【原文链接】
一、HPA控制器简介
HPA（Horizontal Pod Autoscaler）控制器可以获取pod利用率，然后和HPA中定义的指标进行对比，同时计算出需要伸缩的具体值，最后实现pod的数量的调整，其实HPA与之前的Deployment一样，也属于一种Kubernetes对象，它通过追踪分析目标pod的负载变化情况，类确定是否需要针对性地调整目标pod的副本数
二、HPA控制器环境准备
2.1 安装metric-server
（1）参照 Git----安装(CentOS) 首先安装git工具
（2）下载metric-server
git clone -b v0.3.6 https://github.com/kubernetes-incubator/metrics-server（3）配置yaml文件
cd metrics-server/deploy/1.8+/
vi metrics-server-deployment.yaml然后按照下图的位置编辑文件，内容如下：
hostNetwork: true
image: registry.cn-hangzhou.aliyuncs.com/google_containers/metrics-server-amd64:v0.3.6
- --kubelet-insecure-tls
- --kubelet-preferred-address-types=InternalIP,Hostname,InternalNDS,ExternalDNS,ExternalIP
（4）使用如下命令部署
[root@master 1.8+]# kubectl apply -f ./
clusterrole.rbac.authorization.k8s.io/system:aggregated-metrics-reader created
Warning: rbac.authorization.k8s.io/v1beta1 ClusterRoleBinding is deprecated in v1.17+, unavailable in v1.22+; use rbac.authorization.k8s.io/v1 ClusterRoleBinding
clusterrolebinding.rbac.authorization.k8s.io/metrics-server:system:auth-delegator created
Warning: rbac.authorization.k8s.io/v1beta1 RoleBinding is deprecated in v1.17+, unavailable in v1.22+; use rbac.authorization.k8s.io/v1 RoleBinding
rolebinding.rbac.authorization.k8s.io/metrics-server-auth-reader created
Warning: apiregistration.k8s.io/v1beta1 APIService is deprecated in v1.19+, unavailable in v1.22+; use apiregistration.k8s.io/v1 APIService
apiservice.apiregistration.k8s.io/v1beta1.metrics.k8s.io created
serviceaccount/metrics-server created
deployment.apps/metrics-server created
service/metrics-server created
clusterrole.rbac.authorization.k8s.io/system:metrics-server created
clusterrolebinding.rbac.authorization.k8s.io/system:metrics-server created
[root@master 1.8+]#查询如下，有metrics-server-669dfc56ff-v6drv 即表示已经部署成功了
[root@master 1.8+]# kubectl get pod -n kube-system
NAME READY STATUS RESTARTS AGE
coredns-558bd4d5db-7vbmq 1/1 Running 0 12d
coredns-558bd4d5db-sps22 1/1 Running 0 12d
etcd-master 1/1 Running 0 12d
kube-apiserver-master 1/1 Running 0 12d
kube-controller-manager-master 1/1 Running 0 12d
kube-flannel-ds-cd9qk 1/1 Running 0 12d
kube-flannel-ds-gg4jq 1/1 Running 0 12d
kube-flannel-ds-n76xj 1/1 Running 0 12d
kube-proxy-g4j5g 1/1 Running 0 12d
kube-proxy-h27ms 1/1 Running 0 12d
kube-proxy-tqzjl 1/1 Running 0 12d
kube-scheduler-master 1/1 Running 0 12d
metrics-server-669dfc56ff-v6drv 1/1 Running 0 56s
[root@master 1.8+]#（5）此时通过如下命令，可以查询到节点的资源使用情况
[root@master 1.8+]# kubectl top node
W0327 11:53:58.289701 9379 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%
master 71m 0% 1268Mi 3%
node1 24m 0% 687Mi 2%
node2 23m 0% 721Mi 2%
[root@master 1.8+]#（6）通过如下命令，可以查看到pod的资源使用情况
[root@master 1.8+]# kubectl top pod -n kube-system
W0327 11:55:35.833234 10195 top_pod.go:140] Using json format to get metrics. Next release will switch to protocol-buffers, switch early by passing --use-protocol-buffers flag
NAME CPU(cores) MEMORY(bytes)
coredns-558bd4d5db-7vbmq 1m 13Mi
coredns-558bd4d5db-sps22 1m 13Mi
etcd-master 9m 26Mi
