Kubernetes的Deployment Scale Up or Down（伸缩）

FeiYi

Kubernetes

发布于：2020年5月13日

次浏览

字数：1.1k字

时长：5分钟

标题中所说的Scale Up or Down也就是伸缩，Deployment的伸缩也是主要对pod的伸缩，达到高可用以及负载的目的。

使用yaml运行几个nginx

编写文件

[root@node1 ~]# vim nginx.yml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: nginx-deployment
spec:
  selector:
    matchLabels:
      app: nginx
  replicas: 2
  template:
    metadata:
      labels:
        app: nginx
    spec:
      containers:
      - name: nginx
        image: nginx:1.8
        ports:
        - containerPort: 80

运行nginx-deployment

[root@node1 ~]# kubectl apply -f nginx.yml 
deployment.apps/nginx-deployment created

查看运行起来的pod容器

[root@node1 ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE     IP           NODE
nginx-deployment-5f8c6846ff-446sl   1/1     Running   0          9m46s   10.244.2.2   node2
nginx-deployment-5f8c6846ff-4qhvm   1/1     Running   0          9m46s   10.244.1.2   node3

pod的伸缩

其实也特别简单，就是直接修改yaml文件的replicas配置项的数量

[root@node1 ~]# vim nginx.yml
# 修改replicas
  replicas: 5

将pod从2个增加为5个

[root @node1 ~]# kubectl apply -f nginx.yml 
deployment.apps/nginx-deployment configured

再次查看

[root@node1 ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE   IP           NODE    NOMINATED NODE   READINESS GATES
nginx-deployment-5f8c6846ff-446sl   1/1     Running   0          13m   10.244.2.2   node2
nginx-deployment-5f8c6846ff-4qhvm   1/1     Running   0          13m   10.244.1.2   node3
nginx-deployment-5f8c6846ff-5srk7   1/1     Running   0          27s   10.244.1.3   node3
nginx-deployment-5f8c6846ff-h5cgv   1/1     Running   0          27s   10.244.2.3   node2
nginx-deployment-5f8c6846ff-stgdn   1/1     Running   0          27s   10.244.2.4   node2

就这么简单，伸缩已经完成了，不会停掉之前的2个pod，直接在2个的基础上，保证pod总数为5

管理节点参与运算调度

发现所有的pod都运行在node2和node3，k8s中为了管理节点的安全考虑，在分配pod运行主机时，管理节点不参与运算。

如果希望管理节点参与运算，需要执行
kubectl taint node 管理节点主机名 node-role.kubernetes.io/master-

再次恢复管理节点的状态，执行
kubectl taint node 管理节点主机名 node-role.kubernetes.io/master="":NoSchedule

[root@node1 ~]# kubectl taint node node1 node-role.kubernetes.io/master-
node/node1 untainted

重新伸缩，查看管理节点上是否运行pod

[root@node1 ~]# vim nginx.yml
# 修改replicas
  replicas: 9

进行伸缩配置

[root@node1 ~]# kubectl apply -f nginx.yml 
deployment.apps/nginx-deployment configured

查看pod运行节点

经过我的验证，管理节点虽然可以参与运算，但是在少量pod运行时，还是尽量不会去将pod运行到管理节点，直到我修改为15个pod，才在管理节点勉强运行了2个pod，又修改了27个pod，才运行的多了一点

[root@node1 ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE   IP           NODE 
nginx-deployment-5f8c6846ff-4qhvm   1/1     Running   0          81m   10.244.1.2   node3
nginx-deployment-5f8c6846ff-5kk2c   1/1     Running   0          2s    10.244.0.10  node1
nginx-deployment-5f8c6846ff-8s5mk   1/1     Running   0          2s    10.244.2.37  node2
nginx-deployment-5f8c6846ff-b9rdj   1/1     Running   0          19s   10.244.2.34  node2
nginx-deployment-5f8c6846ff-c8dv8   1/1     Running   0          19s   10.244.1.21  node3
nginx-deployment-5f8c6846ff-fk994   1/1     Running   0          2s    10.244.2.36  node2
nginx-deployment-5f8c6846ff-g2sqz   1/1     Running   0          19s   10.244.2.33  node2
nginx-deployment-5f8c6846ff-hks7l   1/1     Running   0          19s   10.244.1.22  node3
nginx-deployment-5f8c6846ff-lhxl4   1/1     Running   0          19s   10.244.2.35  node2
nginx-deployment-5f8c6846ff-lzct7   1/1     Running   0          2s    10.244.0.11  node1
nginx-deployment-5f8c6846ff-m5vs9   1/1     Running   0          19s   10.244.1.23  node3
nginx-deployment-5f8c6846ff-nwbwk   1/1     Running   0          19s   10.244.2.32  node2
nginx-deployment-5f8c6846ff-rcjlg   1/1     Running   0          19s   10.244.2.31  node2
nginx-deployment-5f8c6846ff-sfq2d   1/1     Running   0          2s    10.244.1.24  node3
nginx-deployment-5f8c6846ff-tzh8s   1/1     Running   0          2s    10.244.1.25  node3

恢复管理节点的only状态

设置值为NoScheduler，不让调度器进行调度

[root@node1 ~]# kubectl taint node node1 node-role.kubernetes.io/master="":NoSchedule
node/node1 tainted

将pod伸缩为3个再次查看，结果发现还有node1的pod在运行

[root@node1 ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE    IP           NODE
nginx-deployment-5f8c6846ff-4qhvm   1/1     Running   0          88m    10.244.1.2   node3
nginx-deployment-5f8c6846ff-5kk2c   1/1     Running   0          7m35s  10.244.0.10  node1
nginx-deployment-5f8c6846ff-lzct7   1/1     Running   0          7m35s  10.244.0.11  node1

我以为是反应慢，重复试了好几次都一样，最后将pod伸缩为1个

[root@node1 ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE    IP           NODE
nginx-deployment-5f8c6846ff-4qhvm   1/1     Running   0          88m    10.244.1.2   node3

然后再次伸缩为多个pod

[root@node1 ~]# kubectl get pod -o wide
NAME                                READY   STATUS    RESTARTS   AGE   IP            NODE 
nginx-deployment-5f8c6846ff-4qhvm   1/1     Running   0          90m   10.244.1.2    node3 
nginx-deployment-5f8c6846ff-5j87g   1/1     Running   0          3s    10.244.1.26   node3 
nginx-deployment-5f8c6846ff-6kd6h   1/1     Running   0          3s    10.244.2.38   node2

以上就是k8s的pod的简单伸缩