DeanKelly751 Runbooks
Purpose
Create runbooks that enable on-call engineers to resolve incidents quickly under stress.
Key Principle
Written for someone who is stressed, at 3am, and may not be the author. Be explicit, sequential, and copy-pasteable.
Structure
# [Alert/Incident Name]
## Severity
[P1/P2/P3] — [Impact description]
## Symptoms
- What the alert says
- What the user experiences
## Quick Diagnosis
1. Check [specific dashboard link]
2. Run: `command to check status`
3. Look for [specific log pattern]
## Common Causes and Fixes
### Cause A: [Description]
**Diagnosis**: [How to confirm this is the cause]
**Fix**:
1. Step-by-step commands
2. Verification that fix worked
### Cause B: [Description]
**Diagnosis**: [How to confirm]
**Fix**: [Steps]
## Escalation
- If not resolved in [time]: escalate to [team/person]
- Contact: [on-call channel/page]
## Post-Incident
- [ ] Update incident timeline
- [ ] Create follow-up ticket for root cause
Writing Principles
- Commands should be copy-pasteable (include full paths, specific values)
- Link to dashboards directly (not "check the dashboard")
- Include expected output of diagnostic commands
- Time-bound each step ("if not resolved in 5 minutes, try next cause")
Related Skills
observability/execution-metrics-alerting— alerts that trigger runbooksreviewers/review-production-readiness— runbooks as readiness criterion