Monitoring & Alerting

Critical

• 10 services monitored in SERVICE_REGISTRY (infrastructure: 3, API/integrations: 6, database: 1). n8n and Ollama are not in the registry (retired; nothing probes them; seedDefaultRules() has no rules for them).
• Alert cooldowns are per-rule — do not fire the same alert within its cooldown period.
• Auto-resolve: When a service recovers, alerts are automatically resolved.
• Health check results persist to service_health — used for uptime calculations.
• Old records cleaned after 30 days via cleanupOldRecords().
• Admin-only: Detailed health data requires admin session.

Architecture