一、HPA概述

1.1、HPAjieshao 

  • HPA全称Horizontal Pod Autoscaler,即水平Pod自动伸缩器,可以根据观察到的CPU、内存使用率或自定义度量标准来自动增加或者减少Pod的数量,但是HPA不适用于无法扩、缩容的对象,例如DaemonSet,通常都作用与Deployment
  • HPA控制器会定期调整RC或者Deployment的副本数,使对象数量符合用户定义规则的数量
  • 既然是通过CPU、内存等指标来自动扩、缩容Pod,那么HPA肯定是需要一个能监控这些硬件资源的组件,则例的组件有很多选择,例如metrices-server、Heapster等,这里使用metrices-server

metrices-server从api-server中获取cpu、内存使用率等监控指标

1.2、HPA伸缩过程

1、收集HPA控制下所有Pod最近的cpu使用情况(CPU utilization)
2、对比在扩容条件里记录的cpu限额(CPUUtilization)
3、调整实例数(必须要满足不超过最大/最小实例数)
4、每隔30s做一次自动扩容的判断

CPU utilization的计算方法是用cpu usage(最近一分钟的平均值,通过metrics可以直接获取到)除以cpu request(这里cpu request就是我们在创建容器时制定的cpu使用核心数)得到一个平均值,这个平均值可以理解为:平均每个Pod CPU核心的使用占比。

1.3、HPA 的工作原理

k8s中的某个Metrics Server(Heapster或自定义Metrics Server)持续采集所有Pod副本的指标数据。

HPA控制器通过Metrics Server的API(Heapster的API或聚合API)获取这些数据,基于用户定义的扩缩容规则进行计算,得到目标Pod副本数量。

当目标Pod副本数量与当前副本数量不同时,HPA控制器就访问Pod的副本控制器(Deployment 、RC或者ReplicaSet)发起scale操作,调整Pod的副本数量,完成扩缩容操作。

总结:HPA 通过监控分析一些控制器控制的所有 Pod 的负载变化情况来确定是否需要调整 Pod 的副本数量。

1.4、指标的类型

Master的kube-controller-manager服务持续监测目标Pod的某种性能指标,以计算是否需要调整副本数量。目前k8s支持的指标类型如下。

◎ Pod资源使用率:Pod级别的性能指标,通常是一个比率值,例如CPU使用率。 
◎ Pod自定义指标:Pod级别的性能指标,通常是一个数值,例如接收的请求数量。 
◎ Object自定义指标或外部自定义指标:通常是一个数值,需要容器应用以某种方式提供,例如通过HTTP URL“/metrics”提供,或者使用外部服务提供的指标采集URL。

k8s从1.11版本开始,弃用基于Heapster组件完成Pod的CPU使用率采集的机制,全面转向基于Metrics Server完成数据采集。Metrics Server将采集到的Pod性能指标数据通过聚合API(Aggregated API)如metrics.k8s.io、custom.metrics.k8s.io和external.metrics.k8s.io提供给HPA控制器进行查询。关于聚合API和聚合器(API Aggregator)的概念我们后面详细讲解。

1.5、扩缩容策略

通过 伸缩系数 判断是否要进行扩容或缩容。

HPA会根据获得的指标数值,应用相应的算法算出一个伸缩系数,此系数是指标的期望值与目前值的比值,如果大于1表示扩容,小于1表示缩容。

容忍度
--horizontal-pod-autoscaler-tolerance:容忍度

它允许一定范围内的使用量的不稳定,现在默认为0.1,这也是出于维护系统稳定性的考虑。

例如,设定HPA调度策略为cpu使用率高于50%触发扩容,那么只有当使用率大于55%或者小于45%才会触发伸缩活动,HPA会尽力把Pod的使用率控制在这个范围之间。

算法

具体的每次扩容或者缩容的多少Pod的算法为: 期望副本数 = ceil[当前副本数 * ( 当前指标 / 期望指标 )]

举个栗子

当前metric值是200m,期望值是100m,那么pod副本数将会翻一倍,因为 比率为 200.0 / 100.0 = 2.0;
如果当前值是 50m ,我们会将pod副本数减半,因为 50.0 / 100.0 == 0.5
如果比率接近1.0,如0.9或1.1(即容忍度是0.1),将不会进行缩放(取决于内置的全局参数容忍度,–horizontal-pod-autoscaler-tolerance,默认值为0.1)。

