Cloud & AI Troubleshooting Playbook
When something breaks, start with the symptom, not the service name.
A practical troubleshooting loop
- Define the symptom — What is failing, for whom, and since when?
- Check scope — One request, one workload, one availability zone, one region, or everything?
- Check recent changes — Deployments, configuration, IAM, networking, certificates, dependencies, or model changes.
- Check the dependency chain — DNS → network → identity → compute → storage/database → application → external service.
- Use evidence — Logs, metrics, traces, health checks, audit events, and error messages.
- Test the smallest hypothesis — Change one variable at a time where possible.
- Mitigate first, investigate second — Restore service safely, then identify the root cause.
- Document the fix — Record the symptom, root cause, resolution, and prevention step.
Questions to ask yourself
- Is the problem availability, performance, correctness, security, or cost?
- Is it new or recurring?
- Did the failure begin after a deployment or configuration change?
- What does the monitoring data prove rather than suggest?
- What is the safest reversible action?
Use this playbook as the entry point, then connect it to the relevant cloud concepts, platform documentation, and service-specific guides.
