k8s指定节点部署

在一些业务场景中，会需要将一些pod部署到指定node，按照默认的调度规则，pod会优先分配到负载较小的node中，难免会出现多个pod资源竞争的情况。
k8s有两种常用的方法可以实现将指定pod分配到指定node中。

• nodeName

  在部署的yaml文件中，对xxx.spec.nodeName指定节点名称，则该pod将只会在该node上进行部署：

$ vim websvr.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: websvr1-deployment
spec:
  selector:
    matchLabels:
      app: websvr1
  replicas: 3
  template:
    metadata:
      labels:
        app: websvr1
    spec:
      nodeName: k8s-node1     #指定节点名称
      containers:
      - name: websvr1
        image: websvr:v1
        ports:
        - containerPort: 3000

---

apiVersion: v1
kind: Service
metadata:
  name: websvr1-service
spec:
  selector:
    app: websvr1
  ports:
  - protocol: TCP
    port: 3000
    targetPort: 3000

$ kubectl apply -f websvr1.yaml

NAME                                  READY   STATUS    RESTARTS   AGE   IP             NODE        NOMINATED NODE   READINESS GATES
websvr1-deployment-54cffcc8b4-jw9fl   1/1     Running   0          15s   10.244.36.90   k8s-node1   <none>           <none>
websvr1-deployment-54cffcc8b4-s97ln   1/1     Running   0          15s   10.244.36.92   k8s-node1   <none>           <none>
websvr1-deployment-54cffcc8b4-wglpb   1/1     Running   0          15s   10.244.36.91   k8s-node1   <none>           <none>

可以看到websvr只部署在了k8s-node1上，那么如果现在有三个node，一个pod需要部署在其中两个不同node中，是不是要指定两个node的名称呢，其实这种场景可以使用第二种指定分配的方法：

• nodeSelector

  和nodeName不同，nodeSelector是可以指定某一种类型的node进行分配，这种类型在k8s里被称为标签（label）

#查看当前node
$ kubectl get node -o wide

NAME                STATUS   ROLES                  AGE   VERSION   INTERNAL-IP     EXTERNAL-IP   OS-IMAGE         KERNEL-VERSION                 CONTAINER-RUNTIME
k8s-elasticsearch   Ready    <none>                 16h   v1.21.0   172.16.66.167   <none>        CentOS Linux 8   4.18.0-305.19.1.el8_4.x86_64   docker://20.10.9
k8s-master          Ready    control-plane,master   43h   v1.21.0   172.16.66.169   <none>        CentOS Linux 8   4.18.0-305.19.1.el8_4.x86_64   docker://20.10.9
k8s-node1           Ready    <none>                 43h   v1.21.0   172.16.66.168   <none>        CentOS Linux 8   4.18.0-305.19.1.el8_4.x86_64   docker://20.10.9
k8s-node2           Ready    <none>                 43h   v1.21.0   172.16.66.170   <none>        CentOS Linux 8   4.18.0-305.19.1.el8_4.x86_64   docker://20.10.9

#给k8s-node1 k8s-node2打上标签websvr
$ kubectl label nodes k8s-node1 k8s-node2 type=websvr

#查看type=websvr标签的node
$ kubectl get node -l type=websvr

NAME        STATUS   ROLES    AGE   VERSION
k8s-node1   Ready    <none>   43h   v1.21.0
k8s-node2   Ready    <none>   43h   v1.21.0

#以下附带标签的其他操作：
#修改标签
$ kubectl label nodes k8s-node1 k8s-node2 type=webtest --overwrite

#查看node标签
$ kubectl get nodes k8s-node1 k8s-node2 --show-labels

#删除标签
$ kubectl label nodes k8s-node1 k8s-node2 type-

修改部署yaml

$ vim websvr1.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: websvr1-deployment
spec:
  selector:
    matchLabels:
      app: websvr1
  replicas: 3
  template:
    metadata:
      labels:
        app: websvr1
    spec:
      nodeSelector:                 #选择标签为type:websvr的node部署
        type: websvr
      containers:
      - name: websvr1
        image: websvr:v1
        ports:
        - containerPort: 3000

---

apiVersion: v1
kind: Service
metadata:
  name: websvr1-service
spec:
  selector:
    app: websvr1
  ports:
  - protocol: TCP
    port: 3000
    targetPort: 3000

同理修改websvr2.yaml

$ vim websvr2.yaml

apiVersion: apps/v1
kind: Deployment
metadata:
  name: websvr2-deployment
spec:
  selector:
    matchLabels:
      app: websvr2
  replicas: 3
  template:
    metadata:
      labels:
        app: websvr2
    spec:
      nodeSelector:                    #选择标签为type:websvr的node部署
        type: websvr
      containers:
      - name: websvr2
        image: websvr:v2
        ports:
        - containerPort: 3001

---

apiVersion: v1
kind: Service
metadata:
  name: websvr2-service
spec:
  selector:
    app: websvr2
  ports:
  - protocol: TCP
    port: 3001
    targetPort: 3001

$ kubectl apply -f websvr1.yaml

$ kubectl apply -f websvr2.yaml

$ kubectl get pods -o wide

NAME                                  READY   STATUS    RESTARTS   AGE     IP               NODE        NOMINATED NODE   READINESS GATES
websvr1-deployment-67fd6cf9d4-cfstc   1/1     Running   0          2m22s   10.244.36.93     k8s-node1   <none>           <none>
websvr1-deployment-67fd6cf9d4-lx6tr   1/1     Running   0          2m22s   10.244.169.151   k8s-node2   <none>           <none>
websvr1-deployment-67fd6cf9d4-zxznp   1/1     Running   0          2m22s   10.244.169.152   k8s-node2   <none>           <none>
websvr2-deployment-67dfc4f674-dz44b   1/1     Running   0          2m10s   10.244.36.95     k8s-node1   <none>           <none>
websvr2-deployment-67dfc4f674-wjg5x   1/1     Running   0          2m10s   10.244.169.153   k8s-node2   <none>           <none>
websvr2-deployment-67dfc4f674-xdk9m   1/1     Running   0          2m10s   10.244.36.94     k8s-node1   <none>           <none>

可以看到此时两个websvr deployment已经部署在指定的node节点上

若有不正确，欢迎指出