Kubernetes之利用prometheus监控K8S集群
prometheus它是一个主动拉取的数据库，在K8S中应该展示图形的grafana数据实例化要保存下来，使用分布式文件系统加动态PV，但是在本测试环境中使用本地磁盘，安装采集数据的agent使用DaemonSet来部署，DaemonSet的特性就是在每个node上部署一个服务进程，这一切都是自动的部署。
此处只讲如何用prometheus来监控K8S集群，关于prometheus的知识参考官方文档。另外有
部署前提： 准备好所需要的文件
$ ls -l
Prometheus/prometheus#:/data/Prometheus/prometheus# ls -l
total 28
drwxr-xr-x 2 root root 4096 Jan 15 02:53 grafana
drwxr-xr-x 2 root root 4096 Jan 15 03:11 kube-state-metrics
-rw-r--r-- 1 root root 60 Jan 14 06:48 namespace.yaml
drwxr-xr-x 2 root root 4096 Jan 15 03:22 node-directory-size-metrics
drwxr-xr-x 2 root root 4096 Jan 15 03:02 node-exporter
drwxr-xr-x 2 root root 4096 Jan 15 02:55 prometheus
drwxr-xr-x 2 root root 4096 Jan 15 02:37 rbac $ ls grafana/
grafana-configmap.yaml grafana-core-deployment.yaml grafana-import-dashboards-job.yaml grafana-pvc-claim.yaml grafana-pvc-volume.yaml grafana-service.yaml $ ls prometheus/
configmap.yaml deployment.yaml prometheus-rules.yaml service.yaml
grafana和 prometheus 都是部署文件，node-exporter、kube-state-metrics、node-directory-size-metrics这三个是采集器，相当于prometheus的agent
文件准备好了，现在开始一步一步来部署：
1，,创建所需Namespace
因为prometheus 部署的所有的deploy、pod、svc都是在monitoring完成的，所以需要事先创建之。
$ cat namespace.yaml
apiVersion: v1
kind: Namespace
metadata:
name: monitoring
$ kubectl create -f namespace.yaml
namespace "monitoring" created
2，创建grafana的pv、 pvc
grafana# cat grafana-pvc-volume.yaml
kind: PersistentVolume
apiVersion: v1
metadata:
name: grafana-pv-volume
labels:
type: local
spec:
storageClassName: grafana-pv-volume
capacity:
storage: 10Gi
accessModes:
- ReadWriteOnce
persistentVolumeReclaimPolicy: Recycle
hostPath:
path: "/data/volume/grafana"
grafana# cat grafana-pvc-claim.yaml
kind: PersistentVolumeClaim
apiVersion: v1
metadata:
name: grafana-pvc-volume
namespace: "monitoring"
spec:
accessModes:
- ReadWriteOnce
resources:
requests:
storage: 5Gi
storageClassName: grafana-pv-volume
$ kubectl create -f grafana/grafana-pvc-volume.yaml -f grafana/grafana-pvc-claim.yaml
persistentvolume "grafana-pv-volume" created
persistentvolumeclaim "grafana-pvc-volume" created $ kubectl get pvc -n monitoring
NAME STATUS VOLUME CAPACITY ACCESS MODES STORAGECLASS AGE
grafana-pvc-volume Bound grafana-pv-volume 10Gi RWO grafana-pv-volume 52s
状态bound已绑定到了 grafana-pv-volume
3，创建grafana应用，这些应用都是第三方的，都会有自已的配置，通过configmap来定义
grafana# ls
grafana-configmap.yaml grafana-core-deployment.yaml grafana-import-dashboards-job.yaml grafana-pvc-claim.yaml grafana-pvc-volume.yaml grafana-service.yaml
grafana# kubectl create -f ./ #grafana目录下所有文件都创建
configmap "grafana-import-dashboards" created
deployment "grafana-core" created
job "grafana-import-dashboards" created
service "grafana" created grafana# kubectl get deployment,pod -n monitoring
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
deploy/grafana-core 1 1 1 0 1m NAME READY STATUS RESTARTS AGE
