incident-timeline-creation

**Server Logs:**

SKILL.md

GitHub

Incident Timeline Creation (Verbose)

Timeline Reconstruction Process

Step 1: Gather Data Sources

Server Logs:

2026-04-10 14:00:05 [ERROR] Connection pool exhausted (available: 0/350)
2026-04-10 14:00:07 [ERROR] Query timeout after 5000ms
2026-04-10 14:00:08 [ERROR] Database unavailable

Metrics:

  • CPU: jumped from 30% to 90% at 14:00:05
  • Memory: stable
  • Database connections: 0 → 350 at 14:00:05
  • Request latency: 50ms → 5000ms at 14:00:05

Chat History:

14:00:15 @oncall: "Hey, alerts coming in"
14:00:30 @oncall: "Restarting database"
14:05:00 @incident-commander: "Incident declared"

Metrics Dashboard:

  • Alert fire time: 14:00:07
  • Error rate spike: 14:00:05
  • Recovery time: 14:07:30

Step 2: Establish Definitive Timeline

Use server logs as source of truth (most accurate):

14:00:00 - [OBSERVATION] Deployment completed (from deployment system)
14:00:05 - [ROOT CAUSE] New query deployed with infinite loop
          (From logs: connection pool exhaustion starts)
14:00:07 - [METRIC CHANGE] Database unavailable
          (From metrics: error rate → 100%)
14:00:08 - [DETECTION] Alert fires
          (From monitoring: automated alert triggered)
14:00:15 - [HUMAN RESPONSE] On-call acknowledges
          (From chat: message timestamp)
14:00:30 - [MITIGATION] Kill infinite query / restart database
          (From chat: "Restarting database")
14:05:00 - [INCIDENT DECLARED] Severity: SEV1
          (From Slack: incident created)
14:07:30 - [RECOVERY] Service restored
          (From metrics: error rate drops to <0.1%)
14:10:00 - [RESOLUTION] All-clear sent
          (From Slack: "Service restored")

Step 3: Calculate Critical Metrics

Total Incident Duration:
  Start: 14:00:00 (root cause begins)
  End: 14:07:30 (service restored)
  Duration: 7.5 minutes

Customer Impact Duration:
  Start: 14:00:07 (first error)
  End: 14:07:30 (service restored)
  Duration: 7.38 minutes

Detection Lag:
  Root cause: 14:00:05
  Alert fired: 14:00:08
  Lag: 3 seconds (excellent)

Response Lag:
  Alert: 14:00:08
  Human starts: 14:00:15
  Lag: 7 seconds (excellent)

Remediation Time:
  Start: 14:00:15
  Complete: 14:07:30
  Duration: 7.25 minutes

Step 4: Identify Contributing Factors

What made this worse?
- No timeout on database queries
  → Connections never released
  → Pool exhausted in seconds
  
- No monitoring on connection pool
  → Didn't detect pool > 80%
  → Hit limit before alert

- No circuit breaker
  → Kept hammering database after pool exhausted
  → Made situation worse

Step 5: Document Final Timeline

# Incident #12345: Database Outage April 10, 2026

## Timeline

| Time | Event | Source | Impact |
|------|-------|--------|--------|
| 14:00:00 | Deployment: v1.5.0 | CD System | Root cause deployed |
| 14:00:05 | Connection pool exhausted | Logs | Service degraded |
| 14:00:07 | First 500 errors | Metrics | Users affected |
| 14:00:08 | Alert fires | Monitoring | Detection |
| 14:00:15 | On-call acknowledges | Chat | Response |
| 14:00:30 | Restart database | Chat | Remediation |
| 14:07:30 | Service restored | Metrics | Recovery |
| 14:10:00 | All-clear sent | Slack | Communication |

## Key Metrics

- Customer Impact: 7.38 minutes
- Detection Lag: 3 seconds
- Response Lag: 7 seconds
- Remediation Time: 7.25 minutes
- Total Incident Duration: 7.5 minutes

Accuracy Checklist

  • Sourced timestamps from logs, not memory
  • Cross-referenced with multiple sources
  • Verified with participants
  • Calculated lag times correctly
  • Identified all critical transition points
  • Timeline reflects both system and human actions
  • Root cause time vs detection time clear

Common Timeline Mistakes

❌ Using participant memory → Inaccurate (+/- 10 minutes)
✅ Use server logs/metrics → Accurate to seconds

❌ Mixing different time zones → Off by hours
✅ Convert all to UTC/standard timezone

❌ Starting from alert, not root cause
✅ Start from actual root cause (may be before alert)

❌ Only including human actions → Miss system behavior
✅ Include metric changes, errors, and human actions