k8s滚动更新(六)

容器技术 靠谱运维 725℃ 0评论

实践

滚动更新是一次只更新一小部分副本,成功后,再更新更多的副本,最终完成所有副本的更新。滚动更新的最大的好处是零停机,整个更新过程始终有副本在运行,从而保证了业务的连续性。

下面我们部署三副本应用,初始镜像为 httpd:2.2.31,然后将其更新到 httpd:2.2.32。

第一步: httpd:2.2.31 的配置文件如下:

[root@ken ~]# cat httpd.yml 
apiVersion: apps/v1beta1
kind: Deployment
metadata:
  name: httpd
spec:
  replicas: 3
  template:
    metadata:
      labels:
        run: httpd
    spec:
      containers:
      - name: httpd
        image: httpd:2.2.31
        ports:
        - containerPort: 80

第二步:部署应用并查看

[root@ken ~]# kubectl apply -f httpd.yml
deployment.apps/httpd created

[root@ken ~]# kubectl get deployment -o wide
NAME    READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES         SELECTOR
httpd   0/3     3            0           10s   httpd        httpd:2.2.31   run=httpd
[root@ken ~]# kubectl get replicaset -o wide
NAME             DESIRED CURRENT READY AGE CONTAINERS IMAGES        SELECTOR
httpd-76cfb94bf4 3          3      0   23s  httpd    httpd:2.2.31 pod-template-hash=76cfb94bf4,run=httpd
[root@ken ~]# kubectl get pod -o wide
NAME                     READY   STATUS    RESTARTS   AGE    IP            NODE    NOMINATED NODE   READINESS GATES
httpd-76cfb94bf4-629s2   1/1     Running   0          113s   10.244.1.34   host1   <none>           <none>
httpd-76cfb94bf4-lt9wb   1/1     Running   0          113s   10.244.1.33   host1   <none>           <none>
httpd-76cfb94bf4-n62nj   1/1     Running   0          113s   10.244.2.23   host2   <none>           <none>

部署过程如下:

创建 Deployment httpd

创建 ReplicaSet httpd-76cfb94bf4

创建三个 Pod

当前镜像为 httpd:2.2.31

第三步:将配置文件中 httpd:2.2.31 替换为 httpd:2.2.32,再次执行 kubectl apply。

[root@ken ~]# kubectl apply -f httpd.yml
deployment.apps/httpd configured

[root@ken ~]# kubectl get deployment -o wide
NAME    READY   UP-TO-DATE   AVAILABLE   AGE    CONTAINERS   IMAGES         SELECTOR
httpd   3/3     1            3           4m2s   httpd        httpd:2.2.32   run=httpd

[root@ken ~]# kubectl get replicaset -o wide  #这一步要稍等几分钟才会切换到32版本
NAME               DESIRED   CURRENT   READY   AGE     CONTAINERS   IMAGES         SELECTOR
httpd-6cf6bf9f57   3         3         3       2m47s   httpd        httpd:2.2.32   pod-template-hash=6cf6bf9f57,run=httpd
httpd-76cfb94bf4   0         0         0       6m30s   httpd        httpd:2.2.31   pod-template-hash=76cfb94bf4,run=httpd

[root@ken ~]# kubectl get pod -o wide
NAME                     READY   STATUS    RESTARTS   AGE     IP            NODE    NOMINATED NODE   READINESS GATES
httpd-6cf6bf9f57-5md94   1/1     Running   0          2m58s   10.244.2.24   host2   <none>           <none>
httpd-6cf6bf9f57-nvxnc   1/1     Running   0          75s     10.244.1.35   host1   <none>           <none>
httpd-6cf6bf9f57-v7lpg   1/1     Running   0          22s     10.244.1.36   host1   <none>           <none>

我们发现了如下变化:

Deployment httpd 的镜像更新为 httpd:2.2.32

新创建了 ReplicaSethttpd-6cf6bf9f57,镜像为 httpd:2.2.32,并且管理了三个新的 Pod。

之前的 ReplicaSet httpd-76cfb94bf4里面已经没有任何 Pod。

结论是:ReplicaSethttpd-76cfb94bf4的三个 httpd:2.2.31 Pod 已经被 ReplicaSethttpd-6cf6bf9f57的三个 httpd:2.2.32 Pod 替换了。