po/grafana-core-9c7f66868-7q8lx 0/1 ContainerCreating 0 1m
运行po/grafana-core 容器时会下载镜像： grafana/grafana:4.2.0
grafana创建的应用 简单的自已描述了下：
grafana-pv-volume=/data/volume/grafana =10G
grafana-pvc-volume=5G--->grafana-pv-volume
---configmap=grafana-import-dashboards
Job=grafana-import-dashboards
Deployment=grafana-core replicas: 1 containers=grafana-core mount: grafana-pvc-volume:/var
service=grafana port: 3000 = nodePort: 30161 (3000是grafana服务的默认端口)
4， 现在grafana的核心应用已部署好了，现在来部署prometheus的RBAC
prometheus/rbac# ls
grant_serviceAccount.sh prometheus_rbac.yaml
#先创建RBAC文件：
prometheus/rbac# kubectl create -f prometheus_rbac.yaml
clusterrolebinding "prometheus-k8s" created
clusterrolebinding "kube-state-metrics" created
clusterrole "kube-state-metrics" created
serviceaccount "kube-state-metrics" created
clusterrolebinding "prometheus" created
clusterrole "prometheus" created
serviceaccount "prometheus-k8s" created
prometheus/rbac#
5，创建prometheus的deloyment,service
prometheus/prometheus# ls
configmap.yaml deployment.yaml prometheus-rules.yaml service.yaml
prometheus/prometheus#
在configmap.yaml中要注意的是在1.7以后，获取cadvsion监控pod等的信息时，用的是kubelet的4194端口，
注意以下这段：这是采集cadvision信息，必须是通过kubelet的4194端口，所以Kubelet必须监听着，4194部署了cadvsion来获取pod中容器信息
prometheus/prometheus#cat configmap.yaml
# https://github.com/prometheus/prometheus/blob/master/documentation/examples/prometheus-kubernetes.yml#L37
- job_name: 'kubernetes-nodes'
tls_config:
ca_file: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt
bearer_token_file: /var/run/secrets/kubernetes.io/serviceaccount/token
kubernetes_sd_configs:
- role: node
relabel_configs:
- source_labels: [__address__]
regex: '(.*):10250'
replacement: '${1}:10255'
target_label: __address__
- job_name: 'kubernetes-cadvisor'
scheme: https
tls_config:
ca_file: /var/run/secrets/kubernetes.io/serviceaccount/ca.crt
bearer_token_file: /var/run/secrets/kubernetes.io/serviceaccount/token
kubernetes_sd_configs:
- role: node
relabel_configs:
- action: labelmap
regex: __meta_kubernetes_node_label_(.+)
- target_label: __address__
replacement: kubernetes.default.svc.cluster.local:443
- source_labels: [__meta_kubernetes_node_name]
regex: (.+)
target_label: __metrics_path__
replacement: /api/v1/nodes/${1}:4194/proxy/metrics # https://github.com/prometheus/prometheus/blob/master/documentation/examples/prometheus-kubernetes.yml#L79
prometheus-rules.yaml 这是它的发现规则文件
deployment.yaml service.yaml 这两个是部署的文件， deployment部署中资源限制建议放大一点
现在部署prometheus目录下所有文件：
prometheus/prometheus# kubectl create -f ./
configmap "prometheus-core" created
deployment "prometheus-core" created
configmap "prometheus-rules" created
service "prometheus" created
prometheus/prometheus# prometheus/prometheus# kubectl get deployment,pod -n monitoring
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
deploy/grafana-core 1 1 1 1 16m
deploy/prometheus-core 1 1 1 1 1m NAME READY STATUS RESTARTS AGE
