您当前的位置： 首页 > 慢生活 > 程序人生 网站首页程序人生

27、资源控制器 - RS、Deployment

发布时间：2022-12-18 17:46:25编辑：雪饮阅读（）

Step1

上篇中部署的nginx对应的这3个pod，首先确认是都可以访问的

[root@k8s-master01 ~]# kubectl get pod -o wide

NAME                               READY   STATUS    RESTARTS   AGE   IP            NODE         NOMINATED NODE   READINESS GATES

nginx-deployment-5df65767f-skm5n   1/1     Running   1          17h   10.224.1.51   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-vbw8f   1/1     Running   1          17h   10.224.1.52   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-xrvsh   1/1     Running   1          17h   10.224.2.37   k8s-node02   <none>           <none>

[root@k8s-master01 ~]# curl 10.224.1.51

Hello MyApp | Version: v1 | <a href="hostname.html">Pod Name</a>

[root@k8s-master01 ~]# curl 10.224.1.52

Hello MyApp | Version: v1 | <a href="hostname.html">Pod Name</a>

[root@k8s-master01 ~]# curl 10.224.2.37

Hello MyApp | Version: v1 | <a href="hostname.html">Pod Name</a>

Step2

既然期望值为3是可以，则尝试下扩容到10个

[root@k8s-master01 ~]# kubectl scale deployment nginx-deployment --replicas=10

deployment.extensions/nginx-deployment scaled

10个也照样没有问题

[root@k8s-master01 ~]# kubectl get pod -o wide

NAME                               READY   STATUS    RESTARTS   AGE     IP            NODE         NOMINATED NODE   READINESS GATES

nginx-deployment-5df65767f-76ktt   1/1     Running   0          2m30s   10.224.1.54   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-dnvmh   1/1     Running   0          2m30s   10.224.2.40   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-h8kr2   1/1     Running   0          2m30s   10.224.2.39   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-rst76   1/1     Running   0          2m30s   10.224.2.42   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-scmp4   1/1     Running   0          2m30s   10.224.1.55   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-skm5n   1/1     Running   1          17h     10.224.1.51   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-tzrr8   1/1     Running   0          2m30s   10.224.1.53   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-vbw8f   1/1     Running   1          17h     10.224.1.52   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-xrvsh   1/1     Running   1          17h     10.224.2.37   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-zcr5k   1/1     Running   0          2m30s   10.224.2.41   k8s-node02   <none>           <none>

可以看到有3个pod分别出现了一次重启，这个无碍，我认为是我这里自上篇到这篇之间有关闭整个k8s目前关联的所有服务器以及master的原因。

但是rs好像没有变（这里说的没有变应该是指rs的数量并不像pod那样一下多了10个条目）

[root@k8s-master01 ~]# kubectl get rs

NAME                         DESIRED   CURRENT   READY   AGE

nginx-deployment-5df65767f   10        10        10      17h

Step3

但当批量变更镜像（这种有多个pod的情况下镜像变更不就是相当于批量变更嘛）可以看到是有点像是新增加10个占位，然后逐个将原10个pod的镜像弄过来，可能说的不够准确。

[root@k8s-master01 ~]# kubectl set image deployment/nginx-deployment nginx=wangyanglinux/myapp:v2

