关于 Kubernetes集群性能监控的一些笔记

Posted 山河已无恙

tags:

篇首语：本文由小常识网(cha138.com)小编为大家整理，主要介绍了关于 Kubernetes集群性能监控的一些笔记相关的知识，希望对你有一定的参考价值。

写在前面

  • 学习K8s涉及，整理笔记记忆
  • 博文偏实战，内容涉及：
    • 集群核心指标(Core Metrics)监控工具Metrics Server的简介
    • Metrics Server的安装Demo
    • 集群自定义指标(Custom Metrics)监控平台简介：
      • Prometheus
      • Grafana
      • NodeExporter
    • 通过helm(kube-prometheus-stack)安装监控平台的Demo

带着凡世的梦想，将污秽的灵魂依偎在纯洁的天边。它们是所有流浪、追寻、渴望与乡愁的永恒象征。 ——赫尔曼·黑塞《彼得.卡门青》

Kubernetes集群性能监控管理

Kubernetes平台搭建好后,了解Kubernetes平台及在此平台上部署的应用的运行状况,以及处理系统主要告誓及性能瓶颈,这些都依赖监控管理系统。

Kubernetes的早期版本依靠Heapster来实现完整的性能数据采集和监控功能, Kubernetes从1.8版本开始,性能数据开始以Metrics APl的方式提供标准化接口,并且从1.10版本开始将Heapster替换为MetricsServer。

在Kubernetes新的监控体系中:Metrics Server用于提供核心指标(Core Metrics) ,包括Node, Pod的CPU和内存使用指标。对其他自定义指标(Custom Metrics)的监控则由Prometheus等组件来完成。

监控节点状态,我们使用docker的话可以通过docker stats.

┌──[root@vms81.liruilongs.github.io]-[~]
└─$docker stats
CONTAINER ID   NAME                                                                                                                             CPU %     MEM USAGE / LIMIT     MEM %     NET I/O   BLOCK I/O     PIDS
781c898eea19   k8s_kube-scheduler_kube-scheduler-vms81.liruilongs.github.io_kube-system_5bd71ffab3a1f1d18cb589aa74fe082b_18                     0.15%     23.22MiB / 3.843GiB   0.59%     0B / 0B   0B / 0B       7
acac8b21bb57   k8s_kube-controller-manager_kube-controller-manager-vms81.liruilongs.github.io_kube-system_93d9ae7b5a4ccec4429381d493b5d475_18   1.18%     59.16MiB / 3.843GiB   1.50%     0B / 0B   0B / 0B       6
fe97754d3dab   k8s_calico-node_calico-node-skzjp_kube-system_a211c8be-3ee1-44a0-a4ce-3573922b65b2_14                                            4.89%     94.25MiB / 3.843GiB   2.39%     0B / 0B   0B / 4.1kB    40

那使用k8s的话，我们可以通过Metrics Server监控Pod和Node的CPU和内存资源使用数据

Metrics Server:集群性能监控平台

Metrics Server在部署完成后，将通过Kubernetes核心API Server 的/apis/metrics.k8s.io/v1beta1路径提供Pod和Node的监控数据。

安装Metrics Server

Metrics Server源代码和部署配置可以在GitHub代码库

curl -Ls https://api.github.com/repos/kubernetes-sigs/metrics-server/tarball/v0.3.6 -o metrics-server-v0.3.6.tar.gz

相关镜像

docker pull mirrorgooglecontainers/metrics-server-amd64:v0.3.6

镜像小伙伴可以下载一下，这里我已经下载好了，直接上传导入镜像

┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$ansible all -m copy -a "src=./metrics-img.tar dest=/root/metrics-img.tar"
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$ansible all -m shell -a "systemctl restart docker "
192.168.26.82 | CHANGED | rc=0 >>

