LitmusChaos — Cloud-Native Chaos Engineering for Kubernetes

What it is

LitmusChaos is a CNCF incubating project that provides a framework for practicing chaos engineering on Kubernetes. It ships with a library of 50+ pre-built chaos experiments -- pod kill, network latency, CPU hog, disk fill, and more -- that you inject into running workloads to verify they handle failures gracefully.

The tool targets SREs, platform engineers, and DevOps teams who manage Kubernetes clusters in production. If you need to prove your services survive node failures, network partitions, or resource exhaustion, LitmusChaos gives you repeatable experiments with observable outcomes.

How it saves time or tokens

Manually testing failure scenarios requires writing custom scripts, coordinating team members, and hoping you covered the right cases. LitmusChaos replaces that ad-hoc process with a catalog of pre-built experiments and a ChaosCenter dashboard. You define a chaos scenario once, schedule it, and get automated reports. This turns what was a multi-day manual exercise into a repeatable pipeline step.

How to use

1. Install LitmusChaos via Helm into your Kubernetes cluster:

helm repo add litmuschaos https://litmuschaos.github.io/litmus-helm
helm install litmus litmuschaos/litmus \
  --namespace litmus --create-namespace

1. Access the ChaosCenter dashboard to browse available experiments and create chaos scenarios.

1. Define a ChaosEngine resource targeting your application namespace and select experiments (pod-delete, network-loss, etc.).

1. Run the experiment and observe results in the ChaosCenter UI or via Kubernetes events.

Example

apiVersion: litmuschaos.io/v1alpha1
kind: ChaosEngine
metadata:
  name: nginx-chaos
  namespace: default
spec:
  appinfo:
    appns: default
    applabel: 'app=nginx'
    appkind: deployment
  chaosServiceAccount: litmus-admin
  experiments:
    - name: pod-delete
      spec:
        components:
          env:
            - name: TOTAL_CHAOS_DURATION
              value: '30'
            - name: CHAOS_INTERVAL
              value: '10'

This deletes nginx pods every 10 seconds for 30 seconds, letting you observe how your deployment recovers.

Related on TokRepo

• DevOps Tools -- Browse automation and infrastructure tools for CI/CD pipelines
• Self-Hosted Solutions -- Explore self-hosted platforms for infrastructure management

Common pitfalls

• Running chaos experiments on production without proper blast radius limits can cause real outages. Always set namespace selectors and duration caps.
• Skipping the steady-state hypothesis means you cannot measure whether the experiment actually proved resilience.
• Forgetting RBAC permissions for the chaos service account leads to silent experiment failures with no useful feedback.