Course
Observability with Prometheus and Grafana
Know a service is in trouble before the customer calls: metrics, logs, traces, dashboards, and alerts that don't wake anyone up for nothing.
Intermediate 7 lessons
Lessons
- 1 Introduction Distinguish monitoring from observability, and learn which specific question each of the three signals answers: metrics, logs, and traces.
- 2 Prometheus Understand the time series, the four metric types, the role of labels, and why cardinality takes down Prometheus servers.
- 3 Instrumenting Expose a clean metrics endpoint, measure the four golden signals, name your metrics according to convention, and know what instrumentation really costs.
- 4 PromQL Write correct queries: selectors, rate, aggregations, percentiles from a histogram, and the pitfalls that produce wrong numbers without any error.
- 5 Dashboards Build a dashboard that answers one specific question, parameterize it with variables, version it, and understand why most of them end up abandoned.
- 6 Alerts Alert on symptoms rather than causes, calibrate thresholds and durations, rely on an error budget, and cut the noise that destroys on-call.
- 7 Logs and Traces You'll learn how to structure your logs around a correlation ID, decide when a distributed trace pays off, and connect an alert to the request that triggered it.