192.168.26.83 | CHANGED | rc=0 >>

192.168.26.81 | CHANGED | rc=0 >>

通过docker命令导入镜像

┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$ansible all -m shell -a "docker load -i /root/metrics-img.tar"
192.168.26.83 | CHANGED | rc=0 >>
Loaded image: k8s.gcr.io/metrics-server-amd64:v0.3.6
192.168.26.81 | CHANGED | rc=0 >>
Loaded image: k8s.gcr.io/metrics-server-amd64:v0.3.6
192.168.26.82 | CHANGED | rc=0 >>
Loaded image: k8s.gcr.io/metrics-server-amd64:v0.3.6
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$

修改metrics-server-deployment.yaml

┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$mv kubernetes-sigs-metrics-server-d1f4f6f/ metrics
┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$cd metrics/
┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics]
└─$ls
cmd                 deploy      hack      OWNERS          README.md          version
code-of-conduct.md  Gopkg.lock  LICENSE   OWNERS_ALIASES  SECURITY_CONTACTS
CONTRIBUTING.md     Gopkg.toml  Makefile  pkg             vendor
┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics]
└─$cd deploy/1.8+/
┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics/deploy/1.8+]
└─$ls
aggregated-metrics-reader.yaml  metrics-apiservice.yaml         resource-reader.yaml
auth-delegator.yaml             metrics-server-deployment.yaml
auth-reader.yaml                metrics-server-service.yaml

这里修改一些镜像获取策略，因为Githup上的镜像拉去不下来，或者拉去比较麻烦，所以我们提前上传好

┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics/deploy/1.8+]
└─$vim metrics-server-deployment.yaml

 31       - name: metrics-server
 32         image: k8s.gcr.io/metrics-server-amd64:v0.3.6
 33         #imagePullPolicy: Always
 34         imagePullPolicy: IfNotPresent
 35         command:
 36         - /metrics-server
 37         - --metric-resolution=30s
 38         - --kubelet-insecure-tls
 39         - --kubelet-preferred-address-types=InternalIP
 40         volumeMounts:

运行资源文件，创建相关资源对象

┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics/deploy/1.8+]
└─$kubectl apply -f .

查看pod列表，metrics-server创建成功

┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics/deploy/1.8+]
└─$kubectl  get pods -n kube-system
NAME                                                 READY   STATUS    RESTARTS   AGE
calico-kube-controllers-78d6f96c7b-79xx4             1/1     Running   2          3h15m
calico-node-ntm7v                                    1/1     Running   1          12h
calico-node-skzjp                                    1/1     Running   4          12h
calico-node-v7pj5                                    1/1     Running   1          12h
coredns-545d6fc579-9h2z4                             1/1     Running   2          3h15m
coredns-545d6fc579-xgn8x                             1/1     Running   2          3h16m
etcd-vms81.liruilongs.github.io                      1/1     Running   1          13h
kube-apiserver-vms81.liruilongs.github.io            1/1     Running   2          13h
kube-controller-manager-vms81.liruilongs.github.io   1/1     Running   4          13h
kube-proxy-rbhgf                                     1/1     Running   1          13h
kube-proxy-vm2sf                                     1/1     Running   1          13h
kube-proxy-zzbh9                                     1/1     Running   1          13h
kube-scheduler-vms81.liruilongs.github.io            1/1     Running   5          13h
metrics-server-bcfb98c76-gttkh                       1/1     Running   0          70m

通过kubectl top nodes命令测试，

┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics/deploy/1.8+]
└─$kubectl top nodes
W1007 14:23:06.102605  102831 top_node.go:119] Using json format to get metrics. Next release will switch to protocol-buffers, switch early by passing --use-protocol-buffers flag
NAME                         CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%
vms81.liruilongs.github.io   555m         27%    2025Mi          52%
vms82.liruilongs.github.io   204m         10%    595Mi           15%
vms83.liruilongs.github.io   214m         10%    553Mi           14%
┌──[root@vms81.liruilongs.github.io]-[~/ansible/metrics/deploy/1.8+]
└─$

