Skip to content
hs-sql-agent
2.0.2
Docs 2.0.2
Docs Operations

Observability and Audit Operations

Configure health probes, slow-query thresholds, Prometheus, OTLP, audit retention, and signed outbound delivery.

hs-sql-agent separates runtime observability from the SQL control plane. Production operators can combine health checks, metrics, telemetry export, audit retention, and outbound alert/SIEM delivery without making the database audit table the only operational signal.

Health and slow-query monitoring

The sample environment exposes:

HEALTH_PROBE_ENABLED
HEALTH_PROBE_INTERVAL_SECONDS
HEALTH_PROBE_TIMEOUT_SECONDS
HEALTH_PROBE_MAX_CONCURRENCY
SLOW_QUERY_THRESHOLD_MS

Health probes are optional. The slow-query threshold determines when SQL activity is classified for slow-query operational visibility.

Prometheus and OTLP

Prometheus is served on a separate listener from the application API:

PROMETHEUS_ENABLED=false
PROMETHEUS_HOST=0.0.0.0
PROMETHEUS_PORT=9000

Optional OpenTelemetry export is configured with:

OTLP_ENDPOINT=
OTEL_SERVICE_NAME=hs-sql-agent

Use the Prometheus listener for scrape-based metrics and OTLP when a collector is part of the deployment topology.

Alert and SIEM delivery

The server supports optional signed outbound targets:

ALERT_WEBHOOK_URL=
ALERT_WEBHOOK_SECRET=
SIEM_WEBHOOK_URL=
SIEM_WEBHOOK_SECRET=
DELIVERY_MAX_ATTEMPTS=6
DELIVERY_MAX_CONCURRENCY=4

Leaving a target URL empty disables that integration. Persist and protect the signing secrets like any other deployment credential.

Audit retention

Retention is controlled separately from live request handling. AUDIT_RETENTION_DAYS=0 disables automatic retention processing. The current service implementation accepts retention modes Archive and Purge; Archive preserves expired records to the configured archive path before deletion.

Relevant settings include:

AUDIT_RETENTION_DAYS
AUDIT_RETENTION_MODE
AUDIT_ARCHIVE_PATH
AUDIT_FALLBACK_PATH
AUDIT_RETENTION_RUN_HOUR_UTC

The fallback path is important for operational failure handling; keep the configured storage persistent when the deployment depends on archived or fallback audit output.