k8s1.9 滚动升级（rollingupdate）和自动缩扩容（HPA）

快来打我* 2022-02-16 14:26 276阅读 0赞

参考：[https://jimmysong.io/kubernetes-handbook/concepts/deployment.html][https_jimmysong.io_kubernetes-handbook_concepts_deployment.html]  
首先创建pod，我这有现成的，如果没有参考链接创建pod。

[root@master01 ~]# kubectl get deployment product-htjf-login-web-profile-a
    NAME                               DESIRED   CURRENT   UP-TO-DATE   AVAILABLE   AGE
    product-htjf-login-web-profile-a   2         2         2            2           5h
    [root@master01 ~]# kubectl get pod product-htjf-login-web-profile-a-586bf5dc8-nltkh -o yaml
    apiVersion: v1
    kind: Pod
    metadata:
      creationTimestamp: 2019-04-22T07:42:22Z
      generateName: product-htjf-login-web-profile-a-586bf5dc8-
      labels:
        app: product-htjf-login-web
        pod-template-hash: "142691874"
        profile: profile-a
      name: product-htjf-login-web-profile-a-586bf5dc8-nltkh
      namespace: default
      ownerReferences:
      - apiVersion: extensions/v1beta1
        blockOwnerDeletion: true
        controller: true
        kind: ReplicaSet
        name: product-htjf-login-web-profile-a-586bf5dc8
        uid: 103f4cd6-64c5-11e9-b41c-005056b678c1
      resourceVersion: "43102672"
      selfLink: /api/v1/namespaces/default/pods/product-htjf-login-web-profile-a-586bf5dc8-nltkh
      uid: 2a3b5374-64d2-11e9-b41c-005056b678c1
    spec:
      containers:
      - image: harbor.reg/xx/product-htjf-login-web:9  ##目前的版本是9
        imagePullPolicy: IfNotPresent
        name: product-htjf-login-web-profile-a
        ports:
        - containerPort: 8080
          protocol: TCP
        resources:
          limits:
            memory: 2000Mi
          requests:
            memory: 1500Mi
        terminationMessagePath: /dev/termination-log
        terminationMessagePolicy: File
        volumeMounts:
        - mountPath: xxx
          name: cephfsmount
        - mountPath: xxx
          name: productmfs
        - mountPath: /var/run/secrets/kubernetes.io/serviceaccount
          name: default-token-mqtdl
          readOnly: true
      dnsPolicy: ClusterFirst
      nodeName: node01
      restartPolicy: Always
      schedulerName: default-scheduler
      securityContext: { }
      serviceAccount: default
      serviceAccountName: default
      terminationGracePeriodSeconds: 30
      volumes:
      - cephfs:
          monitors:
          - 10.100.101.xxx:16789
          - 10.100.101.xxx:16789
          - 10.100.101.xxx:16789
          secretFile: /etc/ceph/admin.secret
          user: admin
        name: cephfsmount
      - hostPath:
          path: xxx
          type: ""
        name: productmfs
      - name: default-token-mqtdl
        secret:
          defaultMode: 420
          secretName: default-token-mqtdl
    status:
      conditions:
      - lastProbeTime: null
        lastTransitionTime: 2019-04-22T07:42:22Z
        status: "True"
        type: Initialized
      - lastProbeTime: null
        lastTransitionTime: 2019-04-22T07:42:24Z
        status: "True"
        type: Ready
      - lastProbeTime: null
        lastTransitionTime: 2019-04-22T07:42:22Z
        status: "True"
        type: PodScheduled
      containerStatuses:
      - containerID: docker://89f1004f5bc58b04492bca24b107c952fb3143d613da043ecede897ba063863f
        image: harbor.reg/pro_jinfu/product-htjf-login-web:10
        imageID: docker-pullable://harbor.reg/pro_jinfu/product-htjf-login-web@sha256:d88054e3836f172991134561af06c2f66ea6889f40b69c9b824c40a222ae4be4
        lastState: { }
        name: product-htjf-login-web-profile-a
        ready: true
        restartCount: 0
        state:
          running:
            startedAt: 2019-04-22T07:42:23Z
      hostIP: 10.100.10.190
      phase: Running
      podIP: 10.2.79.9
      qosClass: Burstable
      startTime: 2019-04-22T07:42:22Z

可以看到pod当前的版本是9

###### 升级pod ######

kubectl  set image deployment/product-htjf-login-web-profile-a product-htjf-login-web-profile-a=harbor.reg/xxx/product-htjf-login-web:10 --record=true
    
    [root@master01 ~]# kubectl rollout history deployment product-htjf-login-web-profile-a
    deployments "product-htjf-login-web-profile-a"
    REVISION  CHANGE-CAUSE
    3         kubectl set image deployment/product-htjf-login-web-profile-a product-htjf-login-web-profile-a=harbor.reg/xxx/product-htjf-login-web:10 --record=true
    4

因为我们创建 Deployment 的时候使用了–record参数可以记录命令，我们可以很方便的查看每次 revision 的变化。

###### 查看pod的总体描述 ######

[root@master01 ~]# kubectl rollout history deployment product-htjf-login-web-profile-a
    deployments "product-htjf-login-web-profile-a"
    REVISION  CHANGE-CAUSE
    3         kubectl set image deployment/product-htjf-login-web-profile-a product-htjf-login-web-profile-a=harbor.reg/xxx/product-htjf-login-web:9 --record=true
    4         kubectl set image deployment/product-htjf-login-web-profile-a product-htjf-login-web-profile-a=harbor.reg/xxx/product-htjf-login-web:10 --record=true
    