Prometheus+Grafana+NodeExporter:集群监控平台

在各个计算节点上部署NodeExporter采集CPU、内存、磁盘及IO信息,并将这些信息传输给监控节点上的Prometheus服务器进行存储分析,通过Grafana进行可视化监控,

Prometheus

Prometheus是一款开源的监控解决方案,由SoundCloud公司开发的开源监控系统，是继Kubernetes之后CNCF第2个孵化成功的项目，在容器和微服务领域得到了广泛应用,能在监控Kubernetes平台的同时监控部署在此平台中的应用,它提供了一系列工具集及多维度监控指标。Prometheus依赖Grafana实现数据可视化。

Prometheus的主要特点如下:

  • 使用指标名称及键值对标识的多维度数据模型。
  • 采用灵活的查询语言PromQL。
  • 不依赖分布式存储，为自治的单节点服务。
  • 使用HTTP完成对监控数据的拉取。
  • 支持通过网关推送时序数据。
  • 支持多种图形和Dashboard的展示，例如Grafana。

Prometheus生态系统由各种组件组成，用于功能的扩充:

组件描述
Prometheus Server负责监控数据采集和时序数据存储，并提供数据查询功能。
客户端SDK对接Prometheus的开发工具包。
Push Gateway推送数据的网关组件。
第三方Exporter各种外部指标收集系统,其数据可以被Prometheus采集
AlertManager告警管理器。
其他辅助支持工具–

Prometheus的核心组件Prometheus Server的主要功能包括:

从Kubernetes Master获取需要监控的资源或服务信息;从各种Exporter抓取(Pull)指标数据,然后将指标数据保存在时序数据库(TSDB)中;向其他系统提供HTTP API进行查询;提供基于PromQL语言的数据查询;可以将告警数据推送(Push)给AlertManager,等等。

Prometheus的系统架构:

NodeExporter

NodeExporter主要用来采集服务器CPU、内存、磁盘、IO等信息,是机器数据的通用采集方案。只要在宿主机上安装NodeExporter和cAdisor容器,通过Prometheus进行抓取即可。它同Zabbix的功能相似.

Grafana

Grafana是一个Dashboard工具,用Go和JS开发,它是一个时间序列数据库的界面展示层,通过SQL命令查询出Metrics并将结果展示出来。它能自定义多种仪表盘,可以轻松实现覆盖多个Docker的宿主机监控信息的展现。

搭建Prometheus+Grafana+NodeExporter平台

这里我们通过helm的方式搭建，简单方便快捷，运行之后，相关的镜像都会创建成功.下面是创建成功的镜像列表。

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl get pods
NAME                                                    READY   STATUS    RESTARTS      AGE
alertmanager-liruilong-kube-prometheus-alertmanager-0   2/2     Running   0             61m
liruilong-grafana-5955564c75-zpbjq                      3/3     Running   0             62m
liruilong-kube-prometheus-operator-5cb699b469-fbkw5     1/1     Running   0             62m
liruilong-kube-state-metrics-5dcf758c47-bbwt4           1/1     Running   7 (32m ago)   62m
liruilong-prometheus-node-exporter-rfsc5                1/1     Running   0             62m
liruilong-prometheus-node-exporter-vm7s9                1/1     Running   0             62m
liruilong-prometheus-node-exporter-z9j8b                1/1     Running   0             62m
prometheus-liruilong-kube-prometheus-prometheus-0       2/2     Running   0             61m
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$