po/grafana-core-9c7f66868-wm68j 1/1 Running 0 16m
po/prometheus-core-6dc6777c5b-5nc7j 1/1 Running 0 1m
prometheus应用的部署，简单描述下创建的内容：
Deployment= prometheus-core replicas: 1 containers=prometheus image: prom/prometheus:v1.7.0 containerPort: 9090(webui)
Service name: prometheus NodePort-->port: 9090 -webui
6，prometheus部署完了现在来部署它的agent，也就是采集器：
Prometheus/prometheus# ls node-directory-size-metrics/
daemonset.yaml
Prometheus/prometheus# ls kube-state-metrics/
deployment.yaml service.yaml
Prometheus/prometheus# ls node-exporter/
exporter-daemonset.yaml exporter-service.yaml
Prometheus/prometheus#
#其中两个用的是daemonset Prometheus/prometheus# kubectl create -f node-exporter/ -f kube-state-metrics/ -f node-directory-size-metrics/
daemonset "prometheus-node-exporter" created
service "prometheus-node-exporter" created
deployment "kube-state-metrics" created
service "kube-state-metrics" created
daemonset "node-directory-size-metrics" created
Prometheus/prometheus# Prometheus/prometheus# kubectl get deploy,pod,svc -n monitoring
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
deploy/grafana-core 1 1 1 1 26m
deploy/kube-state-metrics 2 2 2 2 1m
deploy/prometheus-core 1 1 1 1 11m NAME READY STATUS RESTARTS AGE
po/grafana-core-9c7f66868-wm68j 1/1 Running 0 26m
po/kube-state-metrics-694fdcf55f-bqcp8 1/1 Running 0 1m
po/kube-state-metrics-694fdcf55f-nnqqd 1/1 Running 0 1m
po/node-directory-size-metrics-n9wx7 2/2 Running 0 1m
po/node-directory-size-metrics-ppscw 2/2 Running 0 1m
po/prometheus-core-6dc6777c5b-5nc7j 1/1 Running 0 11m
po/prometheus-node-exporter-kchmb 1/1 Running 0 1m
po/prometheus-node-exporter-lks5m 1/1 Running 0 1m NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
svc/grafana NodePort 10.254.231.25 <none> 3000:30161/TCP 26m
svc/kube-state-metrics ClusterIP 10.254.156.51 <none> 8080/TCP 1m
svc/prometheus NodePort 10.254.239.90 <none> 9090:37318/TCP 10m
svc/prometheus-node-exporter ClusterIP None <none> 9100/TCP 1m
Prometheus/prometheus# --------
Prometheus/prometheus# kubectl get pod -o wide -n monitoring
NAME READY STATUS RESTARTS AGE IP NODE
prometheus-node-exporter-kchmb 1/1 Running 0 4m 10.3.1.16 10.3.1.16
prometheus-node-exporter-lks5m 1/1 Running 0 4m 10.3.1.17 10.3.1.17 #这两个是exporter，用的是daemonset 分别在这两个node上运行了。这样就可以采集到所有数据了。
如上部署完成，以下是用自已的话简单描述下：
node-exporter/exporter-daemonset.yaml 文件：
DaemonSet=prometheus-node-exporter
containers： name: prometheus-node-exporter image: prom/node-exporter:v0.14.0
containerPort: 9100 hostPort: 9100 hostNetwork: true #它用的是主机的9100端口
Prometheus/prometheus/node-exporter# kubectl get daemonset,pod -n monitoring
NAME DESIRED CURRENT READY UP-TO-DATE AVAILABLE NODE SELECTOR AGE
ds/node-directory-size-metrics 2 2 2 2 2 <none> 16h
ds/prometheus-node-exporter 2 2 2 2 2 <none> 16h
因为它是daemonset，所以相应的也会运行着两个Pod： prometheus-node-exporter Service=prometheus-node-exporter clusterIP: None port: 9100 type: ClusterIP #它没有clusterIP
