# HTTP 5xx Spike Runbook

## Overview
This runbook covers the response procedure for HTTP 5xx error rate spikes.

## Detection
- Alert triggers when 5xx rate > 5% for 5 minutes
- Check Grafana dashboard: HTTP Error Rate

## Diagnosis Steps
1. Check recent deployments
2. Review application logs for stack traces
3. Check database connection pool utilization
4. Verify downstream service health

## Mitigation
1. If bad deploy: rollback immediately
2. Restart affected pods
3. Scale up if load-related

## Prevention Checklist
- [ ] Run smoke tests before promoting to production
- [ ] Set up canary deployments
- [ ] Monitor error budgets