环境版本

我的K8s集群版本

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$kubectl  get nodes
NAME                         STATUS   ROLES                  AGE   VERSION
vms81.liruilongs.github.io   Ready    control-plane,master   34d   v1.22.2
vms82.liruilongs.github.io   Ready    <none>                 34d   v1.22.2
vms83.liruilongs.github.io   Ready    <none>                 34d   v1.22.2

hrlm版本

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$helm version
version.BuildInfoVersion:"v3.2.1", GitCommit:"fe51cd1e31e6a202cba7dead9552a6d418ded79a", GitTreeState:"clean", GoVersion:"go1.13.10"

prometheus-operator(旧名字)安装出现的问题

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$helm  search repo prometheus-operator
NAME                            CHART VERSION   APP VERSION     DESCRIPTION
ali/prometheus-operator         8.7.0           0.35.0          Provides easy monitoring definitions for Kubern...
azure/prometheus-operator       9.3.2           0.38.1          DEPRECATED Provides easy monitoring definitions...
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$helm install liruilong  ali/prometheus-operator
Error: failed to install CRD crds/crd-alertmanager.yaml: unable to recognize "": no matches for kind "CustomResourceDefinition" in version "apiextensions.k8s.io/v1beta1"
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$helm pull  ali/prometheus-operator
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$

解决办法:新版本安装

直接下载kube-prometheus-stack(新)的chart包，通过命令安装：

https://github.com/prometheus-community/helm-charts/releases/download/kube-prometheus-stack-30.0.1/kube-prometheus-stack-30.0.1.tgz

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$ls
index.yaml  kube-prometheus-stack-30.0.1.tgz  liruilonghelm  liruilonghelm-0.1.0.tgz  mysql  mysql-1.6.4.tgz
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$helm list
NAME    NAMESPACE       REVISION        UPDATED STATUS  CHART   APP VERSION

解压chart包kube-prometheus-stack-30.0.1.tgz

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$tar -zxf kube-prometheus-stack-30.0.1.tgz

创建新的命名空间

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$cd kube-prometheus-stack/
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl create ns monitoring
namespace/monitoring created
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl config  set-context $(kubectl config current-context) --namespace=monitoring
Context "kubernetes-admin@kubernetes" modified.

进入文件夹，直接通过helm install liruilong .安装

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$ls
Chart.lock  charts  Chart.yaml  CONTRIBUTING.md  crds  README.md  templates  values.yaml
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$helm install liruilong .

kube-prometheus-admission-create对应Pod的相关镜像下载不下来问题

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$kubectl get pods
NAME                                                  READY   STATUS             RESTARTS   AGE
liruilong-kube-prometheus-admission-create--1-bn7x2   0/1     ImagePullBackOff   0          33s

查看pod详细信息，发现是谷歌的一个镜像国内无法下载

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$kubectl  describe pod  liruilong-kube-prometheus-admission-create--1-bn7x2
Name:         liruilong-kube-prometheus-admission-create--1-bn7x2
Namespace:    monitoring
Priority:     0
Node:         vms83.liruilongs.github.io/192.168.26.83
Start Time:   Sun, 16 Jan 2022 02:43:07 +0800
Labels:       app=kube-prometheus-stack-admission-create
              app.kubernetes.io/instance=liruilong
              app.kubernetes.io/managed-by=Helm
              app.kubernetes.io/part-of=kube-prometheus-stack
              app.kubernetes.io/version=30.0.1
              chart=kube-prometheus-stack-30.0.1
              controller-uid=2ce48cd2-a118-4e23-a27f-0228ef6c45e7
              heritage=Helm
              job-name=liruilong-kube-prometheus-admission-create
              release=liruilong
Annotations:  cni.projectcalico.org/podIP: 10.244.70.8/32
              cni.projectcalico.org/podIPs: 10.244.70.8/32
Status:       Pending
IP:           10.244.70.8
IPs:
  IP:           10.244.70.8
Controlled By:  Job/liruilong-kube-prometheus-admission-create
Containers:
  create:
    Container ID:
    Image:         k8s.gcr.io/ingress-nginx/kube-webhook-certgen:v1.0@sha256:f3b6b39a6062328c095337b4cadcefd1612348fdd5190b1dcbcb9b9e90bd8068
    Image ID:
    Port:          <none>
    Host Port:
    。。。。。。。。。。。。。。。。。。。。。。。。。。。

