Monitoring and alerting

DenialOS exposes application health and records structured operational events.

Health

GET /api/health reports database and queue checks together with worker liveness, so a stopped worker is visible from the same endpoint as a degraded database.

Operational events

The worker reports its own failures, including a schedule that has stopped advancing (schedule_stalled). Security-relevant events — refused sign-in attempts and rejected requests — are recorded in the security event stream rather than left in unstructured logs.

What monitoring covers

• API health;
• database health;
• Redis/queue health;
• worker liveness and schedule advancement;
• worker failures and dead letters;
• storage/scanner failures;
• authentication/security failures;
• reconciliation exceptions;
• latency and error rates.

Application logs carry correlation/request identifiers and operational metadata, not raw PHI.

Boundary

Alert delivery, retention, and escalation remain deployment responsibilities.