此外,存在几种Pod异常的情况,如下所述。

◎  Pod正在被删除(设置了删除时间戳):将不会计入目标Pod副本数量。

◎ Pod的当前指标值无法获得:本次探测不会将这个Pod纳入目标Pod副本数量,后续的探测会被重新纳入计算范围。

◎ 如果指标类型是CPU使用率,则对于正在启动但是还未达到Ready状态的Pod,也暂时不会纳入目标副本数量范围。可以通过kube-controller-manager服务的启动参数--horizontal-pod-autoscaler-initial-readiness-delay设置首次探测Pod是否Ready的延时时间,默认值为30s。另一个启动参数--horizontal-pod-autoscaler-cpuinitialization-period设置首次采集Pod的CPU使用率的延时时间。

注意:

  • 每次最大扩容pod数量不会超过当前副本数量的2倍
  • 如果某些pod的容器没有需要的的资源metrics,自动伸缩将不会根据这些metrics进行伸缩。
  • 如果指定了targetAverageValue 或者 targetAverageUtilization,currentMetricValue是所有目标pod的metric取均值。在检查容忍度和确定最终值之前,会结合参考pod是否就绪、是否丢失metrics。
  • 设置了删除时间戳的所有Pod(即处于关闭状态的Pod)和所有失败的Pod将被丢弃。

冷却和延迟机制

使用HPA管理一组副本时,有可能因为metrics动态变化而导致副本数频繁波动,这种现象叫做 “颠簸”。

想象一种场景:

当pod所需要的CPU负荷过大,从而在创建一个新pod的过程中,系统的CPU使用量可能会同样在有一个攀升的过程。所以,在每一次作出决策后的一段时间内,将不再进行扩展决策。对于扩容而言,这个时间段为3分钟,缩容为5分钟

  • -- horizontal-pod-autoscaler-downscale- delay :这个参数用于告诉autoscaler做完缩容操作后需要等多久才能进行下一次缩容,默认值是5分钟。
  • --horizontal-pod-autoscaler-upscale-delay: 这个参数用于告诉autoscaler做完扩容操作后需要等多久才能进行下一次扩容,默认值是3分钟。
  • 注意:使用者需要知道调整这个参数可能造成的影响。设置得太长,HPA对负载变化的响应也会变长;太短又会导致自动伸缩“颠簸”。

Pod延迟探测机制

如果指标类型是CPU使用率,则对于正在启动但是还未达到Ready状态的Pod,也暂时不会纳入目标副本数量范围。

  • 可以通过kube-controller-manager服务的启动参数--horizontal-pod-autoscaler-initial-readiness-delay设置首次探测Pod是否Ready的延时时间,默认值为30s。
  • 另一个启动参数--horizontal-pod-autoscaler-cpuinitialization-period设置首次采集Pod的CPU使用率的延时时间。

1.6、HPA获取自定义指标(Custom Metrics)的底层实现(基于Prometheus)

Kubernetes是借助Agrregator APIServer扩展机制来实现Custom Metrics。Custom Metrics APIServer是一个提供查询Metrics指标的API服务(Prometheus的一个适配器),这个服务启动后,kubernetes会暴露一个叫custom.metrics.k8s.io的API,当请求这个URL时,请求通过Custom Metics APIServer去Prometheus里面去查询对应的指标,然后将查询结果按照特定格式返回。

二、HPA的metrics的分类

要支持最新的custom(包括external)的metrics,也需要使用新版本的HPA:autoscaling/v2beta1,里面增加四种类型的Metrics:Resource、Pods、Object、External,每种资源对应不同的场景,下面分别说明:

  • Resource支持k8s里Pod的所有系统资源(包括cpu、memory等),但是一般只会用cpu,memory因为不太敏感而且跟语言相关:大多数语言都有内存池及内置GC机制导致进程内存监控不准确。
  • Pods类型的metrics表示cpu,memory等系统资源之外且是由Pod自身提供的自定义metrics数据,比如用户可以在web服务的pod里提供一个promesheus metrics的自定义接口,里面暴露了本pod的实时QPS监控指标,这种情况下就应该在HPA里直接使用Pods类型的metrics。
  • Object类型的metrics表示监控指标不是由Pod本身的服务提供,但是可以通过k8s的其他资源Object提供metrics查询,比如ingress等,一般Object是需要汇聚关联的Deployment下的所有的pods总的指标。
  • External类型的metrics也属于自定义指标,与Pods和Object不同的是,其监控指标的来源跟k8s本身无关,metrics的数据完全取自外部的系统。