在dokcer仓库里找了一个类似的，通过 kubectl edit修改

image: k8s.gcr.io/ingress-nginx/kube-webhook-certgen:v1.0  替换为 ： docker.io/liangjw/kube-webhook-certgen:v1.1.1

或者也可以修改配置文件从新install(记得要把sha注释掉)

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$ls
index.yaml  kube-prometheus-stack  kube-prometheus-stack-30.0.1.tgz  liruilonghelm  liruilonghelm-0.1.0.tgz  mysql  mysql-1.6.4.tgz
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create]
└─$cd kube-prometheus-stack/
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$ls
Chart.lock  charts  Chart.yaml  CONTRIBUTING.md  crds  README.md  templates  values.yaml
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$cat values.yaml | grep -A 3 -B 2 kube-webhook-certgen
      enabled: true
      image:
        repository: docker.io/liangjw/kube-webhook-certgen
        tag: v1.1.1
        #sha: "f3b6b39a6062328c095337b4cadcefd1612348fdd5190b1dcbcb9b9e90bd8068"
        pullPolicy: IfNotPresent
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$helm del liruilong;helm install liruilong .

之后其他的相关pod正常创建中

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl get pods
NAME                                                  READY   STATUS              RESTARTS   AGE
liruilong-grafana-5955564c75-zpbjq                    0/3     ContainerCreating   0          27s
liruilong-kube-prometheus-operator-5cb699b469-fbkw5   0/1     ContainerCreating   0          27s
liruilong-kube-state-metrics-5dcf758c47-bbwt4         0/1     ContainerCreating   0          27s
liruilong-prometheus-node-exporter-rfsc5              0/1     ContainerCreating   0          28s
liruilong-prometheus-node-exporter-vm7s9              0/1     ContainerCreating   0          28s
liruilong-prometheus-node-exporter-z9j8b              0/1     ContainerCreating   0          28s

kube-state-metrics这个pod的镜像也没有拉取下来。应该也是相同的原因

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl get pods
NAME                                                    READY   STATUS             RESTARTS   AGE
alertmanager-liruilong-kube-prometheus-alertmanager-0   2/2     Running            0          3m35s
liruilong-grafana-5955564c75-zpbjq                      3/3     Running            0          4m46s
liruilong-kube-prometheus-operator-5cb699b469-fbkw5     1/1     Running            0          4m46s
liruilong-kube-state-metrics-5dcf758c47-bbwt4           0/1     ImagePullBackOff   0          4m46s
liruilong-prometheus-node-exporter-rfsc5                1/1     Running            0          4m47s
liruilong-prometheus-node-exporter-vm7s9                1/1     Running            0          4m47s
liruilong-prometheus-node-exporter-z9j8b                1/1     Running            0          4m47s
prometheus-liruilong-kube-prometheus-prometheus-0       2/2     Running            0          3m34s

同样 k8s.gcr.io/kube-state-metrics/kube-state-metrics:v2.3.0 这个镜像没办法拉取

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl  describe  pod liruilong-kube-state-metrics-5dcf758c47-bbwt4
Name:         liruilong-kube-state-metrics-5dcf758c47-bbwt4
Namespace:    monitoring
Priority:     0
Node:         vms82.liruilongs.github.io/192.168.26.82
Start Time:   Sun, 16 Jan 2022 02:59:53 +0800
Labels:       app.kubernetes.io/component=metrics
              app.kubernetes.io/instance=liruilong
              app.kubernetes.io/managed-by=Helm
              app.kubernetes.io/name=kube-state-metrics
              app.kubernetes.io/part-of=kube-state-metrics
              app.kubernetes.io/version=2.3.0
              helm.sh/chart=kube-state-metrics-4.3.0
              pod-template-hash=5dcf758c47
              release=liruilong
Annotations:  cni.projectcalico.org/podIP: 10.244.171.153/32
              cni.projectcalico.org/podIPs: 10.244.171.153/32
Status:       Pending
IP:           10.244.171.153
IPs:
  IP:           10.244.171.153
Controlled By:  ReplicaSet/liruilong-kube-state-metrics-5dcf758c47
Containers:
  kube-state-metrics:
    Container ID:
    Image:         k8s.gcr.io/kube-state-metrics/kube-state-metrics:v2.3.0
    Image ID:
    Port:          8080/TCP
    。。。。。。。。。。。。。。。。。。。。。。

