一、简介

kubernetes监控指标有核心指标和自定义指标两大类，目前使用监控指标依赖于基础组件metrics-server。

核心指标也可称为核心指标流水线，由kubelet、metrics-server和由API server提供的api组成。可以提供基础指标，如CPU累计使用率、内存实时使用率、Pod资源占用率及容器的磁盘占用率，主要由metrics-server组件提供。

自定义监控也称为监控流水线，用于从系统收集各种指标数据并提供给终端用户、存储系统以及HPA，包含了核心指标和非核心指标(非核心指标需要特殊中间控件如被kubernetes进行识别)。

二、安装部署(核心指标监控控件metrics-server)

1.安装metrics-server

GitHub官方metrics-server项目地址点击 这里 或 这里 查看，

下载yaml配置文件

[root@master1 ~]# for file in auth-delegator.yaml auth-reader.yaml metrics-apiservice.yaml metrics-server-deployment.yaml metrics-server-service.yaml resource-reader.yaml; do wget https://raw.githubusercontent.com/kubernetes/kubernetes/master/cluster/addons/metrics-server/$file; done

创建后报错，查看日志如下内容：

E1012 08:26:57.910695 1 memcache.go:134] couldn't get resource list for metrics.k8s.io/v1beta1: the server is currently unable to handle the request
E1012 08:27:13.214427 1 resource_quota_controller.go:430] unable to retrieve the complete list of server APIs: metrics.k8s.io/v1beta1: the server is currently unable to handle the request
W1012 08:27:17.126343 1 garbagecollector.go:647] failed to discover some groups: map[metrics.k8s.io/v1beta1:the server is currently unable to handle the request]

2.创建metrics-server报错处理方法

此yaml配置文件并非可以直接执行成功，笔者亲测此版本(metrics-server-v0.3.6)需要修改两个文件的某些参数，后续版本会做什么调整尚未可知，修改详细内容如下：(若需要相关镜像或文件可通过微信公众号“琼杰笔记”免费获取)

修改文件: metrics-server-deployment.yaml

修改1：

46 spec:
47 priorityClassName: system-cluster-critical
48 serviceAccountName: metrics-server
49 nodeSelector:
50 kubernetes.io/os: linux
51 hostNetwork: true  # 新增
52 containers:
53 - name: metrics-server
54 image: k8s.gcr.io/metrics-server-amd64:v0.3.6
55 imagePullPolicy: IfNotPresent
56 command:
57 - /metrics-server
58 - --metric-resolution=30s
59 - --requestheader-allowed-names=aggregator  # 新增
60 # These are needed for GKE, which doesn't support secure communication yet.
61 # Remove these lines for non-GKE clusters, and when GKE supports token-based auth.
62 # - --kubelet-port=10255  # 注释
63 # - --deprecated-kubelet-completely-insecure=true  # 注释
64 - --kubelet-insecure-tls  # 新增
65 - --kubelet-preferred-address-types=InternalIP,Hostname,InternalDNS,ExternalDNS,ExternalIP

修改2处：

修改之后，重新创建即可成功，如下：

[root@master1 metrics-server]# kubectl top nodes
NAME      CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%   
master1   226m         11%    1093Mi          63%       
node1     91m          4%     340Mi           19%       
node2     127m         6%     406Mi           23%

三、安装部署(自定义控件Prometheus)

若允许prometheus获取metrics的数据，一般需要在metrics中定义annotations参数prometheus.io/scrape: “true”

四、HorizontalPodAutoscaler(简称hpa)

1.创建pod和service资源

apiVersion: v1
kind: Service
metadata:
  name: myapp-svc
  namespace: default
spec:
  selector:
    app: myapp
    release: for1
  ports:
  - name: myapp-svc
    port: 80
    targetPort: 80
  type: ClusterIP
---
apiVersion: apps/v1
kind: Deployment
metadata:
  name: myapp-hpa
  namespace: default
spec:
  selector:
    matchLabels:
      app: myapp
      release: for1
  replicas: 1
  template:
    metadata:
      labels:
        app: myapp
        release: for1
    spec:
      containers:
      - name: myapp-hpa-test
        image: ikubernetes/myapp:v1
        imagePullPolicy: IfNotPresent
        ports:
        - name: http
          containerPort: 80
        resources:
          limits:
            cpu: 80m
            memory: 128Mi
          requests:
            cpu: 80m
            memory: 128Mi

2.创建配置hpa

[root@master1 ~]# kubectl autoscale deployment myapp-hpa --min=1 --max=6 --cpu-percent=50

3.压测

[root@node2 ~]# ab -c 1000 -n 500000000 http://192.168.222.150:32331/index.html
This is ApacheBench, Version 2.3 <$Revision: 1430300

gt; Copyright 1996 Adam Twiss, Zeus Technology Ltd, http://www.zeustech.net/ Licensed to The Apache Software Foundation, http://www.apache.org/ Benchmarking 192.168.222.150 (be patient) apr_socket_recv: Connection reset by peer (104) Total of 2663 requests completed

4.验证（此时发现pod从1个变为2个，然后等压测结束，又变为1个）

[root@master1 hpa]# kubectl describe hpa