在HPA最新的版本 autoscaling/v2beta2 中又对metrics的配置和HPA扩缩容的策略做了完善,特别是对 metrics 数据的目标指标值的类型定义更通用灵活:包括AverageUtilization、AverageValue和Value,但是不是所有的类型的Metrics都支持三种目标值的,具体对应关系如下表。

HPA里的各种类型的Metrics和Metrics Target Type的对应支持关系表

Metrics Type \ Target TypeAverageUtilizationAverageValueValue备注(query metrics)
Resource(pod’s cpu/memory etc.)YesYesNopods metrics list
Pods(pod’s other metrics)NoYesNopods metrics list
Object(k8s object)NoYesYesobject metrics
External(not k8s object)NoYesYesexternal metrics list

三、HPA的使用说明

先看个最简单的HPA的定义的例子

apiVersion: autoscaling/v2beta2
kind: HorizontalPodAutoscaler
metadata:
  name: php-apache
spec:
  scaleTargetRef:
    apiVersion: apps/v1
    kind: Deployment
    name: php-apache
  minReplicas: 1
  maxReplicas: 10
  metrics:
  - type: Resource
    resource:
      name: cpu
      target:
        type: Utilization
        averageUtilization: 50

从上面的例子可以看出,HPA的spec定义由三个必填部分组成:

  1. HPA控制的目标workload,即scaleTargetRef,理论上HPA可以对任意支持scale子接口( sub-resource )的workload做弹性伸缩,不过statefulset一般代表有状态服务,副本不可随便修改,而Job一般代表短生命周期的,所以基本可以认为HPA目前是专门控制deployment的扩缩容的(不建议直接控制RS,否则将无法滚动升级)。
  2. 弹性扩缩容的上下边界,minReplicas和maxReplicas,也就是说HPA的扩缩容也不能是漫无边际,如果计算出的副本数超过max则统一取maxReplicas,maxReplicas是为了保护k8s集群的资源被耗尽,minReplicas则相反,而且minReplicas必须不大于maxReplicas,但是也要大于0(k8s v1.16之后才放开允许Objetct和External类型的metrics的minReplicas为0,需要apiserver开启–feature-gates mapStringBool HPAScaleToZero=true),两者相等就相当于关闭了自动伸缩功能了,总的来说minReplicas和maxReplicas边界机制避免metrics数据异常导致的副本数不受控,特别是HPA在k8s最新的v1.18版本也依然是alpha特性,强烈建议大家谨慎设置这两个边界。
  3. metrics指标类型和目标值,在autoscaling/v1里只有targetCPUUtilizationPercentage,autoscaling/v2beta1开始就扩展为metrics数组了,也就是说一个HPA可以同时设置多个类型维度的metrics目标指标,如果有多个HPA 将会依次考量各个指标,然后最终HPA Controller选择一个会选择扩缩幅度最大的那个为最终扩容副本数。在最新的autoscaling/v2beta2版本的HPA中,metrics type共有4种类型:Resource、Pods、Object、External,target里则定义了metrics的目标期望值,这里target的type也有三种类型Utilization,AverageValue和 Value,不同的metrics type都只能支持部分target type(详见上面表格)
  4. 此外,在autoscaling/v2beta2的HPA的spec里还新增了一个Behavior可选结构,它是用来精确控制HPA的扩容和缩容的速度。

完整的HPA的定义可参考k8s的官方API文档

默认HPA spec里不配置任何metrics的话k8s会默认设置cpu的Resouce,且目标类型是AverageUtilization value为80%。

四、HPA部署

4.1、HPA版本

查看HPA所有版本

[root@master C]# kubectl api-versions |grep autoscaling
autoscaling/v1			#只支持通过CPU为参考依据来改变Pod的副本数
autoscaling/v2beta1		#支持通过CPU、内存、连接数或者自定义规则为参考依据
autoscaling/v2beta2		#和v2beta1差不多

查看当前版本

[root@master C]# kubectl explain hpa
KIND:     HorizontalPodAutoscaler
VERSION:  autoscaling/v1  #可以看到使用的默认版本是v1

DESCRIPTION:
     configuration of a horizontal pod autoscaler.

