Node selectors for simple scheduling in Kubernetes - Commands & Configuration

Introduction

Sometimes you want your app to run only on certain computers in your cluster. Node selectors help you pick those computers by matching labels you set on them.

When you want a pod to run only on nodes with special hardware like GPUs.

When you want to keep certain apps on nodes in a specific data center or zone.

When you want to separate workloads by node type, like fast SSD nodes versus regular nodes.

When you want to avoid running some pods on nodes reserved for system tasks.

When you want to test an app on a specific node before wider rollout.

Config File - pod.yaml

pod.yaml

apiVersion: v1
kind: Pod
metadata:
  name: example-pod
spec:
  containers:
  - name: example-container
    image: nginx
  nodeSelector:
    disktype: ssd

This file creates a pod named example-pod running an nginx container.

The nodeSelector section tells Kubernetes to schedule this pod only on nodes labeled with disktype=ssd.

Commands

This command adds the label disktype=ssd to the node named node1. This label is used by the pod's node selector to find a matching node.

Terminal

kubectl label nodes node1 disktype=ssd

Expected Output

node/node1 labeled

This command creates the pod defined in pod.yaml. Kubernetes will try to schedule it on a node with the label disktype=ssd.

Terminal

kubectl apply -f pod.yaml

Expected Output

pod/example-pod created

This command shows the pods with extra details including the node they are running on. You can verify the pod is on the correct node.

Terminal

kubectl get pods -o wide

Expected Output

NAME          READY   STATUS    RESTARTS   AGE   IP           NODE    NOMINATED NODE   READINESS GATES

example-pod   1/1     Running   0          10s   10.244.1.5   node1   <none>           <none>

→

-o wide - Shows detailed information including the node name

This command shows detailed information about the pod, including the node selector and the node where it is scheduled.

Terminal

kubectl describe pod example-pod

Expected Output

Name:         example-pod

Namespace:    default

Node:         node1/192.168.1.10

Start Time:   Thu, 01 Jan 2024 12:00:00 +0000

Labels:       <none>

Annotations:  <none>

Status:       Running

IP:           10.244.1.5

Containers:

  example-container:

    Image:        nginx

    State:        Running

    Ready:        True

    Restart Count: 0

Conditions:

  Type           Status

  Initialized    True 

  Ready          True 

  ContainersReady True 

  PodScheduled   True 

Node-Selectors:  disktype=ssd

Key Concept

If you remember nothing else from this pattern, remember: node selectors let you tell Kubernetes to run pods only on nodes with matching labels.

Common Mistakes

Not labeling any node with the key-value used in nodeSelector.

Kubernetes cannot find a matching node, so the pod stays pending and never runs.

Always label at least one node with the exact key and value used in the pod's nodeSelector before creating the pod.

Using a different label key or value in the pod than what is on the node.

The pod will not match any node and remain unscheduled.

Double-check that the label key and value in the pod's nodeSelector exactly match the node's label.

Forgetting to apply the pod manifest after labeling nodes.

The pod won't be created or scheduled, so no workload runs.

Run 'kubectl apply -f pod.yaml' after labeling nodes to create and schedule the pod.

Summary

Label nodes with key-value pairs to identify their characteristics.

Use nodeSelector in pod specs to require pods run only on nodes with matching labels.

Apply the pod manifest and verify the pod runs on the correct node using kubectl commands.