The Chaos Button: Fault Injection for Mere Mortals

Software breaks.
Sometimes because of traffic spikes, network hiccups, or the cosmic will of misconfigured YAML.

But what happens when you break your app on purpose?

I added a tiny Chaos API to this engineering lab to answer that:

export async function getChaos() {
const roll = Math.random();

if (roll < 0.7) return { status: 200, message: "All good." };
if (roll < 0.9) throw new Error("Random Server Explosion");

await new Promise(r => setTimeout(r, 1500));
throw new Error("Timeout");
}

It’s silly, tiny, and brilliant.

Why Inject Chaos?

Chaos testing teaches you things that perfect days never reveal:

• Your UI should handle failure with grace
• Your retry logic should be real, not theoretical
• Your app needs backpressure, timeouts, and escape hatches
• Logs matter. A lot.

What I Learned

• Even a toy chaos endpoint reveals brittle assumptions
• UIs are rarely prepared for partial failure
• Humans react better to a helpful error message than a blank screen
• Observability is less about dashboards and more about writing good stories

Reflection

Breaking your own system is like doing push-ups with a weighted vest. It feels weird at first, but you walk away stronger.

Try pressing the chaos button before shipping next time. Your app will thank you.