Production: Forensics, Observability, and the Frontier
Assessment contract
75 minutes. Part A (code): StatsDigest against an unseen production dump; watchdog vs three synthetic state-event failure schedules plus one real far-peer-close kill (state-based window, per m6-l5); a WebCodecs recovery-pipeline variant. Part B (scenarios): four fresh incident bundles (one is 'everything is actually fine' — overdiagnosis is penalized); an observability design review with seeded metric fallacies to catch. Part C (concepts): metric design, WebCodecs/WebTransport boundaries, when-not-WebRTC, escalation ladders.
The examination rewards working code and discriminating evidence. Each assertion gives local feedback; the scorecard combines the code problems with the concept review. Reference answers remain available after a serious attempt.
Join the stats graph. Follow identifiers from RTP through codec, transport, pair, and candidates.
Lab 1
Join the stats graph
Graded in the browser against 1 assertion; the editor and harness require JavaScript.
Diagnose the incident bundle. Return one cause and one decisive evidence key rather than a shotgun list.
Lab 2
Diagnose the incident bundle
Graded in the browser against 2 assertions; the editor and harness require JavaScript.
Compute a production percentile. Use nearest-rank ordering over the raw segment samples.
Lab 3
Compute a production percentile
Graded in the browser against 1 assertion; the editor and harness require JavaScript.
Choose the transport boundary. Distinguish integrated peer media from massive broadcast and custom client-server media.
Lab 4
Choose the transport boundary
Graded in the browser against 3 assertions; the editor and harness require JavaScript.
Check your understanding
Auto-graded check
5 auto-graded questions with an explanation for every wrong answer. Requires JavaScript. (m8-exam-quiz)
Examination score
The scorecard totals this page’s graded work as you go. It requires JavaScript.
WebRTC: From First Principles · Module 8 Examination