Troubleshooting runbooks
Choose the symptom you can observe. Each runbook starts with checks an administrator can make in Tutela, then describes the evidence and access a deployment operator needs. If you are responding to an active incident, record when it began before changing settings.
Requests fail or take too long
| Symptom | Start here |
|---|---|
| Tutela cannot be reached or a service is unhealthy | Service unavailable |
| Requests fail repeatedly | High error rate |
| Requests complete slowly | Slow requests |
| Inspection cannot complete or traffic is unexpectedly blocked | Inspection unavailable or blocked |
| Traffic continues while an inspection dependency is unavailable | Inspection dependency unavailable |
Activity or usage does not match expectations
- Missing activity events: find a request that is absent from Activity.
- Delayed activity processing: investigate events that arrive late.
- AI usage and cost problems: distinguish missing usage, missing prices and incorrect totals.
Infrastructure alerts
- Low disk space
- High CPU utilization
- High memory utilization
- Low cache hit rate
- Cache dependency errors
- Cache dependency latency
Routine operations and follow-up
Use Daily checks and routine maintenance between incidents. After recovery, use Service reliability review to record the customer impact and prioritize corrective work.
If you cannot access your deployment, contact support@h2htech.com. Include your deployment address, version if known, the time and timezone, and the visible error. Send credentials and customer content through neither email nor screenshots.