System Operations

Policy continues work after users save, approve, or archive policies. Background services refresh discovery, generate approved configuration, publish it, and check delivery progress. Platform operators maintain these services. Authors and approvers decide policy content.

Choose an Operational Task

Choose the task that matches the required evidence or change:

• Refer to background processing to distinguish the workers, their effective cadence, and their stop controls.
• Refer to automatic recompilation to understand fresh-input cycles, compilation minors, unchanged output, and failed-cycle recovery.
• Refer to observation and restarts to inspect process health, metrics, operation history, and restart boundaries.
• Refer to backup and restore for database recovery. Do not restore data merely to clear a publication warning.

Follow an Incident From Evidence to Recovery

Complete the following steps:

1. Record the affected policy identifier, approved version, error message, and time.
2. Check API logs for the last recompilation outcome using observation and restarts. If compilation succeeded but delivery stopped, inspect the policy’s latest request in Operations.
3. Use the reported error to select recompilation troubleshooting or publication troubleshooting. Each guide identifies the administrator who can resolve the problem.
4. Repeat the failed check after the administrator corrects the cause.
5. Follow the Data or Guardrails verification guide after publication succeeds.

Deployment retry resends retained artifacts. It does not request recompilation. Policy exposes no public automatic-recompilation run-now endpoint. Do not restart the API as a forced compilation trigger.

Continue to recovery playbooks for publication or ownership incidents, and API recovery for authorized inspection and commands.