Skip to content

Monitoring API

Health endpoints for infrastructure components (PostgreSQL, Redis), real-time collector status, and the dangerous admin-only clear-all operation.

Endpoint Summary

Method Path Description
GET /api/monitoring/health PostgreSQL + Redis health (latency, Redis keys)
GET /api/monitoring/collectors Per-source status + events in last 24h
DELETE /api/monitoring/clear-all Destroy all data (admin/superadmin only)

System Health

GET /api/monitoring/health
Field Description
postgres.status up / down
postgres.latency_ms Round-trip SELECT 1 latency
redis.status up / down
redis.latency_ms PING round-trip
redis.extra.total_keys Number of keys in Redis DB (dedup windows, cache) 
{
  "postgres": { "status": "up", "latency_ms": 0.8, "extra": {} },
  "redis": { "status": "up", "latency_ms": 0.5, "extra": { "total_keys": 4123 } }
}

Kafka is not pinged here — its health is derived from consumer lag and Vector connectivity.

Collector Status

GET /api/monitoring/collectors

For each source, returns:

Field Description
name Source name
connector jira, prometheus, azure-devops, custom
status up if enabled + status == active, otherwise raw status
last_event_at Timestamp of last ingested event
events_24h Live count of events in the last 24h

Used by the frontend to render per-collector health badges.

Clear All (Danger)

DELETE /api/monitoring/clear-all

Irreversible

Requires role admin or superadmin. Deletes every application, source, event, alert, template and DLQ record.

Steps performed:

  1. Delete events, templates, dlq_events, alerts, sources, applications
  2. Invalidate Redis KPI cache (kpis_global:*)
  3. REFRESH MATERIALIZED VIEW CONCURRENTLY on OLAP views (best-effort)

Returns {"status": "ok"}.

Rate limited by clear_all_limit. This endpoint intentionally requires elevated RBAC so no authenticated user can trigger it.