同样的，我们通过docker仓库找一下相同的,然后通过kubectl edit pod修改一下

k8s.gcr.io/kube-state-metrics/kube-state-metrics 替换为： docker.io/dyrnq/kube-state-metrics:v2.3.0

可以先在节点机上拉取一下

┌──[root@vms81.liruilongs.github.io]-[~/ansible]
└─$ ansible node -m shell -a "docker pull dyrnq/kube-state-metrics:v2.3.0"
192.168.26.82 | CHANGED | rc=0 >>
v2.3.0: Pulling from dyrnq/kube-state-metrics
e8614d09b7be: Pulling fs layer
53ccb90bafd7: Pulling fs layer
e8614d09b7be: Verifying Checksum
e8614d09b7be: Download complete
e8614d09b7be: Pull complete
53ccb90bafd7: Verifying Checksum
53ccb90bafd7: Download complete
53ccb90bafd7: Pull complete
Digest: sha256:c9137505edaef138cc23479c73e46e9a3ef7ec6225b64789a03609c973b99030
Status: Downloaded newer image for dyrnq/kube-state-metrics:v2.3.0
docker.io/dyrnq/kube-state-metrics:v2.3.0
192.168.26.83 | CHANGED | rc=0 >>
v2.3.0: Pulling from dyrnq/kube-state-metrics
e8614d09b7be: Pulling fs layer
53ccb90bafd7: Pulling fs layer
e8614d09b7be: Verifying Checksum
e8614d09b7be: Download complete
e8614d09b7be: Pull complete
53ccb90bafd7: Verifying Checksum
53ccb90bafd7: Download complete
53ccb90bafd7: Pull complete
Digest: sha256:c9137505edaef138cc23479c73e46e9a3ef7ec6225b64789a03609c973b99030
Status: Downloaded newer image for dyrnq/kube-state-metrics:v2.3.0
docker.io/dyrnq/kube-state-metrics:v2.3.0

修改完之后，会发现所有的pod都创建成功

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl get pods
NAME                                                    READY   STATUS    RESTARTS      AGE
alertmanager-liruilong-kube-prometheus-alertmanager-0   2/2     Running   0             61m
liruilong-grafana-5955564c75-zpbjq                      3/3     Running   0             62m
liruilong-kube-prometheus-operator-5cb699b469-fbkw5     1/1     Running   0             62m
liruilong-kube-state-metrics-5dcf758c47-bbwt4           1/1     Running   7 (32m ago)   62m
liruilong-prometheus-node-exporter-rfsc5                1/1     Running   0             62m
liruilong-prometheus-node-exporter-vm7s9                1/1     Running   0             62m
liruilong-prometheus-node-exporter-z9j8b                1/1     Running   0             62m
prometheus-liruilong-kube-prometheus-prometheus-0       2/2     Running   0             61m
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$