deployment.extensions/nginx-deployment image updated

[root@k8s-master01 ~]# kubectl get rs

NAME                          DESIRED   CURRENT   READY   AGE

nginx-deployment-5c478875d8   5         5         3       11s

nginx-deployment-5df65767f    6         6         6       17h

[root@k8s-master01 ~]# kubectl get rs

NAME                          DESIRED   CURRENT   READY   AGE

nginx-deployment-5c478875d8   10        10        10      19s

nginx-deployment-5df65767f    0         0         0       17h

[root@k8s-master01 ~]# kubectl get rs

NAME                          DESIRED   CURRENT   READY   AGE

nginx-deployment-5c478875d8   10        10        10      21s

nginx-deployment-5df65767f    0         0         0       17h

[root@k8s-master01 ~]# kubectl get rs

NAME                          DESIRED   CURRENT   READY   AGE

nginx-deployment-5c478875d8   10        10        10      22s

nginx-deployment-5df65767f    0         0         0       17h

[root@k8s-master01 ~]# kubectl get rs

NAME                          DESIRED   CURRENT   READY   AGE

nginx-deployment-5c478875d8   10        10        10      22s

nginx-deployment-5df65767f    0         0         0       17h

[root@k8s-master01 ~]# kubectl get rs

NAME                          DESIRED   CURRENT   READY   AGE

nginx-deployment-5c478875d8   10        10        10      23s

nginx-deployment-5df65767f    0         0         0       17h

那么镜像改变后，各个pod的ip也发生变化了，这里以最后的这个pod的ip作为访问入口，以验证版本确实是改变了。

[root@k8s-master01 ~]# kubectl get pod -o wide

NAME                                READY   STATUS    RESTARTS   AGE     IP            NODE         NOMINATED NODE   READINESS GATES

nginx-deployment-5c478875d8-47rnz   1/1     Running   0          4m29s   10.224.1.59   k8s-node01   <none>           <none>

nginx-deployment-5c478875d8-4drq8   1/1     Running   0          4m41s   10.224.1.56   k8s-node01   <none>           <none>

nginx-deployment-5c478875d8-4sjvg   1/1     Running   0          4m27s   10.224.1.60   k8s-node01   <none>           <none>

nginx-deployment-5c478875d8-6jc64   1/1     Running   0          4m32s   10.224.1.57   k8s-node01   <none>           <none>

nginx-deployment-5c478875d8-d45b8   1/1     Running   0          4m30s   10.224.2.45   k8s-node02   <none>           <none>

nginx-deployment-5c478875d8-fk874   1/1     Running   0          4m30s   10.224.1.58   k8s-node01   <none>           <none>

nginx-deployment-5c478875d8-m2xzh   1/1     Running   0          4m32s   10.224.2.44   k8s-node02   <none>           <none>

nginx-deployment-5c478875d8-mm4v7   1/1     Running   0          4m28s   10.224.2.47   k8s-node02   <none>           <none>

nginx-deployment-5c478875d8-r46pp   1/1     Running   0          4m41s   10.224.2.43   k8s-node02   <none>           <none>

nginx-deployment-5c478875d8-rwbd7   1/1     Running   0          4m29s   10.224.2.46   k8s-node02   <none>           <none>

[root@k8s-master01 ~]# curl 10.224.2.46

Hello MyApp | Version: v2 | <a href="hostname.html">Pod Name</a>

之前step1哪里访问是v1，现在是v2，是变了。

Step4

上面不是更换镜像了嘛，那么我们还可以回滚到之前的镜像

[root@k8s-master01 ~]# kubectl rollout undo deployment/nginx-deployment

deployment.extensions/nginx-deployment rolled back

可以看到又变成v1了

[root@k8s-master01 ~]# kubectl get pod -o wide

NAME                               READY   STATUS    RESTARTS   AGE   IP            NODE         NOMINATED NODE   READINESS GATES

nginx-deployment-5df65767f-2kxfk   1/1     Running   0          46s   10.224.1.63   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-4xbdl   1/1     Running   0          47s   10.224.1.62   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-gjxpq   1/1     Running   0          43s   10.224.1.65   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-h7pg7   1/1     Running   0          48s   10.224.1.61   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-hdvsm   1/1     Running   0          46s   10.224.2.50   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-lxfpw   1/1     Running   0          41s   10.224.1.66   k8s-node01   <none>           <none>

nginx-deployment-5df65767f-mvbgr   1/1     Running   0          48s   10.224.2.48   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-thbk4   1/1     Running   0          47s   10.224.2.49   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-w6427   1/1     Running   0          45s   10.224.2.51   k8s-node02   <none>           <none>

nginx-deployment-5df65767f-zfg4z   1/1     Running   0          43s   10.224.1.64   k8s-node01   <none>           <none>