FIELDS:
   apiVersion   <string>
     APIVersion defines the versioned schema of this representation of an
     object. Servers should convert recognized schemas to the latest internal
     value, and may reject unrecognized values. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources

   kind <string>
     Kind is a string value representing the REST resource this object
     represents. Servers may infer this from the endpoint the client submits
     requests to. Cannot be updated. In CamelCase. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds

   metadata     <Object>
     Standard object metadata. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadata

   spec <Object>
     behaviour of autoscaler. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status.

   status       <Object>
     current information about the autoscaler.

指定使用版本,这里并不是修改,相当于执行这条命令时,指定了下版本

[root@master C]# kubectl explain hpa --api-version=autoscaling/v2beta1
KIND:     HorizontalPodAutoscaler
VERSION:  autoscaling/v2beta1

DESCRIPTION:
     HorizontalPodAutoscaler is the configuration for a horizontal pod
     autoscaler, which automatically manages the replica count of any resource
     implementing the scale subresource based on the metrics specified.

FIELDS:
   apiVersion   <string>
     APIVersion defines the versioned schema of this representation of an
     object. Servers should convert recognized schemas to the latest internal
     value, and may reject unrecognized values. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#resources

   kind <string>
     Kind is a string value representing the REST resource this object
     represents. Servers may infer this from the endpoint the client submits
     requests to. Cannot be updated. In CamelCase. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#types-kinds

   metadata     <Object>
     metadata is the standard object metadata. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#metadata

   spec <Object>
     spec is the specification for the behaviour of the autoscaler. More info:
     https://git.k8s.io/community/contributors/devel/sig-architecture/api-conventions.md#spec-and-status.

   status       <Object>
     status is the current information about the autoscaler.

4.2、 HPA部署

(1)部署metrics-server

[root@master kube-system]# kubectl top nodes  #查看节点状态,因为没有安装,所以会报错
Error from server (NotFound): the server could not find the requested resource (get services http:heapster:)
  • 编写yaml文件,注意端口和镜像
[root@master kube-system]# vim components-v0.5.0.yaml
apiVersion: v1
kind: ServiceAccount
metadata:
  labels:
    k8s-app: metrics-server
  name: metrics-server
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  labels:
    k8s-app: metrics-server
    rbac.authorization.k8s.io/aggregate-to-admin: "true"
    rbac.authorization.k8s.io/aggregate-to-edit: "true"
    rbac.authorization.k8s.io/aggregate-to-view: "true"
  name: system:aggregated-metrics-reader
rules:
- apiGroups:
  - metrics.k8s.io
  resources:
  - pods
  - nodes
  verbs:
  - get
  - list
  - watch
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  labels:
    k8s-app: metrics-server
  name: system:metrics-server
rules:
- apiGroups:
  - ""
  resources:
  - pods
  - nodes
  - nodes/stats
  - namespaces
  - configmaps
  verbs:
  - get
  - list
  - watch
---
apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:
  labels:
    k8s-app: metrics-server
  name: metrics-server-auth-reader
  namespace: kube-system
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: Role
  name: extension-apiserver-authentication-reader
subjects:
- kind: ServiceAccount
  name: metrics-server
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  labels:
    k8s-app: metrics-server
  name: metrics-server:system:auth-delegator
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: system:auth-delegator
subjects:
- kind: ServiceAccount
  name: metrics-server
  namespace: kube-system
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  labels:
    k8s-app: metrics-server
  name: system:metrics-server
roleRef:
  apiGroup: rbac.authorization.k8s.io
  kind: ClusterRole
  name: system:metrics-server
subjects:
- kind: ServiceAccount
  name: metrics-server
  namespace: kube-system
---
apiVersion: v1
kind: Service
metadata:
  labels:
    k8s-app: metrics-server
  name: metrics-server
  namespace: kube-system
spec:
  ports:
  - name: https
    port: 443
    protocol: TCP
    targetPort: https
  selector:
    k8s-app: metrics-server
---
apiVersion: apps/v1
kind: Deployment
metadata:
  labels:
    k8s-app: metrics-server
  name: metrics-server
  namespace: kube-system
spec:
  selector:
    matchLabels:
      k8s-app: metrics-server
  strategy:
    rollingUpdate:
      maxUnavailable: 0
  template:
    metadata:
      labels:
        k8s-app: metrics-server
    spec:
      containers:
      - args:
        - --cert-dir=/tmp
        - --secure-port=4443
        - --kubelet-preferred-address-types=InternalIP,ExternalIP,Hostname
        - --kubelet-use-node-status-port
        - --metric-resolution=15s
        - --kubelet-insecure-tls
        image: registry.cn-shenzhen.aliyuncs.com/zengfengjin/metrics-server:v0.5.0
        imagePullPolicy: IfNotPresent
        livenessProbe:
          failureThreshold: 3
          httpGet:
            path: /livez
            port: https
            scheme: HTTPS
          periodSeconds: 10
        name: metrics-server
        ports:
        - containerPort: 4443
          name: https
          protocol: TCP
        readinessProbe:
          failureThreshold: 3
          httpGet:
            path: /readyz
            port: https
            scheme: HTTPS
          initialDelaySeconds: 20
          periodSeconds: 10
        resources:
          requests:
            cpu: 100m
            memory: 200Mi
        securityContext:
          readOnlyRootFilesystem: true
          runAsNonRoot: true
          runAsUser: 1000
        volumeMounts:
        - mountPath: /tmp
          name: tmp-dir
      nodeSelector:
        kubernetes.io/os: linux
      priorityClassName: system-cluster-critical
      serviceAccountName: metrics-server
      volumes:
      - emptyDir: {}
        name: tmp-dir
---
apiVersion: apiregistration.k8s.io/v1
kind: APIService
metadata:
  labels:
    k8s-app: metrics-server
  name: v1beta1.metrics.k8s.io
spec:
  group: metrics.k8s.io
  groupPriorityMinimum: 100
  insecureSkipTLSVerify: true
  service:
    name: metrics-server
    namespace: kube-system
  version: v1beta1
  versionPriority: 100
  • 部署
[root@master kube-system]# kubectl apply -f components-v0.5.0.yaml
serviceaccount/metrics-server created
clusterrole.rbac.authorization.k8s.io/system:aggregated-metrics-reader created
clusterrole.rbac.authorization.k8s.io/system:metrics-server created
rolebinding.rbac.authorization.k8s.io/metrics-server-auth-reader created
clusterrolebinding.rbac.authorization.k8s.io/metrics-server:system:auth-delegator created
clusterrolebinding.rbac.authorization.k8s.io/system:metrics-server created
service/metrics-server created
deployment.apps/metrics-server created
apiservice.apiregistration.k8s.io/v1beta1.metrics.k8s.io created