第四步:具体过程可以通过 kubectl describe deployment httpd 查看。

[root@ken ~]# kubectl describe deployment httpd
...
Events:
  Type    Reason             Age    From                   Message
  ----    ------             ----   ----                   -------
  Normal  ScalingReplicaSet  9m11s  deployment-controller  Scaled up replica set httpd-76cfb94bf4 to 3
  Normal  ScalingReplicaSet  5m28s  deployment-controller  Scaled up replica set httpd-6cf6bf9f57 to 1
  Normal  ScalingReplicaSet  3m45s  deployment-controller  Scaled down replica set httpd-76cfb94bf4 to 2
  Normal  ScalingReplicaSet  3m45s  deployment-controller  Scaled up replica set httpd-6cf6bf9f57 to 2
  Normal  ScalingReplicaSet  2m52s  deployment-controller  Scaled down replica set httpd-76cfb94bf4 to 1
  Normal  ScalingReplicaSet  2m52s  deployment-controller  Scaled up replica set httpd-6cf6bf9f57 to 3
  Normal  ScalingReplicaSet  2m50s  deployment-controller  Scaled down replica set httpd-76cfb94bf4 to 0

每次只更新替换一个 Pod:

ReplicaSet httpd-6cf6bf9f57 增加一个 Pod,总数为 1。

ReplicaSet httpd-76cfb94bf4 减少一个 Pod,总数为 2。

ReplicaSet httpd-6cf6bf9f57 增加一个 Pod,总数为 2。

ReplicaSet httpd-76cfb94bf4 减少一个 Pod,总数为 1。

ReplicaSet httpd-6cf6bf9f57 增加一个 Pod,总数为 3。

ReplicaSet httpd-76cfb94bf4 减少一个 Pod,总数为 0。

每次替换的 Pod 数量是可以定制的。Kubernetes 提供了两个参数 maxSurge 和 maxUnavailable 来精细控制 Pod 的替换数量,我们将在后面结合 Health Check 特性一起讨论。

更新回滚

kubectl apply 每次更新应用时 Kubernetes 都会记录下当前的配置,保存为一个 revision(版次),这样就可以回滚到某个特定 revision。

默认配置下,Kubernetes 只会保留最近的几个 revision,可以在 Deployment 配置文件中通过 revisionHistoryLimit 属性增加 revision 数量。

第一步:下面实践回滚功能。

应用有如下三个配置文件 httpd.v1.yml,httpd.v2.yml 和 httpd.v3.yml,分别对应不同的 httpd 镜像 2.4.16,2.4.17 和 2.4.18:

第二步:部署应用并更新

后面一个部署的应用,会覆盖掉前面的(名称相同)

[root@ken ~]# kubectl apply -f httpd_v1.yml --record
deployment.apps/httpd created
[root@ken ~]# kubectl get deployment httpd -o wide
NAME    READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES         SELECTOR
httpd   3/3     3            3           18s   httpd        httpd:2.4.16   run=httpd


[root@ken ~]# kubectl apply -f httpd_v2.yml --record
deployment.apps/httpd configured
[root@ken ~]# kubectl get deployment httpd -o wide
NAME    READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES         SELECTOR
httpd   3/3     2            3           75s   httpd        httpd:2.4.17   run=httpd

[root@ken ~]# kubectl apply -f httpd_v3.yml --record
deployment.apps/httpd configured
[root@ken ~]# kubectl get deployment httpd -o wide
NAME    READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES         SELECTOR
httpd   3/3     1            3           94s   httpd        httpd:2.4.18   run=httpd

–record 的作用是将当前命令记录到 revision 记录中,这样我们就可以知道每个 revison 对应的是哪个配置文件。

第三步:通过 kubectl rollout history deployment httpd 查看 revison 历史记录。

[root@ken ~]# kubectl rollout history deployment httpd
deployment.extensions/httpd 
REVISION  CHANGE-CAUSE
1         kubectl apply --filename=httpd_v1.yml --record=true
2         kubectl apply --filename=httpd_v2.yml --record=true
3         kubectl apply --filename=httpd_v3.yml --record=true

CHANGE-CAUSE 就是 –record 的结果。

第四步:如果要回滚到某个版本,比如 revision 1,可以执行命令 kubectl rollout undo deployment httpd –to-revision=1:

[root@ken ~]# kubectl rollout undo deployment httpd --to-revision=1
deployment.extensions/httpd rolled back
[root@ken ~]# kubectl get deployment httpd -o wide
NAME    READY   UP-TO-DATE   AVAILABLE   AGE     CONTAINERS   IMAGES         SELECTOR
httpd   3/3     3            3           5m26s   httpd        httpd:2.4.16   run=httpd

第五步:此时,revison 历史记录也会发生相应变化。

[root@ken ~]# kubectl rollout history deployment httpd
deployment.extensions/httpd 
REVISION  CHANGE-CAUSE
2         kubectl apply --filename=httpd_v2.yml --record=true
3         kubectl apply --filename=httpd_v3.yml --record=true
4         kubectl apply --filename=httpd_v1.yml --record=true

revison 1 变成了 revison 4。不过我们可以通过 CHANGE-CAUSE 知道每个 revison 的具体含义。所以一定要在执行 kubectl apply 时加上 –record参数。

转载请注明:靠谱运维 » k8s滚动更新(六)

喜欢 (0)or分享 (0)
发表我的评论
取消评论

表情