[root@k8s-master01 ~]# curl 10.224.1.64

Hello MyApp | Version: v1 | <a href="hostname.html">Pod Name</a>

也可以查看回滚状态

[root@k8s-master01 ~]# kubectl rollout status deployment/nginx-deployment

deployment "nginx-deployment" successfully rolled out

Step5

也可以查看回滚历史

[root@k8s-master01 ~]# kubectl rollout history deployment/nginx-deployment

deployment.extensions/nginx-deployment

REVISION  CHANGE-CAUSE

2         kubectl apply --filename=deployment.yaml --record=true

3         kubectl apply --filename=deployment.yaml --record=true

这里可以看到change-cause并非空值，是因为我们创建deployment时候

kubectl apply -f deployment.yaml –record

的—record参数的作用

我们删除全部deployment，并重建并不带—record试试

[root@k8s-master01 ~]# kubectl delete deployment --all

deployment.extensions "nginx-deployment" deleted

[root@k8s-master01 ~]# kubectl apply -f deployment.yaml

deployment.extensions/nginx-deployment created

然后再次设置镜像，这次搞个v3出来

[root@k8s-master01 ~]# kubectl set image deployment/nginx-deployment nginx=wangyanglinux/myapp:v3

deployment.extensions/nginx-deployment image updated

然后也可以马上查看回滚状态，如果你电脑性能快，并且你手速慢，就看到下面这样的结果，否则会看到一整个类似各个pod就绪的过程，最后才是下面这个结果

[root@k8s-master01 ~]# kubectl rollout status deployment/nginx-deployment

deployment "nginx-deployment" successfully rolled out

这次可以看到change-cause字段就没有内容了

[root@k8s-master01 ~]# kubectl rollout history deployment/nginx-deployment

deployment.extensions/nginx-deployment

REVISION  CHANGE-CAUSE

1         <none>

2         <none>

Step6

貌似是每更换一次镜像就能产生一次回滚历史记录，比如下面我们再次更新到指定（这里不叫指定，误导了。。。）版本镜像（v2），就又多出一条记录

[root@k8s-master01 ~]# kubectl set image deployment/nginx-deployment nginx=wangyanglinux/myapp:v2

deployment.extensions/nginx-deployment image updated

[root@k8s-master01 ~]# kubectl rollout history deployment/nginx-deployment

deployment.extensions/nginx-deployment

REVISION  CHANGE-CAUSE

1         <none>

2         <none>

3         <none>

 

Step7

回滚到指定版本

[root@k8s-master01 ~]# kubectl rollout undo deployment/nginx-deployment --to-revision=1

deployment.extensions/nginx-deployment rolled back

这里回滚到指定版本，那个—to-revision的参数值应该就是上面那个回滚历史里面的那个最前面的历史记录id吧

[root@k8s-master01 ~]# kubectl get pod -o wide

NAME                               READY   STATUS    RESTARTS   AGE   IP            NODE         NOMINATED NODE   REA                                                                                               DINESS GATES

nginx-deployment-5df65767f-28v7g   1/1     Running   0          19s   10.224.1.72   k8s-node01   <none>           <no                                                                                               ne>

nginx-deployment-5df65767f-c74cg   1/1     Running   0          20s   10.224.2.57   k8s-node02   <none>           <no                                                                                               ne>

nginx-deployment-5df65767f-n4gl9   1/1     Running   0          20s   10.224.1.71   k8s-node01   <none>           <no                                                                                               ne>

[root@k8s-master01 ~]# curl 10.224.1.71

Hello MyApp | Version: v1 | <a href="hostname.html">Pod Name</a>

Step8

回滚状态的验证

对于回滚状态据说还可以验证变量$?,若值为0，表示也是ok的

[root@k8s-master01 ~]# kubectl rollout status deployment/nginx-deployment

deployment "nginx-deployment" successfully rolled out

[root@k8s-master01 ~]# echo $?

0

这里稍微科普下

$?：是上一指令的返回值，成功是0，不成功是1。一般来说，UNIX(linux) 系统的进程以执行系统调用exit() 来结束的。这个回传值就是status值。回传给父进程，用来检查子进程的执行状态。一般指令程序倘若执行成功，其回传值为 0；失败为 1。

这属于linux的知识，并非k8s独有

Step9

Rs的条目数量，好像也是对应着回滚历史记录的条目数量的哈

[root@k8s-master01 ~]# kubectl rollout history deployment/nginx-deployment

deployment.extensions/nginx-deployment

REVISION  CHANGE-CAUSE

2         <none>

3         <none>

4         <none>

 

[root@k8s-master01 ~]# kubectl get rs

NAME                          DESIRED   CURRENT   READY   AGE

nginx-deployment-5c478875d8   0         0         0       9m2s

nginx-deployment-5df65767f    3         3         3       14m

nginx-deployment-68bbd86499   0         0         0       13m