#查看创建的pod
[root@master kube-system]# kubectl get pods -n kube-system| egrep 'NAME|metrics-server'
NAME                              READY   STATUS              RESTARTS   AGE
metrics-server-5944675dfb-q6cdd   0/1     ContainerCreating   0          6s

 #查看日志
[root@master kube-system]# kubectl logs metrics-server-5944675dfb-q6cdd  -n kube-system 
I0718 03:06:39.064633       1 serving.go:341] Generated self-signed cert (/tmp/apiserver.crt, /tmp/apiserver.key)
I0718 03:06:39.870097       1 configmap_cafile_content.go:202] Starting client-ca::kube-system::extension-apiserver-authentication::client-ca-file
I0718 03:06:39.870122       1 configmap_cafile_content.go:202] Starting client-ca::kube-system::extension-apiserver-authentication::requestheader-client-ca-file
I0718 03:06:39.870159       1 shared_informer.go:240] Waiting for caches to sync for client-ca::kube-system::extension-apiserver-authentication::client-ca-file
I0718 03:06:39.870160       1 shared_informer.go:240] Waiting for caches to sync for client-ca::kube-system::extension-apiserver-authentication::requestheader-client-ca-file
I0718 03:06:39.870105       1 requestheader_controller.go:169] Starting RequestHeaderAuthRequestController
I0718 03:06:39.871166       1 shared_informer.go:240] Waiting for caches to sync for RequestHeaderAuthRequestController
I0718 03:06:39.872804       1 dynamic_serving_content.go:130] Starting serving-cert::/tmp/apiserver.crt::/tmp/apiserver.key
I0718 03:06:39.875741       1 secure_serving.go:197] Serving securely on [::]:4443
I0718 03:06:39.876050       1 tlsconfig.go:240] Starting DynamicServingCertificateController
I0718 03:06:39.970469       1 shared_informer.go:247] Caches are synced for client-ca::kube-system::extension-apiserver-authentication::client-ca-file
I0718 03:06:39.970575       1 shared_informer.go:247] Caches are synced for client-ca::kube-system::extension-apiserver-authentication::requestheader-client-ca-file
I0718 03:06:39.971610       1 shared_informer.go:247] Caches are synced for RequestHeaderAuthRequestController

