Troubleshooting Lab Guides¶
Reproducible, hypothesis-driven experiments that recreate real Azure Monitor failure modes end-to-end, then prove the fix with collected evidence. Where the Playbooks tell you how to diagnose a live incident, lab guides let you reproduce the failure in a controlled environment so you can build intuition before you face it in production.
How Lab Guides Differ from Playbooks¶
| Aspect | Playbooks | Lab Guides |
|---|---|---|
| Goal | Diagnose and resolve a live incident | Reproduce a failure to learn its signature |
| Starting point | An active symptom | A clean environment you deliberately break |
| Outcome | Restored service | A confirmed hypothesis backed by evidence |
| Reusability | Reference during an incident | Run as a self-contained exercise |
Lab Methodology¶
Every lab guide follows the scientific-method skeleton so the result is reproducible and falsifiable, not anecdotal:
- Background — the question the lab investigates and the environment it needs.
- Hypothesis — the expected cause and an "IF … THEN …" prediction.
- Runbook — the steps to deploy the environment, trigger the failure, and apply the fix.
- Experiment Log — the observed evidence, tagged for strength (
[Observed],[Measured],[Inferred]), including a Falsification step that proves the fix works and the original theory was correct. - Verification — the KQL queries and portal evidence that confirm the outcome.
Available Labs¶
Each published lab is added as a file under docs/troubleshooting/lab-guides/ and as a nav entry in mkdocs.yml. Live evidence capture status is called out explicitly so documentation-first labs do not imply fabricated proof.
| Lab | Failure Mode | Status |
|---|---|---|
| AMA Heartbeat Loss | DCR association deletion makes Heartbeat stale even though the VM and AMA extension still exist | Documentation published; pending live capture |
| Application Insights Telemetry Gap | Invalid APPLICATIONINSIGHTS_CONNECTION_STRING stops fresh AppRequests from arriving until the original value is restored | Documentation published; pending live capture |
See Also¶
- Playbooks
- Troubleshooting Architecture Overview
- Troubleshooting Mental Model
- Decision Tree
- Evidence Map
- KQL Query Packs