kube-prometheus 是一整套监控解决方案，它使用 Prometheus 采集集群指标，Grafana 做展示，包含如下组件：

• The Prometheus Operator
• Highly available Prometheus
• Highly available Alertmanager
• Prometheus node-exporter
• Prometheus Adapter for Kubernetes Metrics APIs （k8s-prometheus-adapter）
• kube-state-metrics
• Grafana

其中 k8s-prometheus-adapter 使用 Prometheus 实现了 metrics.k8s.io 和 custom.metrics.k8s.io API，所以不需要再部署 metrics-server。 如果要单独部署 metrics-server，请参考：C.metrics-server插件.md

如果没有特殊指明，本文档的所有操作均在 k8s-01 节点上执行；

一、 下载和安装

原本是要把Registry的地址换成微软中国的地址的，但是发现貌似从2020年4月份开始就用不了了，网上也是有很多的网友反映，暂时未能找到可以替代的镜像仓库地址。

解决方法：

我的办法是在一台能科学上网的服务器上把需要用到的镜像先pull下来，然后推送到阿里云的镜像仓库中，在安装的时候，在三台主机(k8s-01、k8s-02、k8s-03)上将阿里云镜像仓库中的相关镜像pull下来，然后再改一下tag就行了。为了保证所需镜像的版本保持不变，我已经将https://github.com/coreos/kube-prometheus.git fork到了自己的github中，地址为https://github.com/wangchaoforever/kube-prometheus.git

$ cd /opt/k8s/work

$ git clone https://github.com/wangchaoforever/kube-prometheus.git

# 拉取镜像
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/alertmanager:v0.20.0
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/grafana:6.6.0
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/kube-state-metrics:v1.9.5
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/kube-rbac-proxy:v0.4.1
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/node-exporter:v0.18.1
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/k8s-prometheus-adapter-amd64:v0.5.0
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/prometheus:v2.15.2
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/prometheus-operator:v0.38.1
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/configmap-reload:v0.3.0
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/prometheus-config-reloader:v0.38.1
docker pull registry.cn-hangzhou.aliyuncs.com/wc181/pause:3.1

# 修改tag
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/alertmanager:v0.20.0 quay.io/prometheus/alertmanager:v0.20.0
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/grafana:6.6.0 grafana/grafana:6.6.0
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/kube-state-metrics:v1.9.5 quay.io/coreos/kube-state-metrics:v1.9.5
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/kube-rbac-proxy:v0.4.1 quay.io/coreos/kube-rbac-proxy:v0.4.1
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/node-exporter:v0.18.1 quay.io/prometheus/node-exporter:v0.18.1
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/k8s-prometheus-adapter-amd64:v0.5.0 quay.io/coreos/k8s-prometheus-adapter-amd64:v0.5.0
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/prometheus:v2.15.2 quay.io/prometheus/prometheus:v2.15.2
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/prometheus-operator:v0.38.1 quay.io/coreos/prometheus-operator:v0.38.1
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/configmap-reload:v0.3.0 jimmidyson/configmap-reload:v0.3.0
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/prometheus-config-reloader:v0.38.1 quay.io/coreos/prometheus-config-reloader:v0.38.1
docker tag registry.cn-hangzhou.aliyuncs.com/wc181/pause:3.1 k8s.gcr.io/pause:3.1

# 删除多余镜像
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/alertmanager:v0.20.0
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/grafana:6.6.0
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/kube-state-metrics:v1.9.5
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/kube-rbac-proxy:v0.4.1
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/node-exporter:v0.18.1
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/k8s-prometheus-adapter-amd64:v0.5.0
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/prometheus:v2.15.2
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/prometheus-operator:v0.38.1
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/configmap-reload:v0.3.0
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/prometheus-config-reloader:v0.38.1
docker rmi -f registry.cn-hangzhou.aliyuncs.com/wc181/pause:3.1

$ cd kube-prometheus/

# 安装 prometheus-operator
$ kubectl apply -f manifests/setup 

# 安装 promethes metric adapter
$ kubectl apply -f manifests/ 

二、查看运行状态

$ kubectl get pods -n monitoring
NAME                                   READY   STATUS    RESTARTS   AGE
alertmanager-main-0                    2/2     Running   0          2m18s
alertmanager-main-1                    2/2     Running   0          2m18s
alertmanager-main-2                    2/2     Running   0          2m18s
grafana-86b55cb79f-dg9wf               1/1     Running   0          2m12s
kube-state-metrics-dbb85dfd5-27p9q     3/3     Running   0          2m12s
node-exporter-fgjcj                    2/2     Running   0          2m12s
node-exporter-rrhmv                    2/2     Running   0          2m12s
node-exporter-wpz9p                    2/2     Running   0          2m12s
prometheus-adapter-5cd5798d96-b42b7    1/1     Running   0          2m13s
prometheus-k8s-0                       3/3     Running   1          2m10s
prometheus-k8s-1                       3/3     Running   1          2m10s
prometheus-operator-5cfbdc9b67-hbrfh   2/2     Running   0          2m43s

$ kubectl top pods -n monitoring
NAME                                  CPU(cores)   MEMORY(bytes)   
alertmanager-main-0                   6m           23Mi            
alertmanager-main-1                   5m           19Mi            
grafana-86b55cb79f-dg9wf              16m          27Mi            
kube-state-metrics-dbb85dfd5-27p9q    1m           28Mi            
node-exporter-fgjcj                   2m           26Mi            
node-exporter-wpz9p                   3m           17Mi            
prometheus-adapter-5cd5798d96-b42b7   0m           17Mi            
prometheus-k8s-0                      30m          323Mi           

三、访问 Prometheus UI

启动服务代理：

$ kubectl port-forward --address 0.0.0.0 pod/prometheus-k8s-0 -n monitoring 9090:9090
Forwarding from 0.0.0.0:9090 -> 9090

• port-forward 依赖 socat。

浏览器访问：http://192.168.0.71:9090/new/graph?g0.expr=&g0.tab=1&g0.stacked=0&g0.range_input=1h

四、访问 Grafana UI

启动代理：

[root@k8s-01 ~]# kubectl port-forward --address 0.0.0.0 svc/grafana -n monitoring 3000:3000 
Forwarding from 0.0.0.0:3000 -> 3000

浏览器访问：http://192.168.0.71:3000/

用 admin/admin 登录：

然后就可以看到各种预定义的 dashboard 了：