# kubectl get service -n monitoring
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
prometheus-node-exporter ClusterIP None <none> 9100/TCP 16h
kube-state-metrics/deployment.yaml 文件：
Deployment=kube-state-metrics replicas: 2 containers-->name: kube-state-metrics image: gcr.io/google_containers/kube-state-metrics:v0.5.0
containerPort: 8080
Service name: kube-state-metrics port: 8080 #没有映射
#kubectl get deployment,pod,svc -n monitoring
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
deploy/kube-state-metrics 2 2 2 2 16h NAME READY STATUS RESTARTS AGE
po/kube-state-metrics-694fdcf55f-2mmd5 1/1 Running 0 11h
po/kube-state-metrics-694fdcf55f-bqcp8 1/1 Running 0 16h NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
svc/kube-state-metrics ClusterIP 10.254.156.51 <none> 8080/TCP 16h
node-directory-size-metrics/daemonset.yaml 文件：
#因为是daemonset，所以未定义replicas数量，直接运行在每个node之上，但是它没有创建service
DaemonSet : name: node-directory-size-metrics
containers-->name: read-du image: giantswarm/tiny-tools mountPath: /mnt/var mountPath: /tmp
containers--> name: caddy image: dockermuenster/caddy:0.9.3 containerPort: 9102
mountPath: /var/www hostPath /var
kubectl get daemonset,pod,svc -n monitoring
NAME DESIRED CURRENT READY UP-TO-DATE AVAILABLE NODE SELECTOR AGE
ds/node-directory-size-metrics 2 2 2 2 2 <none> 16h NAME READY STATUS RESTARTS AGE
po/node-directory-size-metrics-n9wx7 2/2 Running 0 16h
po/node-directory-size-metrics-ppscw 2/2 Running 0 16h NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
没有node-directory-size-metrics的service
到此 prometheus算是部署完成了，最后来看下它暴露的端口：
Prometheus/prometheus# kubectl get svc -o wide -n monitoring
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE SELECTOR
grafana NodePort 10.254.231.25 <none> 3000:30161/TCP 31m app=grafana,component=core
kube-state-metrics ClusterIP 10.254.156.51 <none> 8080/TCP 6m app=kube-state-metrics
prometheus NodePort 10.254.239.90 <none> 9090:37318/TCP 16m app=prometheus,component=core
prometheus-node-exporter ClusterIP None <none> 9100/TCP 6m app=prometheus,component=node-exporter
Prometheus/prometheus#
7，访问、使用prometheus
如上可以看到grafana的端口号是30161，NodeIP:30161 就可以打开grafana，默认admin/admin
登录后，添加数据源：
添加Prometheus的数据源：
将Prometheus的作为数据源的相关参数如下图所示：
添加完后，导入模板文件：
部署完成。
Kubernetes之利用prometheus监控K8S集群的更多相关文章
- 基于prometheus监控k8s集群
本文建立在你已经会安装prometheus服务的基础之上,如果你还不会安装,请参考:prometheus多维度监控容器 如果你还没有安装库k8s集群,情参考: 从零开始搭建基于calico的kuben ...
- Prometheus 监控K8S集群资源监控
Prometheus 监控K8S集群中Pod 目前cAdvisor集成到了kubelet组件内,可以在kubernetes集群中每个启动了kubelet的节点使用cAdvisor提供的metrics接 ...
- 如何用prometheus监控k8s集群中业务pod的metrics
一般,我们从网上看到的帖子和资料, 都是用prometheus监控k8s的各项资源, 如api server, namespace, pod, node等. 那如果是自己的业务pod上的自定义metr ...
- 如何使用helm优雅安装prometheus-operator，并监控k8s集群微服务
前言:随着云原生概念盛行,对于容器.服务.节点以及集群的监控变得越来越重要.Prometheus 作为 Kubernetes 监控的事实标准,有着强大的功能和良好的生态.但是它不支持分布式,不支持数据 ...
