root cause
The underlying reason a failure happened, as opposed to the trigger that set it off.
In more detail
Incidents usually need several conditions to align, so a single root cause can end inquiry too early. Look for contributing conditions and fix more than one path to harm.
Example
Refused, timeout and reset are observations, not root causes. A good incident note pairs each observation with the next narrow test, and changes one thing at a time. Keep evidence so the next engineer can see where the path stopped.Related terms
- contributing factor — A condition that helped an incident happen or made it worse, without being its only cause.
- postmortem — A written review after an incident that explains impact, timeline, contributing conditions and the follow-up actions that will change the system.
- diagnostic ladder — The course's ordered checklist for a failing service: is it up, is it named, is it routed, is it listening, is it answering, is traffic arriving?
- vantage point — The place a test is run from, such as the server itself, another host on its subnet or a client across the internet.
Taught in
- The Blameless Postmortem — DevOps & Site Reliability Engineering
- Writing A Postmortem: The SRE Appendix — DevOps & Site Reliability Engineering
- What Testing Is For, And What Debugging Is — ISTQB Certified Tester Foundation Level (CTFL v4.0) Exam Prep
- Refused, Timed Out, Reset: Reading A Failure — Linux Fundamentals & Systems Administration
- The Diagnostic Ladder: Is It Up, Is It Named, Is It Routed — Linux Fundamentals & Systems Administration
References
- Google SRE book: Example Postmortem (appendix)
- Google SRE workbook: Postmortem Culture
- ISTQB Glossary: root cause
- ISTQB CTFL Syllabus v4.0.1, section 1.2.3
- Google SRE book: Postmortem Culture
- Google SRE book: Effective Troubleshooting