusterIP   10.43.235.11    <none>        9100/TCP                     121m
prometheus-operated                     ClusterIP   None            <none>        9090/TCP                     103m

# op-grafana 的 NodePort
http://192.168.0.67:31844

# op-kube-prometheus-stack-prometheus 的 NodePort
http://192.168.0.67:32088