概述

在Kubernetes（K8s）中，升级和回滚可通过Deployment来实现对应用服务无中断或少中断的更新。升级过程通常采用滚动更新策略，逐步替换Pod实例以新版本镜像启动新Pod，同时确保服务始终有足够副本在线提供服务。回滚则是指在升级出现问题时快速恢复到先前稳定版本的操作，K8s会自动基于Deployment的历史记录轻松回滚到之前的一个修订版本。整个过程旨在保障服务连续性和稳定性。

升级

当集群中的某个服务需要升级，我们需要停止该服务下的所有Pod，然后下载新版本并创建新Pod，如果集群规模大，这个行为就成为了挑战。k8s提供了滚动更新功能。

如果Pod是deployment创建，我们直接可以修改镜像名称，系统即可完成自动更新操作。如果在更新过程中发生错误，则可以回滚。

yaml 复制代码

apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-deployment
spec:
  selector:
    matchLabels:
      app: nginx
  replicas: 3
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.7.9
        ports:
        - containerPort: 80

运行并查看结果

bash 复制代码

# 运行
[root@master1 pod]# kubectl apply -f 26.yaml
deployment.apps/nginx-deployment created

# 查看结果
[root@master1 pod]# kubectl get pod
NAME                                READY   STATUS    RESTARTS   AGE
nginx-deployment-5d59d67564-hmpdx   1/1     Running   0          5s
nginx-deployment-5d59d67564-l4qs5   1/1     Running   0          5s
nginx-deployment-5d59d67564-l5gkr   1/1     Running   0          5s

更新镜像版本到1.9.1

bash 复制代码

# 编辑
[root@master1 pod]# kubectl edit deployment/nginx-deployment

# 更改 image 版本号
# wq保存推出

# 查看新创建
[root@master1 pod]# kubectl get pod
NAME                                READY   STATUS    RESTARTS   AGE
nginx-deployment-69c44dfb78-djrfv   1/1     Running   0          3m16s
nginx-deployment-69c44dfb78-jtk2g   1/1     Running   0          2m29s
nginx-deployment-69c44dfb78-nsvdt   1/1     Running   0          2m4s

# 查看更新过程
[root@master1 ~]# kubectl rollout status deployment/nginx-deployment
Waiting for deployment "nginx-deployment" rollout to finish: 1 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 1 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 1 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 2 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 2 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 2 out of 3 new replicas have been updated...
Waiting for deployment "nginx-deployment" rollout to finish: 1 old replicas are pending termination...
Waiting for deployment "nginx-deployment" rollout to finish: 1 old replicas are pending termination...
deployment "nginx-deployment" successfully rolled out

其升级过程就是此消彼长过程

在升级过程中，系统会至少有两个Pod可用，并且最多同时运行4个Pod。这样能够保证在升级的过程中，老系统可以不中断服务。

更新策略有两种：

Recreate：重建，全部删掉Pod，然后重建。设置spec.strategy.type=Recreate
RollingUpdate：滚动更新。设置spec.strategy.type=RollingUpdate
- maxUnavailable: 可用Pod数量上限
- maxSurge：更新过程中Pod总数超过Pod期望副本数的最大值

回滚

默认情况下，所有Deployment的发布和历史记录都被保存在系统中，以便我们随时进行回滚。用以上的例子。

bash 复制代码

[root@master1 pod]# kubectl get pod
NAME                                READY   STATUS    RESTARTS   AGE
nginx-deployment-5d59d67564-4l9c4   1/1     Running   0          3s
nginx-deployment-5d59d67564-62xvx   1/1     Running   0          3s
nginx-deployment-5d59d67564-hltmg   1/1     Running   0          3s

假设我们在更新镜像的时候，版本写错了，写成1.91

bash 复制代码

kubectl set image deployment/nginx-deployment nginx=nginx:1.91

# 查看部署过程会被卡住
[root@master1 pod]# kubectl rollout status deployment/nginx-deployment
Waiting for deployment "nginx-deployment" rollout to finish: 1 out of 3 new replicas have been updated...

# 查看ReplicaSet 可以看见新建的RS(nginx-deployment-d645d84b6)
[root@master1 pod]# kubectl get rs
NAME                          DESIRED   CURRENT   READY   AGE
nginx-deployment-5d59d67564   3         3         3       6m57s
nginx-deployment-d645d84b6    1         1         0       3m3s

# 查看Pod 看到拉去镜像失败
[root@master1 pod]# kubectl get pod
NAME                                READY   STATUS             RESTARTS   AGE
nginx-deployment-5d59d67564-4l9c4   1/1     Running            0          8m50s
nginx-deployment-5d59d67564-62xvx   1/1     Running            0          8m50s
nginx-deployment-5d59d67564-hltmg   1/1     Running            0          8m50s
nginx-deployment-d645d84b6-57w4s    0/1     ImagePullBackOff   0          4m56s

为了解决上述问题，我想需要回滚稳定版本注意：执行命令时增加 --record=true，才可以看到每个版本执行的命令，否则看到的结果如下：

bash 复制代码

[root@master1 pod]# kubectl rollout history deployment/nginx-deployment
deployment.apps/nginx-deployment
REVISION  CHANGE-CAUSE
1         <none>
2         <none>

如果增加了 --record=true 查看结果为

bash 复制代码

[root@master1 pod]# kubectl rollout history deployment/nginx-deployment
deployment.apps/nginx-deployment
REVISION  CHANGE-CAUSE
1         kubectl apply --filename=26.yaml --record=true
2         kubectl set image deployment/nginx-deployment nginx=nginx:1.91 --record=true

回滚操作可以有两种

默认返回上个版本

bash 复制代码

[root@master1 pod]# kubectl rollout undo deployment/nginx-deployment
deployment.apps/nginx-deployment rolled back

# 查看版本
[root@master1 pod]# kubectl get pod
  NAME                                READY   STATUS    RESTARTS   AGE
  nginx-deployment-5d59d67564-ckgk4   1/1     Running   0          6m22s
  nginx-deployment-5d59d67564-nqb52   1/1     Running   0          6m22s
  nginx-deployment-5d59d67564-s7h7m   1/1     Running   0          6m22s

通过版本

bash 复制代码

[root@master1 pod]# kubectl rollout undo deployment/nginx-deployment --to-revision=2
  deployment.apps/nginx-deployment rolled back
  # 查询结果
  [root@master1 pod]# kubectl get pod
  NAME                                READY   STATUS              RESTARTS   AGE
  nginx-deployment-5d59d67564-ckgk4   1/1     Running             0          8m27s
  nginx-deployment-5d59d67564-nqb52   1/1     Running             0          8m27s
  nginx-deployment-5d59d67564-s7h7m   1/1     Running             0          8m27s
  nginx-deployment-d645d84b6-4pwkx    0/1     ContainerCreating   0          4s

以上，Kubernetes 提供了一种声明式的、安全的方式来管理和控制应用服务的版本升级与回滚，极大地增强了系统的可靠性和运维效率。

kubernetes Pod-06 升级与回滚

概述

升级

回滚