helm 安装 prometheus loki grafana

前言

我有个应用需要接入一下指标跟日志收集，就采用k8s安装了 prometheus loki grafana

安装prometheus跟grafana

这个就比较简单，直接使用bitnami的charts就行

这里感谢一下大佬，谢谢大佬的镜像仓库，很好用！大佬地址

配置文件如下：

bash 复制代码

# prometheus-values.yml
global:
  imageRegistry: "dockerpull.org"
server:
#我这里使用clusterIP的原因是因为我有traefik作为网关代理
  service:
    type: ClusterIP
alertmanager:
  service:
    type: ClusterIP

bash 复制代码

# grafana-values.yml
global:
  imageRegistry: "dockerpull.org"
admin:
  password: "yourPassword"
persistence:
  enabled: false

安装

bash 复制代码

helm install -f prometheus-values.yml prometheus bitnami/prometheus -n monitor
helm install -f grafana-values.yml grafana bitnami/grafana -n monitor

装完了之后，使用traefik配置一下两个的域名，也就是配一下ingressRoute

bash 复制代码

apiVersion: traefik.io/v1alpha1
kind: IngressRoute
metadata:
  name: grafana-web--com-tls
  namespace: monitor  
spec:
  entryPoints:
    - websecure
  routes:
  - match: Host(`grafana.xxx.com`)
    kind: Rule
    services:
    - name: grafana
      port: 3000

bash 复制代码

apiVersion: traefik.io/v1alpha1
kind: IngressRoute
metadata:
  name: prometheus-web-com-tls
  namespace: monitor  
spec:
  entryPoints:
    - websecure
  routes:
  - match: Host(`prometheus.xxx.com`)
    kind: Rule
    services:
    - name: prometheus-server
      port: 80

在grafana里面配置一下数据源

安装loki

这里为什么要把loki单独拿出来讲呢，因为bitnami没有，至少截止至2024/12/04是没有的

这里我就只能使用官方的chart了也就是 grafana/loki 但是呢官方的chart配置太臃肿了，我们想定制也很麻烦，比如service的type, persitence 的 storageclass或者 resource-size，还有官方的 golbal.image.registry居然不生效，我没那么多时间去单独写一个chart了，我只能在官方的chart的values.yml文件上改了，我改了拉取镜像的registry以及storage

文件如下下载

安装

bash 复制代码

helm install -f loki-values.yml loki grafana/loki -n monitor

对接grafana

chart安装成功后 console会打印如下注释

bash 复制代码

***********************************************************************
Connecting Grafana to Loki
***********************************************************************

If Grafana operates within the cluster, you'll set up a new Loki datasource by utilizing the following URL:

http://loki-gateway.monitor.svc.cluster.local/

***********************************************************************
Multi-tenancy
***********************************************************************

Loki is configured with auth enabled (multi-tenancy) and expects tenant headers (`X-Scope-OrgID`) to be set for all API calls.

You must configure Grafana's Loki datasource using the `HTTP Headers` section with the `X-Scope-OrgID` to target a specific tenant.
For each tenant, you can create a different datasource.

The agent of your choice must also be configured to propagate this header.
For example, when using Promtail you can use the `tenant` stage. https://grafana.com/docs/loki/latest/send-data/promtail/stages/tenant/

When not provided with the `X-Scope-OrgID` while auth is enabled, Loki will reject reads and writes with a 404 status code `no org id`.

You can also use a reverse proxy, to automatically add the `X-Scope-OrgID` header as suggested by https://grafana.com/docs/loki/latest/operations/authentication/

For more information, read our documentation about multi-tenancy: https://grafana.com/docs/loki/latest/operations/multi-tenancy/

按照文档操作即可

我这里的url 是 http://loki-gateway ,跟文档不一样的原因是我的grafana跟loki在同一个 namespace，可以省略 .svc.cluster.local