CPU requests and limits in Kubernetes - Commands & Configuration

Introduction

When running applications in Kubernetes, you need to tell the system how much CPU your app needs and the maximum it can use. This helps the system share CPU fairly and avoid one app slowing down others.

When you want to make sure your app always gets enough CPU to run smoothly.

When you want to prevent your app from using too much CPU and affecting other apps.

When you run many apps on the same cluster and want to share CPU fairly.

When you want to avoid your app being killed because it uses too much CPU.

When you want to optimize resource costs by not over-allocating CPU.

Config File - pod-cpu.yaml

pod-cpu.yaml

apiVersion: v1
kind: Pod
metadata:
  name: cpu-demo-pod
spec:
  containers:
  - name: cpu-demo-container
    image: busybox
    command: ["sh", "-c", "while true; do echo hello; sleep 10; done"]
    resources:
      requests:
        cpu: "250m"
      limits:
        cpu: "500m"

This file creates a pod named cpu-demo-pod with one container.

The resources section sets CPU requests to 250 milliCPU (0.25 CPU) and CPU limits to 500 milliCPU (0.5 CPU).

Requests tell Kubernetes the minimum CPU guaranteed for the container.

Limits tell Kubernetes the maximum CPU the container can use.

Commands

This command creates the pod with CPU requests and limits defined in the YAML file.

Terminal

kubectl apply -f pod-cpu.yaml

Expected Output

pod/cpu-demo-pod created

This command checks that the pod is running after creation.

Terminal

kubectl get pods cpu-demo-pod

Expected Output

NAME           READY   STATUS    RESTARTS   AGE

cpu-demo-pod   1/1     Running   0          10s

This command shows detailed information about the pod, including CPU requests and limits.

Terminal

kubectl describe pod cpu-demo-pod

Expected Output

Name:         cpu-demo-pod

Namespace:    default

Node:         minikube/192.168.49.2

Start Time:   Thu, 01 Jun 2023 12:00:00 +0000

Labels:       <none>

Annotations:  <none>

Status:       Running

Containers:

  cpu-demo-container:

    Container ID:   docker://abcdef123456

    Image:          busybox

    Image ID:       docker-pullable://busybox@sha256:abcdef123456

    Port:           <none>

    Host Port:      <none>

    Command:

      sh

      -c

      while true; do echo hello; sleep 10; done

    State:          Running

      Started:      Thu, 01 Jun 2023 12:00:05 +0000

    Ready:          True

    Restart Count:  0

    Limits:

      cpu:  500m

    Requests:

      cpu:  250m

Conditions:

  Type              Status

  Initialized       True 

  Ready             True 

  ContainersReady   True 

  PodScheduled      True 

Volumes:            <none>

QoS Class:          Burstable

Node-Selectors:     <none>

Tolerations:        <none>

Events:             <none>

Key Concept

If you remember nothing else from this pattern, remember: CPU requests guarantee minimum CPU for your app, and CPU limits cap the maximum CPU it can use.

Common Mistakes

Setting CPU limits lower than CPU requests

Kubernetes will reject the pod because limits must be equal or higher than requests.

Always set CPU limits equal to or higher than CPU requests.

Not setting CPU requests at all

Without requests, Kubernetes may not reserve CPU, causing your app to get less CPU than needed.

Always set CPU requests to ensure your app gets enough CPU.

Setting very high CPU limits without reason

This can cause your app to use too much CPU and affect other apps.

Set realistic CPU limits based on your app's needs.

Summary

Create a pod YAML file with CPU requests and limits under the resources section.

Apply the YAML file with kubectl apply to create the pod.

Use kubectl get pods to check the pod status.

Use kubectl describe pod to see detailed CPU requests and limits.

CPU requests guarantee minimum CPU; CPU limits cap maximum CPU usage.