    [root@master01 ~]# kubectl describe deployment product-htjf-login-web-profile-a 
    Name:                   product-htjf-login-web-profile-a
    Namespace:              default
    CreationTimestamp:      Mon, 22 Apr 2019 10:59:02 +0800
    Labels:                 app=product-htjf-login-web
                            profile=profile-a
    Annotations:            deployment.kubernetes.io/revision=4
                            kubectl.kubernetes.io/last-applied-configuration={ "apiVersion":"extensions/v1beta1","kind":"Deployment","metadata":{ "annotations":{ },"name":"product-htjf-login-web-profile-a","namespace":"default"},"s... kubernetes.io/change-cause=kubectl set image deployment/product-htjf-login-web-profile-a product-htjf-login-web-profile-a=harbor.reg/xxx/product-htjf-login-web:10 --record=true Selector: app=product-htjf-login-web,profile=profile-a Replicas: 2 desired | 2 updated | 2 total | 2 available | 0 unavailable StrategyType: RollingUpdate MinReadySeconds: 0 RollingUpdateStrategy: 1 max unavailable, 1 max surge Pod Template: Labels: app=product-htjf-login-web profile=profile-a Containers: product-htjf-login-web-profile-a: Image: harbor.reg/pro_jinfu/product-htjf-login-web:10 Port: 8080/TCP Limits: memory: 2000Mi Requests: memory: 1500Mi Environment: <none> Mounts: /data/WEBLOG from cephfsmount (rw) /home/nfs from productmfs (rw) Volumes: cephfsmount: Type: CephFS (a CephFS mount on the host that shares a pod's lifetime) Monitors: [10.100.xx:16789 10.100.xx:16789 10.100.xx:16789] Path: User: admin SecretFile: /xxx/admin.secret SecretRef: <nil> ReadOnly: false productmfs: Type: HostPath (bare host directory volume) Path: xxx HostPathType: Conditions: Type Status Reason ---- ------ ------ Available True MinimumReplicasAvailable OldReplicaSets: <none> NewReplicaSet: product-htjf-login-web-profile-a-586bf5dc8 (2/2 replicas created) Events: Type Reason Age From Message ---- ------ ---- ---- ------- Normal ScalingReplicaSet 36m (x2 over 2h) deployment-controller Scaled up replica set product-htjf-login-web-profile-a-586bf5dc8 to 1 Normal ScalingReplicaSet 36m (x2 over 2h) deployment-controller Scaled down replica set product-htjf-login-web-profile-a-7c7dcbdd94 to 1 Normal ScalingReplicaSet 36m (x2 over 2h) deployment-controller Scaled up replica set product-htjf-login-web-profile-a-586bf5dc8 to 2 Normal DeploymentRollback 36m deployment-controller Rolled back deployment "product-htjf-login-web-profile-a" to revision 2
      Normal  ScalingReplicaSet   36m (x2 over 2h)  deployment-controller  Scaled down replica set product-htjf-login-web-profile-a-7c7dcbdd94 to 0

`Image: harbor.reg/pro_jinfu/product-htjf-login-web:10`已经升级到10版本了

##### 查看状态和历史升级的版本 #####

kubectl  rollout status deployment  product-htjf-login-web-profile-a
     [root@master01 ~]# kubectl rollout history deployment product-htjf-login-web-profile-a
    deployments "product-htjf-login-web-profile-a"
    REVISION  CHANGE-CAUSE
    3         kubectl set image deployment/product-htjf-login-web-profile-a product-htjf-login-web-profile-a=harbor.reg/xxx/product-htjf-login-web:9 --record=true
    4         kubectl set image deployment/product-htjf-login-web-profile-a product-htjf-login-web-profile-a=harbor.reg/xxx/product-htjf-login-web:10 --record=true

##### 回退到历史版本 #####

回退到上一个版本

kubectl  rollout undo deployment  product-htjf-login-web-profile-a

回退到指定版本(指定版本这里指history版本号 查看上面的history 可以回滚3或者4)

kubectl  rollout undo deployment  product-htjf-login-web-profile-b --to-revision=4

可以通过设置.spec.revisonHistoryLimit项来指定 deployment 最多保留多少 revision 历史记录。默认的会保留所有的 revision；如果将该项设置为0，Deployment就不允许回退了。

spec:
      minReadySeconds: 5
      strategy:
        type: RollingUpdate
        rollingUpdate:
          maxUnavailable: 1
          maxSurge: 1
      replicas: 2
      revisionHistoryLimit: 10
      template:

#### 自动扩容 ####

可以使用以下命令扩容 Deployment：

$ kubectl scale deployment nginx-deployment --replicas 10
    deployment "nginx-deployment" scaled

假设您的集群中启用了horizontal pod autoscaling，您可以给 Deployment 设置一个 autoscaler，基于当前 Pod的 CPU 利用率选择最少和最多的 Pod 数。

$ kubectl autoscale deployment nginx-deployment --min=10 --max=15 --cpu-percent=80
    deployment "nginx-deployment" autoscaled

[https_jimmysong.io_kubernetes-handbook_concepts_deployment.html]: https://jimmysong.io/kubernetes-handbook/concepts/deployment.html