One live simulation for everyone on the internet, served by the backend. Kill a pod and the whole world sees it die. Expect chaos from strangers — that's the fun. May be slow (or down!) under heavy load, by design.
Enter the shared cluster →Your own private cluster, running 100% in your browser — same engine, zero server involvement. Nobody can touch your pods, it works at any speed, and it never goes down. Best for focused, distraction-free practice.
Launch your own sandbox →A hidden fault is injected into a Kubernetes cluster and users are complaining. Read the events, logs and charts, diagnose the root cause, apply the fix and watch it heal. How long is your streak?
Take the on-call shift →Partitions, consumer groups and the physics of lag. Create a hot key and watch one partition drown, trigger rebalancing storms, kill a broker mid-stream — all in your browser.
Start producing →A three-service chain where a slow database plus naive retries equals a self-inflicted outage. Tune timeouts, backoff and circuit breakers to ride out the storm — or cause one on purpose.
Brew a storm →One outage a day — the same hidden fault for everyone on the planet. Diagnose it, keep your streak alive, and share your result. New incident every midnight UTC.
Play today's incident →Interactive walkthroughs that drive the simulators for you — each step waits until you've watched the concept actually happen. Kubernetes, Kafka and resilience, minus the wall of text.
Start learning →