Course
Observability: Seeing Production Clearly
Logs, metrics, and traces the way operators use them — Prometheus and Grafana without the wallboard theater, alerting that doesn't cry wolf, and SLOs that turn 'it feels slow' into an engineering decision.
The Three Signals
Logs for detail, metrics for shape, traces for causality — what each is actually for.
- 01Logs, Metrics, Traces: What Each Signal Is ForThree different questions: what happened, how much/how often, and where the time went — pick the signal that answers yours.20mOpen
- 02Instrumentation: Metrics Your App Should ExportThe RED method for services, the USE method for resources, and the /metrics endpoint that powers both.20mMembers
The Stack
Prometheus scraping, label discipline, and dashboards that answer questions instead of decorating.
- 01Prometheus: Pull Model, PromQL, and Label DisciplineHow scraping actually works, the five PromQL shapes you'll use daily, and the label rules that keep Prometheus alive.25mMembers
- 02Dashboards That Answer QuestionsDashboards as arguments — the service overview, the drill-down, and why red-on-green wallboards help nobody.20mMembers
Operating on Signals
Alert on symptoms not causes, error budgets, and using observability during an actual incident.
- 01Alerting Without the Wolves: Symptom Alerts That PagePage on user-facing symptoms, ticket on causes — multi-window burn rates, and the alert audit that keeps the channel honest.22mMembers
- 02SLOs and Error Budgets: The Engineering Decision ToolSLIs, SLOs, error budgets — turning 'be reliable' into a number you can trade features against.20mMembers
- 03Debugging an Incident With SignalsThe 10-minute triage flow: is it us, which signal, where's the tail — turning observability into a practiced drill.20mMembers