Unclassified
Runbooks
Runbooks are reactive operational procedures for incidents, failures, rollback, and recovery.
Use them when something is broken or when you need a tightly scoped operational response.
Available Runbooks
| Runbook | Use when |
|---|---|
| incident-response.md | Customer-facing outage, major degradation, or unclear platform incident |
| deployment-and-rollback.md | A release fails, a deploy regresses behavior, or you need a safe rollback path |
| connector-oauth-troubleshooting.md | OAuth authorization or callback flows fail for a connector |
| sync-failure-and-dlq-redrive.md | Connector sync, pipeline, or queue processing is failing and you may need redrive |
| data-recovery.md | Data was corrupted, lost, or written incorrectly and recovery planning is needed |
Operating Rules
- Prefer containment first, then diagnosis, then recovery
- Preserve auditability: keep timestamps, correlation IDs, tenant IDs, and deployment versions
- Avoid bulk replay or rollback without understanding blast radius
- When in doubt, protect the Spine and stop new writes before attempting recovery