Horizontal Pod Autoscaler官方文档:Pod 水平自动扩缩 | Kubernetes
Pod 水平自动扩缩(Horizontal Pod Autoscaler) 可以基于 CPU 利用率自动扩缩 ReplicationController、Deployment、ReplicaSet 和 StatefulSet 中的 Pod 数量。 除了 CPU 利用率,也可以基于其他应程序提供的 自定义度量指标 来执行自动扩缩。 Pod 自动扩缩不适用于无法扩缩的对象,比如 DaemonSet。
Pod 水平自动扩缩特性由 Kubernetes API 资源和控制器实现。资源决定了控制器的行为。 控制器会周期性地调整副本控制器或 Deployment 中的副本数量,以使得类似 Pod 平均 CPU 利用率、平均内存利用率这类观测到的度量值与用户所设定的目标值匹配。
Kubectl top ->apiserver->metrics server-> kubelet(cadvisor)->pod
HPA是根据指标来进行自动伸缩的,目前HPA有两个版本–v1和v2beta
HPA的API有三个版本,通过kubectl api-versions | grep autoscal可看到
[root@master1 yaml]# kubectl api-versions | grep autosca
autoscaling/v1
autoscaling/v2beta1
autoscaling/v2beta2
查看使用的版本:
kubectl explain hpa
查看指定其他版本:
kubectl explain hpa --api-version=autoscaling/v2beta1
autoscaling/v1只支持基于CPU指标的缩放;
autoscaling/v2beta1支持Resource Metrics(资源指标,如pod内存)和Custom Metrics(自定义指标)的缩放;
autoscaling/v2beta2支持Resource Metrics(资源指标,如pod的内存)和Custom Metrics(自定义指标)和ExternalMetrics
1.部署一下metrics-server,收集集群资源利用率
metrics-server版本获取:
https://github.com/kubernetes-sigs/metrics-server/releases
vim metrics-server.yaml
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: system:aggregated-metrics-reader
labels:
rbac.authorization.k8s.io/aggregate-to-view: "true"
rbac.authorization.k8s.io/aggregate-to-edit: "true"
rbac.authorization.k8s.io/aggregate-to-admin: "true"
rules:
- apiGroups: ["metrics.k8s.io"]
resources: ["pods", "nodes"]
verbs: ["get", "list", "watch"]
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
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: RoleBinding
metadata:
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: apiregistration.k8s.io/v1
kind: APIService
metadata:
name: v1beta1.metrics.k8s.io
spec:
service:
name: metrics-server
namespace: kube-system
group: metrics.k8s.io
version: v1beta1
insecureSkipTLSVerify: true
groupPriorityMinimum: 100
versionPriority: 100
---
apiVersion: v1
kind: ServiceAccount
metadata:
name: metrics-server
namespace: kube-system
---
apiVersion: apps/v1
kind: Deployment
metadata:
name: metrics-server
namespace: kube-system
labels:
k8s-app: metrics-server
spec:
selector:
matchLabels:
k8s-app: metrics-server
template:
metadata:
name: metrics-server
labels:
k8s-app: metrics-server
spec:
serviceAccountName: metrics-server
volumes:
# mount in tmp so we can safely use from-scratch images and/or read-only containers
- name: tmp-dir
emptyDir: {}
containers:
- name: metrics-server
image: registry.cn-shenzhen.aliyuncs.com/lishanbin/metrics-server:v0.3.7
imagePullPolicy: IfNotPresent
args:
- --cert-dir=/tmp
- --secure-port=4443
- --kubelet-insecure-tls
- --kubelet-preferred-address-types=InternalIP
ports:
- name: main-port
containerPort: 4443
protocol: TCP
securityContext:
readOnlyRootFilesystem: true
runAsNonRoot: true
runAsUser: 1000
volumeMounts:
- name: tmp-dir
mountPath: /tmp
#nodeSelector:
# kubernetes.io/os: linux
# kubernetes.io/arch: "amd64"
---
