01Configuration & auth
Tokens rejected after deploy
The same API token works locally and fails in production.
Trace a request through configuration loading and token validation.
35 minHardOpen incident
Debrief opens after your accepted Practice fix.
Node.js · Pack 01
Backend
reliability.
Five failures across the life of a request. Follow the evidence, fix the cause, and prove the behavior holds.
- Scope
- 5 multi-file incidents
- Level
- Advanced · Node.js / JavaScript
- Time
- About 175 min · at your pace
The investigation sequence
Start with a request, then follow resources, state, and side effects. Work in order or choose the failure you want to understand.
- 02Resource lifecycles
A connection pool that runs dry
A read API starts healthy, then queues requests and returns 5xx under load.
Follow a borrowed resource through successful and failing requests.
35 minHardOpen incidentDebrief opens after your accepted Practice fix.
- 03Cache isolation
An order from the wrong region
An order endpoint returns another region’s data after a traffic burst.
Follow an ownership boundary across a repository and its cache.
35 minHardOpen incidentDebrief opens after your accepted Practice fix.
- 04Retries & idempotency
One payment, two charges
A checkout times out. The retry succeeds, but the customer is charged twice.
Reconstruct what happened on both sides of an ambiguous timeout.
35 minHardOpen incidentDebrief opens after your accepted Practice fix.
- 05Concurrency & recovery
Two workers, one job
A redelivered queue job races with another worker and applies its effect twice.
Reason about interleaving workers and recovery after a failure.
35 minHardOpen incidentDebrief opens after your accepted Practice fix.
Hiring for this kind of work?
Choose one or two relevant incidents for a work sample, then discuss the candidate’s checks, patch, and explanation. These labs are public; candidates may have practised them.