Incident drills · simulated
You're on call.
Two made-up incidents, one on each platform I run. Pick one, and the clock only moves when you do.
It is 02:14 and the pager goes off: readings from some of the platform's buildings have stopped arriving fresh.
Prefer a terminal?
How it works
- A page fires. Every look and every action costs simulated minutes, and the error budget burns while you work.
- The Grafana dashboards, the runbook and its guardrail are the platform's real ones. The freshness objective behind the page is a proposal, not yet agreed, and the incident is invented for practice.
- When it is over you get your own postmortem: timeline, time to recover, budget spent and what went well.
It is 10:42 and the pager goes off: facility managers are asking the document assistant about their buildings, and its answers have slowed to a crawl.
Prefer a terminal?
How it works
- A page fires. Every look and every action costs simulated minutes, and the error budget burns while you work.
- The services, traces and dashboards are the platform's, as this site describes them. The objective, the alert, the runbook and the incident are invented for practice.
- When it is over you get your own postmortem: timeline, time to recover, budget spent and what went well.