apiVersion: v1
kind: Service
metadata:
name: metrics-server
namespace: kube-system
labels:
kubernetes.io/name: "Metrics-server"
kubernetes.io/cluster-service: "true"
spec:
selector:
k8s-app: metrics-server
ports:
- port: 443
protocol: TCP
targetPort: main-port
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
name: system:metrics-server
rules:
- apiGroups:
- ""
resources:
- pods
- nodes
- nodes/stats
- namespaces
- configmaps
verbs:
- get
- list
- watch
---
apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
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
[root@master1 yaml]# kubectl api-versions |grep metrics
metrics.k8s.io/v1beta1
[root@master1 yaml]# kubectl top nodes
NAME CPU(cores) CPU% MEMORY(bytes) MEMORY%
192.168.203.219 180m 2% 2936Mi 18%
192.168.203.220 499m 6% 1662Mi 10%
192.168.203.221 203m 2% 1883Mi 11%
192.168.203.223 151m 1% 9805Mi 61%
192.168.203.226 343m 4% 10645Mi 67%
192.168.203.228 299m 3% 10698Mi 67%
2.hpa基于cpu自动扩缩容
HPA伸缩过程:
收集HPA控制下所有Pod最近的cpu使用情况(CPU utilization)
对比在扩容条件里记录的cpu限额(CPUUtilization)
调整实例数(必须要满足不超过最大/最小实例数)
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。
HPA进行伸缩算法:
计算公式:TargetNumOfPods = ceil(sum(CurrentPodsCPUUtilization) / Target)
ceil()表示取大于或等于某数的最近一个整数
每次扩容后冷却3分钟才能再次进行扩容,而缩容则要等5分钟后。
当前Pod Cpu使用率与目标使用率接近时,不会触发扩容或缩容:
触发条件:avg(CurrentPodsConsumption) / Target >1.1 或 <0.9
kubectl autoscale deployment f1 -n lishanbin --cpu-percent=1 --min=1 --max=2
yaml例子
---
apiVersion: autoscaling/v1
kind: HorizontalPodAutoscaler
metadata:
name: ubb
namespace: hpa
spec:
maxReplicas: 4
minReplicas: 1
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: ubb
targetCPUUtilizationPercentage: 20
查看hpa
[root@master1 yaml]# kubectl get hpa -n lishanbin
NAME REFERENCE TARGETS MINPODS MAXPODS REPLICAS AGE
f1 Deployment/f1 2%/1% 1 2 2 3m33s
验证:可以查看到创建了一个新的pod
[root@master1 yaml]# kubectl get pod -n lishanbin
NAME READY STATUS RESTARTS AGE
f1-856675bdbf-n4fmk 1/1 Running 0 11d
f1-856675bdbf-n9gg8 1/1 Running 0 4m5s
[root@master1 yaml]# kubectl get deployment -n baichuan
NAME READY UP-TO-DATE AVAILABLE AGE
f1 2/2 2 2 447d
f1web 1/1 1 1 447d
发现hpa已经超过了预定值,随之pod的副本数也变成了1个,最多可变成2个,停止负载后,副本数也会变成一个
如果出现了 failed to get cpu utilization: missing request for cpu
这样的错误信息。这是因为我们上面创建的 Pod 对象没有添加 request 资源声明,这样导致 HPA 读取不到 CPU 指标信息,所以如果要想让 HPA 生效,对应的 Pod 资源必须添加 requests 资源声明,
假如targets字段有显示unknown
原因:
刚建立,等待一段时间再查看
需要自动伸缩的目标资源并没有进行资源限制
3.hpa基于内存扩缩容
targetAverageUtilization 表示的是百分比
targetAverageValue 表示的是数值,比如100m的CPU、100Mi的内存
apiVersion: autoscaling/v2beta1
kind: HorizontalPodAutoscaler
metadata:
name: kevin-t-hap
namespace: lishanbin
spec:
scaleTargetRef:
apiVersion: apps/v1
kind: Deployment
name: f1
minReplicas: 1
maxReplicas: 2
metrics:
- type: Resource
resource:
name: cpu
targetAverageUtilization: 80
- type: Resource
resource:
name: memory
targetAverageValue: 30Mi
如上,设置了f1的deployment控制的pod的HPA限制,当cpu使用超过设置的80%,内存使用超过30Mi时就触发自动扩容,副本数最小为1,最大为2。