kube-apiserver-master 22m 297Mi
kube-controller-manager-master 6m 53Mi
kube-flannel-ds-cd9qk 2m 15Mi
kube-flannel-ds-gg4jq 2m 15Mi
kube-flannel-ds-n76xj 1m 16Mi
kube-proxy-g4j5g 1m 19Mi
kube-proxy-h27ms 1m 19Mi
kube-proxy-tqzjl 1m 19Mi
kube-scheduler-master 2m 21Mi
metrics-server-669dfc56ff-v6drv 1m 15Mi
[root@master 1.8+]#2.2 创建deployment和service
编辑service_deployment.yaml文件，内容如下：
apiVersion: v1
kind: Namespace
metadata:
name: dev
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: deploy-nginx
namespace: dev
spec:
replicas: 1
selector:
matchLabels:
run: nginx
template:
metadata:
labels:
run: nginx
spec:
containers:
- name: nginx
image: nginx:1.17.1
ports:
- containerPort: 80
protocol: TCP
---
apiVersion: v1
kind: Service
metadata:
name: service-nginx
namespace: dev
spec:
ports:
- port: 80
protocol: TCP
targetPort: 80
nodePort: 30030
selector:
run: nginx
type: NodePort使用如下命令创建
[root@master pod_controller]# kubectl apply -f service_deployment.yaml
namespace/dev created
deployment.apps/deploy-nginx created
service/service-nginx created
[root@master pod_controller]#查看创建的资源如下：
[root@master pod_controller]# kubectl get deployment,pod,service -n dev
NAME READY UP-TO-DATE AVAILABLE AGE
deployment.apps/deploy-nginx 3/3 3 3 5m6s
NAME READY STATUS RESTARTS AGE
pod/deploy-nginx-66ffc897cf-jhc9w 1/1 Running 0 5m6s
NAME TYPE CLUSTER-IP EXTERNAL-IP PORT(S) AGE
service/service-nginx NodePort 10.106.128.172 <none> 80:30030/TCP 5m6s
[root@master pod_controller]#2.3 部署HPA
编辑pc_hpa.yaml文件，内容如下：注意这里把cpu使用率设置为1%是为了测试用。在实际部署中需要根据具体情况去定
apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
name: pc-hpa
namespace: dev
spec:
minReplicas: 1
maxReplicas: 10
targetCPUUtilizationPercentage: 1
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: deploy-nginx然后使用如下命令创建
[root@master pod_controller]# kubectl apply -f pc_hpa.yaml
horizontalpodautoscaler.autoscaling/pc-hpa created
[root@master pod_controller]#查看hpa如下
[root@master pod_controller]# kubectl get hpa -n dev
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
pc-hpa Deployment/deploy-nginx <unknown>/1% 1 10 1 5m18s
[root@master pod_controller]#2.4 压力测试观测HPA控制器
开三个窗口分别对pod，deployment，hpa进行监视分别使用如下命令
kubectl get pod -n dev -w
kubectl get deploy -n dev -w
kubectl get hpa -n dev -w编写测试脚本 test.sh，用于发送get请求
#! /bin/bash
for((i=1;i<1000000;i++));
do
curl http://192.168.16.40:30030;
done执行测试脚本，然后观察pod数量变化
如下为deploy的观测结果：
[root@master ~]# kubectl get deploy -n dev -w
NAME READY UP-TO-DATE AVAILABLE AGE
deploy-nginx 1/1 1 1 82m
deploy-nginx 1/3 1 1 84m
deploy-nginx 1/3 1 1 84m
deploy-nginx 1/3 1 1 84m
deploy-nginx 1/3 3 1 84m
deploy-nginx 2/3 3 2 84m
deploy-nginx 3/3 3 3 84m如下为HPA的观测结果
[root@master ~]# kubectl get hpa -n dev -w
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
pc-hpa Deployment/deploy-nginx 0%/1% 1 10 1 63m
pc-hpa Deployment/deploy-nginx 3%/1% 1 10 1 64m
pc-hpa Deployment/deploy-nginx 3%/1% 1 10 3 64m
pc-hpa Deployment/deploy-nginx 1%/1% 1 10 3 65m
pc-hpa Deployment/deploy-nginx 1%/1% 1 10 3 66m可以看到，HPA确实可以做到动态扩缩容
