ENGINEERING WIKI STRUCTURE

===== TAXONOMY =====

Root: /engineering

/engineering/onboarding
• First day checklist
• Development environment setup
• Access provisioning (GitHub, AWS, etc.)
• Tools walkthrough (Slack, Notion, CI/CD)

/engineering/runbooks
• Deployments
  ├ Staging deploy checklist
  ├ Production deploy checklist
  └ Rollback procedure
• Incident response
  ├ High-severity incident playbook
  ├ On-call escalation matrix
  └ Post-mortem template
• On-call duties
  ├ On-call schedule
  ├ Alert routing rules
  └ Common alert responses

/engineering/architecture
• System design overview
• API specifications
• Database schema
• Deployment architecture (diagram)
• Third-party integrations

/engineering/standards
• Code review guidelines
• Commit message format
• Naming conventions
• Testing requirements (unit + integration)

/engineering/troubleshooting
• Common issues + fixes
• Performance tuning guide
• Log analysis playbook

===== TEMPLATE: RUNBOOK =====

Filename: /engineering/runbooks/deployment-production.md

PRODUCTION DEPLOYMENT

Prerequisites
• You have merge permissions on main
• You have AWS access (prod account)
• Latest code is pushed to main
• All tests pass on CI
• Staging deployment completed and verified

Step 1: Verify Pre-deployment
→ Run: npm run build
→ Run: npm run test:integration
→ Check: Are all high-priority PRs merged?

Step 2: Deploy to Production
→ Command: npm run deploy --env=prod
→ Monitor: Deployment logs in CloudWatch
→ Wait for: Green status on all health checks

Step 3: Smoke Test
→ Open prod.ourapp.com
→ Log in as test@ourapp.com
→ Perform 5 critical flows (signup, purchase, export)
→ Check database for new records

Step 4: Monitor Metrics (next 10 minutes)
→ CPU usage: Should stay < 60%
→ Error rate: Should stay < 0.1%
→ Response time p95: Should stay < 300ms
→ If anomaly detected → Go to ROLLBACK section

Step 5: Notify
→ Post in #deployments: Production deployed by Name at Time. Version: git tag

ROLLBACK PROCEDURE (if anything goes wrong)
→ Command: npm run rollback --env=prod --to-version=previous-tag
→ Verify health checks (same as Step 3)
→ Post in #incidents: Rolled back to Version due to Reason

Estimated time: 15 minutes
On-call escalation: If unclear, page DevOps lead

===== TEMPLATE: INCIDENT RESPONSE =====

Incident severity: HIGH

Detection: Error rate > 1% for 2 min

Immediate actions (first 5 min):
1. Page on-call engineer
2. Post in #incidents: Production error spike detected
3. Gather: Error logs, affected users, root cause hypothesis

Investigation (next 10 min):
• Query logs: SELECT * FROM errors WHERE severity = HIGH ORDER BY timestamp DESC LIMIT 100
• Check metrics: CPU, memory, database connections
• Review recent deployments: Was anything shipped in last hour?

Resolution (parallel to investigation):
• If deployment caused: Rollback (see runbook)
• If database issue: Trigger failover (see runbook)
• If third-party API: Switch to backup provider

Post-incident:
→ Schedule post-mortem within 24 hours
→ Document: Timeline, root cause, prevention step

===== SEARCHABILITY =====

Each page tagged with: Category, Severity, Time-to-complete
Example: Deployment checklist tagged Deployment Low-risk 15 min

Search index updated daily
Common searches alias to fastest answer (e.g., how do I deploy → /runbooks/deployment-production)