oncallsim
Join the WaitlistWaitlist Try Demo

Your first page shouldn’t be your first incident.

oncallsim puts you on call for a broken production system, with the incident already on the graphs. Find the cause, fix the code, and prove the fix under the traffic that paged you.

Try Demo Join the Waitlist

Access opens in waves. One short form.

orders-api · p95 latency p95 threshold 500 ms

How it works

Three stages, one session

  1. 1Diagnose

    Read the metrics, logs and traces in Grafana. Name the service that failed and how it failed.

    passes when the culprit and failure class are right

  2. 2Fix

    Clone the repo, patch it in your own editor, push. The sandbox rebuilds and runs the regression suite.

    passes when build OK · tests green

  3. 3Replay

    We replay the original incident traffic against your fix. The alert that paged you has to stay quiet.

    passes when the alert stays silent for the whole replay window

Every incident is timed against a par. Clear it for gold, silver or bronze.

Who it’s for

First rotation
You’re joining an on-call rotation and have never debugged production at 3 a.m.
Interview prep
You want real reps before an SRE or platform interview, not flashcards.
After the fact
You only ever read postmortems. This time you’re the one holding the pager.

Questions before your first round

What incidents can I work?

The first three:

  • Checkout is timing outAn N+1 query exhausting the database connection pool.Go
  • Worker keeps restartingA memory leak ending in OOM restarts.Python
  • Gateway is throwing 5xxA retry storm cascading across services.Go
Is this a quiz?

No. You get a running system, its Grafana, and its repository. You pick the cause from what the telemetry shows, write the fix yourself, and the grade is whether production holds.

Is the incident real?

It was recorded from a real run of the broken system and is replayed into your own Grafana when your session starts. The replay at the end is live traffic against your code.

What do I need?

A browser, git, and the editor you already use. Nothing to install from us.

Which languages?

Go and Python to start.

How long does a session take?

A session runs for up to two hours. Each incident has a par time, and beating it is how you earn gold.

What does it cost?

Pricing will be announced before public launch; the waitlist hears first.

When can I play?

Access opens in waves. The form is the queue.

What happens to my email?

It’s used to send your invite and nothing else.

The pager is going to go off eventually.

Try Demo Join the Waitlist

Access opens in waves. One short form.