Configuring Specified Node Scheduling (nodeSelector)

Prerequisites

A custom label has been added to the target node so that workload pods can be scheduled based on the node label. For details, see Adding or Deleting a Node Label.

Creating a Workload Scheduled to a Specified Node

1. Use kubectl to access the cluster. For details, see Accessing a Cluster Using kubectl.
2. Create a YAML file named nginx.yaml. The file name can be customized.
   Configure nodeSelector for the workload. For example, if the key is deploy_qa and the value is true, the pod will be scheduled to the node with the deploy_qa=true label. Example:

   apiVersion: apps/v1
   kind: Deployment
   metadata:
     name: nginx
   spec:
     replicas: 3
     selector:
       matchLabels:
         app: nginx
     template:
       metadata:
         labels:
           app: nginx
       spec:
         nodeSelector: 
           deploy_qa: "true"
         containers:
         - image: nginx:latest
           imagePullPolicy: IfNotPresent
           name: nginx
         imagePullSecrets:
         - name: default-secret

3. Create a workload.

   kubectl apply -f nginx.yaml

4. Verify that all pods run on the target node.

   kubectl get pod -o wide

   In the following command output, node 192.168.0.103 is labeled with deploy_qa=true:

   NAME                     READY   STATUS    RESTARTS   AGE     IP              NODE                                   NOMINATED NODE   READINESS GATES
   nginx-66859f4f48-xgp2g   1/1     Running   0          6h57m   172.16.3.0   192.168.0.103   <none>           <none>
   nginx-66859f4f48-t9gqj   1/1     Running   0          6h57m   172.16.3.1   192.168.0.103   <none>           <none>
   nginx-66859f4f48-2grhq   1/1     Running   0          6h57m   172.16.3.2   192.168.0.103   <none>           <none>

Helpful Links

• Configuring Node Affinity Scheduling (nodeAffinity)
• Configuring Workload Affinity or Anti-affinity Scheduling (podAffinity or podAntiAffinity)
• If a pod is not scheduled to the specified node, rectify the fault by referring to Why Cannot a Pod Be Scheduled to a Node?