- Prometheus 监控 Redis 集群的正确姿势
Prometheus 监控Redis的正确姿势(redis集群) Prometheus 监控 Redis cluster,其实套路都是一样的,使用 exporter. exporter 负责采集指标, ...
- shell脚本监控k8s集群job状态，若出现error通过触发阿里云的进程监控报警
#!/bin/bash while [ 1 ] do job_error_no=`kubectl get pod -n weifeng |grep -i "job"|grep -c ...
- Kubernetes 学习3 kubeadm初始化k8s集群
一.k8s集群 1.k8s整体架构图 2.k8s网络架构图 二.基于kubeadm安装k8s步骤 1.master,nodes:安装kubelet,kubeadm,docker 2.master: k ...
- Prometheus神器之监控K8s集群
Prometheus 简介 Prometheus是SoundCloud开源的一款开源软件.它的实现参考了Google内部的监控实现,与源自Google的Kubernetes结合起来非常合适.另外相比i ...
- Prometheus监控elasticsearch集群（以elasticsearch-6.4.2版本为例）
部署elasticsearch集群,配置文件可"浓缩"为以下: cluster.name: es_cluster node.name: node1 path.data: /app/ ...
随机推荐
- Python文本文件的输入输出操作学习
Python具有基本的文本文件读写功能.Python的标准库提供有更丰富的读写功能. 文本文件的读写主要通过open()所构建的文件对象来实现. 创建文件对象 我们打开一个文件,并使用一个对象来表示该 ...
- JVM新生代老年代详解
1.为什么会有年轻代 我们先来屡屡,为什么需要把堆分代?不分代不能完成他所做的事情么?其实不分代完全可以,分代的唯一理由就是优化GC性能.你先想想,如果没有分代,那我们所有的对象都在一块,GC的时候我 ...
- 互评Beta版本——Thunder组爱阅app（探路者团队测评)
基于NABCD评论作品,及改进建议 每个小组评论其他小组beta发布的作品. 1.根据(不限于)NABCD评论作品的选题; N(Need,需求):在Beta中加入了书友QQ群,以及反馈建议,更好的 ...
- oracle和mysql对时间与字符串的转换
1,oracle to_date(#{item.value},'YYYY-MM-DD hh24-mi-ss') to_char(CRERATE_TIME,'YYYY-MM-DD hh24-mi-ss' ...
- 【贪心算法】POJ-3190 区间问题
一.题目 Description Oh those picky N (1 <= N <= 50,000) cows! They are so picky that each one wil ...
- 『编程题全队』Alpha 阶段冲刺博客Day3
1.每日站立式会议 1.会议照片 2.昨天已完成的工作统计 孙志威: 1.添加团队模块的标题栏 2.测试客户端和服务器之间的通讯基本连通性 3.完成团队模块的燃尽图模块 孙慧君: 1.完成了水印的设计 ...
- 第二次结对作业-WordCount进阶需求
原博客 队友博客 github项目地址 目录 具体分工 需求分析 PSP表格 解题思路描述与设计实现说明 爬虫使用 代码组织与内部实现设计(类图) 算法的关键与关键实现部分流程图 附加题设计与展示 设 ...
- windows多线程（六） 互斥量Mutex与关键段CriticalSection比较
一.关键段CS 和 互斥量Mutex 的相同点:都有线程拥有权 关键段和互斥量都有线程拥有权,即可以被一个线程拥有.在 前面讲关键段CS的文章中有说到,关键段结构体的第四个参数保存着拥有该关键段的线程 ...
- [转帖] IIS 与 HTTP/2 的介绍.
HTTP/2 on IIS https://blogs.iis.net/davidso/http2 Friday, September 11, 2015 Windows 10 HTTP2 In Oct ...
- js获取元素到屏幕左上角的距离
开发过程中经常会遇到 获取元素到屏幕左上角的距离, 当我们使用jQuery开发时,我们可以使用 $.offset()来获取准确的距离. 如果我们的项目中并没有引入jQuer的话,跟希望通过原生方法实现 ...