k8s学习（30）：HPA-自动伸缩

发表于 更新于 分类于

Horizontal Pod Autoscaling(HPA)

Horizontal Pod Autoscaling可以根据CPU利用率自动伸缩一个Replication Controller、Deployment或Replica Set中的Pod数量

创建一个deployment

1
2
3
4
5
6
7
8
9
10
11
12
13
14
[root@k8s-master k8s-install]# kubectl run php-apache --image=gcr.io/google_containers/hpa-example --image-pull-policy='IfNotPresent'  --requests=cpu=200m --expose --port=80
kubectl run --generator=deployment/apps.v1 is DEPRECATED and will be removed in a future version. Use kubectl run --generator=run-pod/v1 or kubectl create instead.
service/php-apache created
deployment.apps/php-apache created

# 查看pod
[root@k8s-master k8s-install]# kubectl get pods
NAME                          READY   STATUS    RESTARTS   AGE
php-apache-799f99c985-zqn77   1/1     Running   0          4s

## 查看资源使用情况
[root@k8s-master k8s-install]# kubectl top pod php-apache-799f99c985-zqn77
NAME                          CPU(cores)   MEMORY(bytes)   
php-apache-799f99c985-zqn77   0m           8Mi

创建HPA控制器

1
kubectl autoscale deployment php-apache --cpu-percent=50 --min=1 --max=10
1
2
3
4
# 查看
[root@k8s-master ~]# kubectl get hpa
NAME         REFERENCE               TARGETS    MINPODS   MAXPODS   REPLICAS   AGE
php-apache   Deployment/php-apache   183%/50%   1         10        4          29m

增加负载，查看负载节点数目

1
2
$ kubectl run -i --tty load-generator --image=busybox --image-pull-policy='IfNotPresent'   /bin/sh
$ while true; do wget -q -O- http://php-apache.default.svc.cluster.local;done
1
2
3
4
5
6
7
8
9
10
11
# 查看pod负载
[root@k8s-master ~]# kubectl top pod php-apache-799f99c985-zqn77  
NAME                          CPU(cores)   MEMORY(bytes)   
php-apache-799f99c985-zqn77   298m         8Mi             
[root@k8s-master ~]# kubectl get pod 
NAME                              READY   STATUS    RESTARTS   AGE
load-generator-76f9c7b5f8-nx5mz   1/1     Running   0          24m
php-apache-799f99c985-nxhzp       1/1     Running   0          79s
php-apache-799f99c985-r9zlp       1/1     Running   0          49s
php-apache-799f99c985-zqn77       1/1     Running   0          30m
[root@k8s-master ~]#

资源限制-Pod

Kubernetes对资源的限制实际上是通过cgroup来控制的，cgroup是容器的一组来控制内核如何运行进程的相关属性集合。针对内存、CPU和各种设备都有对应的cgroup

默认情况下，pod运行没有CPU和内存的限额。这意味着系统中的任何Pod将能够执行该Pod所在的节点一样，消耗足够多的CPU和内存。一般会针对某些应用的pod资源进行资源限制，这个资源限制是通过resources的requests和limits来实现

1
2
3
4
5
6
7
8
9
10
11
spec:
  containers:
  - image: xxx
    name: auth
    resources:
      limits:
        cpu: "4"
        memory: 2Gi
      requests:
      cpu: 250m
      memory: 250Mi

requests要分配的资源，limits为最高请求的资源值，可以简单理解为初始值和最大值

资源限制-名称空间

1 计算资源配额

1
2
3
4
5
6
7
8
9
10
11
12
apiVersion: v1
kind: ResourceQuota
metadata:
  name: compute-resources
  namespace: spark-cluster
spec:
  hard:
    pods: "20"
    requests.cpu: "20"
    requests.memory: 100Gi
    limits.cpu: "40"
    limits.memory: 200Gi

2 配置对象数量配额限制

1
2
3
4
5
6
7
8
9
10
11
12
13
apiVesion: v1
kindL ResourceQuota
metadata:
  name: object-counts
  namespace:spark-cluster
spec:
  hard:
    configmaps: "10"
    persistentvolumeclaims: "4"
    replicationcontrollers: "20"
    secrets: "10"
    services: "10"
    services.loadbalancers: "2"

3 配置CPU和内存limitRange

1
2
3
4
5
6
7
8
9
10
11
12
13
apiVersion: v1
kind: LimitRange
metadata:
  name: mem-limit-range
spec:
  limits:
  - default:
      memory: 50Gi
      cpu: 5
    defaultRequest:
      memory: 1Gi
      cpu: 1
    type: Container
  • defualt即limit的值
  • defaultRequest即request的值