#如果报错的化,可以修改apiserver的yaml文件,这是k8s的yaml文件
[root@master kube-system]# vim /etc/kubernetes/manifests/kube-apiserver.yaml

 40     - --tls-cert-file=/etc/kubernetes/pki/apiserver.crt
 41     - --tls-private-key-file=/etc/kubernetes/pki/apiserver.key
 42     - --enable-aggregator-routing=true     #添加这行
 43     image: registry.aliyuncs.com/google_containers/kube-apiserver:v1.18.0
 44     imagePullPolicy: IfNotPresent

#保存退出
[root@master kube-system]# systemctl restart kubelet  #修改后重启kubelet

#再次查看节点信息

[root@master kube-system]# kubectl top node
NAME     CPU(cores)   CPU%   MEMORY(bytes)   MEMORY%
master   327m         4%     3909Mi          23%
node     148m         1%     1327Mi          8%

(2)创建Deployment

  • 这里创建一个nginx的deployment
[root@master test]# cat nginx.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx
spec:
  selector:
    matchLabels:
      run: nginx
  replicas: 1
  template:
    metadata:
      labels:
        run: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.15.2
        ports:
        - containerPort: 80
        resources:
          limits:
            cpu: 500m
          requests:  #想要HPA生效,必须添加requests声明
            cpu: 200m

---
apiVersion: v1
kind: Service
metadata:
  name: nginx
  labels:
    run: nginx
spec:
  ports:
  - port: 80
  selector:
    run: nginx
  • 访问测试
[root@master test]# kubectl get pods -o wide
NAME                    READY   STATUS    RESTARTS   AGE   IP            NODE   NOMINATED NODE   READINESS GATES
nginx-9cb8d65b5-tq9v4   1/1     Running   0          14m   10.244.1.22   node   <none>           <none>
[root@master test]# kubectl get svc nginx
NAME    TYPE        CLUSTER-IP      EXTERNAL-IP   PORT(S)   AGE
nginx   ClusterIP   172.16.169.27   <none>        80/TCP    15m
[root@master test]# kubectl describe svc nginx
Name:              nginx
Namespace:         default
Labels:            run=nginx
Annotations:       Selector:  run=nginx
Type:              ClusterIP
IP:                172.16.169.27
Port:              <unset>  80/TCP
TargetPort:        80/TCP
Endpoints:         10.244.1.22:80
Session Affinity:  None
Events:            <none>
[root@node test]# curl 172.16.169.27  #访问成功
<!DOCTYPE html>
<html>
<head>
<title>Welcome to nginx!</title>
<style>
    body {
        width: 35em;
        margin: 0 auto;
        font-family: Tahoma, Verdana, Arial, sans-serif;
    }
</style>
</head>
<body>
<h1>Welcome to nginx!</h1>
<p>If you see this page, the nginx web server is successfully installed and
working. Further configuration is required.</p>

<p>For online documentation and support please refer to
<a href="http://nginx.org/">nginx.org</a>.<br/>
Commercial support is available at
<a href="http://nginx.com/">nginx.com</a>.</p>

<p><em>Thank you for using nginx.</em></p>
</body>
</html>

(3)基于CPU创建HPA

#创建一个cpu利用率达到20,最大10个pod,最小1个,这里没有指定版本所以默认是v1版本,而v1版本只能以CPU为标准
[root@master test]# kubectl autoscale deployment nginx --cpu-percent=20 --min=1 --max=10
horizontalpodautoscaler.autoscaling/nginx autoscaled

#TARGETS可以看到使用率
[root@master test]# kubectl get hpa
NAME    REFERENCE          TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
nginx   Deployment/nginx   0%/20%    1         10        1          86s

#创建一个测试pod增加负载,访问地址要和pod的svc地址相同
[root@master ~]# kubectl  run busybox -it --image=busybox -- /bin/sh -c 'while true; do wget -q -O- http://10.244.1.22; done'


#过一分钟后看hap的使用率,REPLICAS是当前pod的数量
[root@master test]# kubectl get hpa
NAME    REFERENCE          TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
nginx   Deployment/nginx   27%/20%   1         10        5          54m

