Lab: build an observability stack
Run Prometheus, Grafana, Loki, Tempo and an OpenTelemetry Collector with Docker Compose, instrument a fake LLM server (no GPU needed), generate load, build a dashboard, write burn-rate alerts, follow a slow request from metric to trace to log, and run three failure drills.
75 min·Verified 2026-11Hands-on labPro+100 XP
Goal: stand up a complete observability stack locally, wire a (simulated) LLM service into it, and practise the workflow from the last four lessons: notice a problem in metrics, locate it in a trace, explain it in logs. No GPU is required.
You need: Docker with Compose, Python 3.11+, and about 3 GB of free RAM. Time: about 75 minutes.
Pro content
Keep reading with Inference Academy Pro
Unlock this full lab, plus every advanced lesson, hands-on lab, the larger playground models and completion certificates.