Your first outage shouldn't be a real one.
Run a broken production system on your laptop. Find the root cause, ship the fix, and keep leadership calm while the clock runs.
0 · THE PROBLEM
Nobody practises outages. They just have them.
You learn on the job.
Your first real incident is also your first practice run, with customers watching.
Reading isn't doing.
Postmortems and courses tell you what happened. They don't build the muscle.
Agents write the code. You own the outage.
As more code gets generated, the skill that matters is judgment: finding the root cause, reviewing what changed, and knowing which tool to reach for.
Half of an outage is talking.
Escalations, an exec asking for an ETA, a support team that needs an answer. Communicating under pressure needs practice too.
Build the muscle on simulations curated by experts. Real broken systems, real escalations, and real metrics that decide when you're done.
1 · WHAT IT FEELS LIKE
It's 14:02. Checkout is down.
A live replay: the graph goes red, the messages start, you fix it, it goes green.
checkout-api · error rate
#incident
2 · HOW IT WORKS
Three steps. One of them is hard.
Pick a problem
Each one is a small, realistic system with something quietly wrong.
Run it locally
One docker compose up starts the system, its dashboards and the clock.
Fix it
We watch the real metrics. When they recover, you're solved.
3 · WHAT YOU GET
Everything a real incident has, except the consequences.
Stakeholders who won't wait
Support, managers and leadership escalate mid-incident. Practise updates that keep them calm.
Checked by metrics, not by you
No "I think it's fixed". The system's own Prometheus metrics decide.
A time on the board
Every solve is timed. Climb the leaderboard and show it on your public profile.
Hints when you're stuck
Ordered hints, so you learn instead of quitting.
Write your own
Seen a nasty one at work? Turn it into a problem for everyone else.
Your laptop, your tools
No browser sandbox. Use the terminal, dashboards and debugger you'd use for real, and learn when each one helps.
Be the one who fixes it.
Free to start. Your first incident is ten minutes away.