[root@master test]# kubectl get pods   #再看pod数量,发现已经增加到了5个
NAME                    READY   STATUS    RESTARTS   AGE
bustbox                 1/1     Running   0          119s
nginx-9cb8d65b5-24dg2   1/1     Running   0          57s
nginx-9cb8d65b5-c6n98   1/1     Running   0          87s
nginx-9cb8d65b5-ksjzv   1/1     Running   0          57s
nginx-9cb8d65b5-n77fm   1/1     Running   0          87s
nginx-9cb8d65b5-tq9v4   1/1     Running   0          84m
[root@master test]# kubectl get deployments.apps
NAME    READY   UP-TO-DATE   AVAILABLE   AGE
nginx   5/5     5            5           84m


#此时,停止压测,过好几分钟后再次查看pod数量和使用率
[root@master test]# kubectl delete pod busybox  #终止后,删除pod
[root@master test]# kubectl get hpa  #虽然使用率已经降到0了,但是可以看到当前REPLICAS的数量还5,这个需要等一会就会缩容
NAME    REFERENCE          TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
nginx   Deployment/nginx   0%/20%    1         10        5          58m

#过了几分钟后,可以看到pod数量已经回到了1
[root@master test]# kubectl get hpa
NAME    REFERENCE          TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
nginx   Deployment/nginx   0%/20%    1         10        1          64m  
[root@master test]# kubectl get pods
NAME                    READY   STATUS    RESTARTS   AGE
nginx-9cb8d65b5-tq9v4   1/1     Running   0          95m

(4)基于内存创建的HPA

#先把上面创建的资源删除
[root@master test]# kubectl delete horizontalpodautoscalers.autoscaling  nginx
horizontalpodautoscaler.autoscaling "nginx" deleted
[root@master test]# kubectl delete -f nginx.yaml
deployment.apps "nginx" deleted
service "nginx" deleted
  • 重新编写yaml文件
[root@master test]# cat nginx.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx
spec:
  selector:
    matchLabels:
      run: nginx
  replicas: 1
  template:
    metadata:
      labels:
        run: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.15.2
        ports:
        - containerPort: 80
        resources:
          limits:
            cpu: 500m
            memory: 60Mi
          requests:
            cpu: 200m
            memory: 25Mi

---
apiVersion: v1
kind: Service
metadata:
  name: nginx
  labels:
    run: nginx
spec:
  ports:
  - port: 80
  selector:
    run: nginx

[root@master test]# kubectl apply -f nginx.yaml
deployment.apps/nginx created
service/nginx created
  • 创建HPA
[root@master test]# vim hpa-nginx.yaml
apiVersion: autoscaling/v2beta1  #上面的hpa版本有提到过,使用基于内存的hpa需要换个版本
kind: HorizontalPodAutoscaler
metadata:
  name: nginx-hpa
spec:
  maxReplicas: 10  #1-10的pod数量限制
  minReplicas: 1
  scaleTargetRef:               #指定使用hpa的资源对象,版本、类型、名称要和上面创建的相同
    apiVersion: apps/v1
    kind: Deployment
    name: nginx
  metrics:
  - type: Resource
    resource:
      name: memory
      targetAverageUtilization: 50   #限制%50的内存
[root@master test]# kubectl apply -f hpa-nginx.yaml
horizontalpodautoscaler.autoscaling/nginx-hpa created
[root@master test]# kubectl get hpa
NAME        REFERENCE          TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
nginx-hpa   Deployment/nginx   7%/50%    1         10        1          59s
  • 更换终端进行测试
#在pod中执行命令,增加内存负载
[root@master ~]# kubectl exec -it nginx-78f4944bb8-2rz7j -- /bin/sh -c 'dd if=/dev/zero of=/tmp/file1'
  • 等待负载上去,然后查看pod数量与内存使用率
[root@master test]# kubectl get hpa
NAME        REFERENCE          TARGETS    MINPODS   MAXPODS   REPLICAS   AGE
nginx-hpa   Deployment/nginx   137%/50%   1         10        1          12m
[root@master test]# kubectl get hpa
NAME        REFERENCE          TARGETS   MINPODS   MAXPODS   REPLICAS   AGE
nginx-hpa   Deployment/nginx   14%/50%   1         10        3          12m
[root@master test]# kubectl get pods
NAME                     READY   STATUS    RESTARTS   AGE
nginx-78f4944bb8-2rz7j   1/1     Running   0          21m
nginx-78f4944bb8-bxh78   1/1     Running   0          34s
nginx-78f4944bb8-g8w2h   1/1     Running   0          34s
#与CPU相同,内存上去了也会自动创建pod

Logo

开源、云原生的融合云平台

更多推荐