Skip to content
Code by Pawpu

Engineering Lab

Production Is On Fire

Choose incident actions and watch downtime, latency, cost, and users affected change.

Loading incident…

How the challenge works

  • You have a short clock. Each action moves downtime, users, cost, latency, and reliability.
  • Helpful actions are the ones a sane on-call would take. The score is how healthy the system is when time runs out.

Key concepts

Mitigate, then investigate

Rollback, flag-off, fail open, and rate limits are mitigations. Logs confirm you picked the right one.

Do not scale a broken artifact

More replicas of a bad deploy or a blocked thread pool make the bill worse.

Common mistakes

  • Rolling back when the signal is traffic or a vendor.
  • Touching the database because it is the most expensive knob.

Interview tips

  • State the user impact in one sentence, then the first mitigation.
  • Say what you would stop doing, not only what you would add.

Related challenges

Built by Al Beltran