Chapter D · 4 of 4~9 min

On-call lab

**Page, acknowledge, escalate** — on-call protects users without burning out engineers.

Reviewed July 2026 · interactive lab

Paths & resources for this lesson

1Try it yourself

Playground

On-call lab

Page for user impact, acknowledge noise, escalate when comms needed.

Sev-1: chat API 50% error rate

2Read & reflect

Recap

#What you just did

You triaged Sev-1 pages, flapping warnings, and vendor outages needing escalation.

Teach

#Pattern

sev-1: page immediately + start runbook
sev-2: ack within 15m + investigate
vendor outage: status page + customer comms template
rotation: max 7 consecutive days, handoff doc required

Pair pagers with runbooks — never page without a first step to try.

Use it

#When you'd use this

  • Production chat or RAG APIs with real users
  • Capstone projects you host publicly
  • Team projects before demo day traffic

Watch out

#Watch out

Alert fatigue causes ignored pages — tune thresholds and use multi-window burn alerts.

Try next

#Try this next

Write your capstone on-call handoff: dashboards, rollback link, escalation phone tree.

3Spark check

Continue learning · glossary & guides