然后我们需要修改liruilong-grafana SVC的类型为NodePort，这样，物理机就可以访问了

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack/templates]
└─$kubectl  get svc
NAME                                     TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)                      AGE
alertmanager-operated                    ClusterIP   None             <none>        9093/TCP,9094/TCP,9094/UDP   33m
liruilong-grafana                        ClusterIP   10.99.220.121    <none>        80/TCP                       34m
liruilong-kube-prometheus-alertmanager   ClusterIP   10.97.193.228    <none>        9093/TCP                     34m
liruilong-kube-prometheus-operator       ClusterIP   10.101.106.93    <none>        443/TCP                      34m
liruilong-kube-prometheus-prometheus     ClusterIP   10.105.176.19    <none>        9090/TCP                     34m
liruilong-kube-state-metrics             ClusterIP   10.98.94.55      <none>        8080/TCP                     34m
liruilong-prometheus-node-exporter       ClusterIP   10.110.216.215   <none>        9100/TCP                     34m
prometheus-operated                      ClusterIP   None             <none>        9090/TCP                     33m
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack]
└─$kubectl edit svc liruilong-grafana
service/liruilong-grafana edited
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack/templates]
└─$kubectl  get svc
NAME                                     TYPE        CLUSTER-IP       EXTERNAL-IP   PORT(S)                      AGE
alertmanager-operated                    ClusterIP   None             <none>        9093/TCP,9094/TCP,9094/UDP   35m
liruilong-grafana                        NodePort    10.99.220.121    <none>        80:30443/TCP                 36m
liruilong-kube-prometheus-alertmanager   ClusterIP   10.97.193.228    <none>        9093/TCP                     36m
liruilong-kube-prometheus-operator       ClusterIP   10.101.106.93    <none>        443/TCP                      36m
liruilong-kube-prometheus-prometheus     ClusterIP   10.105.176.19    <none>        9090/TCP                     36m
liruilong-kube-state-metrics             ClusterIP   10.98.94.55      <none>        8080/TCP                     36m
liruilong-prometheus-node-exporter       ClusterIP   10.110.216.215   <none>        9100/TCP                     36m
prometheus-operated                      ClusterIP   None             <none>        9090/TCP                     35m
物理机访问

通过secrets解密获取用户名密码

┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack/templates]
└─$kubectl get secrets | grep grafana
liruilong-grafana                                                  Opaque                                3      38m
liruilong-grafana-test-token-q8z8j                                 kubernetes.io/service-account-token   3      38m
liruilong-grafana-token-j94p8                                      kubernetes.io/service-account-token   3      38m
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack/templates]
└─$kubectl get secrets liruilong-grafana -o yaml
apiVersion: v1
data:
  admin-password: cHJvbS1vcGVyYXRvcg==
  admin-user: YWRtaW4=
  ldap-toml: ""
kind: Secret
metadata:
  annotations:
    meta.helm.sh/release-name: liruilong
    meta.helm.sh/release-namespace: monitoring
  creationTimestamp: "2022-01-15T18:59:40Z"
  labels:
    app.kubernetes.io/instance: liruilong
    app.kubernetes.io/managed-by: Helm
    app.kubernetes.io/name: grafana
    app.kubernetes.io/version: 8.3.3
    helm.sh/chart: grafana-6.20.5
  name: liruilong-grafana
  namespace: monitoring
  resourceVersion: "1105663"
  uid: c03ff5f3-deb5-458c-8583-787f41034469
type: Opaque
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack/templates]
└─$kubectl get secrets liruilong-grafana -o jsonpath='.data.admin-user'| base64 -d
adminbase64: 输入无效
┌──[root@vms81.liruilongs.github.io]-[~/ansible/k8s-helm-create/kube-prometheus-stack/templates]
└─$kubectl get secrets liruilong-grafana -o jsonpath='.data.admin-password'| base64 -d
prom-operatorbase64: 输入无效

得到用户名密码：admin/prom-operator

正常登录，查看监控信息

以上是关于关于 Kubernetes集群性能监控的一些笔记的主要内容，如果未能解决你的问题，请参考以下文章

关于 Kubernetes使用kubeadm重置集群的一些笔记

关于 Kubernetes使用kubeadm重置集群的一些笔记

关于Kubernetes集群中常见问题的排查方法的一些笔记

关于Kubernetes中如何优雅的访问集群外服务的一些笔记

关于Kubernetes中如何优雅的访问集群外服务的一些笔记

关于Kubernetes集群中常见问题的排查方法的一些笔记