OOMKilled containers in Kubernetes - Commands & Configuration

Introduction

Sometimes, containers in Kubernetes stop working because they use too much memory. This is called OOMKilled, which means the system killed the container to protect the server. Understanding why this happens helps keep your apps running smoothly.

When a container suddenly stops and shows OOMKilled status in Kubernetes.

When you want to prevent your app from using too much memory and crashing.

When you need to check if your memory limits are set correctly for your containers.

When troubleshooting why a pod restarts frequently without clear errors.

When optimizing resource use to avoid wasting server memory.

Config File - pod-memory-limit.yaml

pod-memory-limit.yaml

apiVersion: v1
kind: Pod
metadata:
  name: memory-demo
spec:
  containers:
  - name: memory-demo-container
    image: busybox
    command: ["sh", "-c", "sleep 3600"]
    resources:
      limits:
        memory: "100Mi"
      requests:
        memory: "50Mi"

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

The resources section sets memory limits and requests:

• requests.memory: The amount of memory Kubernetes reserves for the container.
• limits.memory: The maximum memory the container can use before it is killed.

If the container uses more than 100Mi of memory, Kubernetes will kill it with OOMKilled.

Commands

This command creates the pod with memory limits set. It tells Kubernetes to start the container with the specified memory rules.

Terminal

kubectl apply -f pod-memory-limit.yaml

Expected Output

pod/memory-demo created

Check the status of the pod to see if it is running or if it was killed due to memory issues.

Terminal

kubectl get pods

Expected Output

NAME           READY   STATUS    RESTARTS   AGE

memory-demo    1/1     Running   0          10s

Run a command inside the container to use more memory than the limit, causing OOMKilled. This simulates a memory overload.

Terminal

kubectl exec memory-demo -- sh -c "stress --vm 1 --vm-bytes 150M --vm-hang 0"

Expected Output

No output (command runs silently)

Check the reason why the container was terminated. It should show 'OOMKilled' if the container was killed due to memory overuse.

Terminal

kubectl get pod memory-demo -o jsonpath='{.status.containerStatuses[0].lastState.terminated.reason}'

Expected Output

OOMKilled

Get detailed information about the pod, including events that show if the container was killed because it used too much memory.

Terminal

kubectl describe pod memory-demo

Expected Output

Name:         memory-demo

Namespace:    default

Containers:

  memory-demo-container:

    Container ID:   docker://abcdef123456

    Image:          busybox

    State:          Waiting

      Reason:       CrashLoopBackOff

    Last State:     Terminated

      Reason:       OOMKilled

      Exit Code:    137

    Ready:          False

    Restart Count:  3

Events:

  Type     Reason     Age                    From               Message

  ----     ------     ----                   ----               -------

  Warning  OOMKilled  2m                     kubelet            Container memory limit exceeded

Key Concept

If a container uses more memory than its limit, Kubernetes kills it with OOMKilled to protect the system.

Common Mistakes

Not setting memory limits on containers.

Without limits, containers can use unlimited memory, causing the node to become unstable or crash.

Always set memory requests and limits in your pod specs to control resource use.

Setting memory limits too low for the app's needs.

The container will be killed frequently with OOMKilled, causing downtime.

Monitor your app's memory use and set limits high enough to avoid killing but low enough to protect the node.

Ignoring OOMKilled status and not checking pod events.

You miss the root cause of crashes and cannot fix memory issues.

Use kubectl describe and check container termination reasons to diagnose OOMKilled.

Summary

Set memory requests and limits in pod specs to control container memory use.

Use kubectl commands to create pods, check status, and diagnose OOMKilled events.

If a container exceeds its memory limit, Kubernetes kills it to protect the system.