Skip to main content

Documentation Index

Fetch the complete documentation index at: https://developers.safarapi.com/llms.txt

Use this file to discover all available pages before exploring further.

status.safarapi.com

Real-time uptime, latency, and incident history for every SafarAPI component.

Components monitored

ComponentEndpointSLA target
Core APIapi.safarapi.com/v1/*99.5 % monthly
Authenticationapi.safarapi.com/v1/auth/*99.5 % monthly
Consoleconsole.safarapi.com99.0 % monthly
Documentationdevelopers.safarapi.combest-effort
Webhooks delivery(queue depth)< 5 min p95

Subscribing to notifications

1

Visit status.safarapi.com

The public status page is hosted independently of the API infrastructure, so it remains reachable even during a major incident.
2

Click 'Subscribe'

Choose email, SMS, or webhook. Webhooks are recommended for automated alerting integration in your bank’s monitoring stack (PagerDuty, Opsgenie, Slack).
3

Select components

Subscribe only to the components your integration depends on. Banks consuming /quotes + /bookings typically subscribe to Core API + Authentication.

Incident communication

For incidents impacting active partners:
  • First update within 15 minutes of detection (acknowledged + impact assessment).
  • Resolution update at most every 30 minutes for active incidents.
  • Postmortem within 5 business days for any incident classified Major or Critical.
For security‑specific incidents (data breach, suspected compromise), the 72‑hour GDPR / Loi 09‑08 notification policy applies in addition to status‑page updates.

SLA credits

If monthly uptime falls below 99.5 %, SafarAPI credits 5 % of your monthly invoice for every 0.1 % below target. Credits are applied automatically on the following month’s settlement — no claim required. Example: 98.8 % uptime in May = (99.5 − 98.8) / 0.1 × 5 % = 35 % credit of May invoice on June settlement.

How we measure

Uptime is measured by external probes (UptimeRobot + Pingdom redundancy) hitting GET /v1/health from at least 3 geographically distinct points, every minute. A check is “down” only if 2 of 3 probes fail in the same 60-second window. This eliminates single-probe false positives. Latency p95 is sampled server-side from the audit log (partner_api_audit_log.latency_ms), aggregated per 5-minute window, computed from the 95th percentile of all customer-facing requests (excluding /health and admin endpoints).