Service level
This is a brief, honest statement of what IRIS aims to deliver. It is intentionally modest: IRIS is a single-box service with manual operator onboarding, not a redundant managed platform. We would rather under-promise and meet it than publish a commitment the deployment cannot honor.
Uptime — single-box, best-effort
IRIS runs on a single box. There is no high-availability or redundancy, and the single-box setup is a single point of failure (SPOF) that is explicitly accepted at this stage. Accordingly:
- Uptime is best-effort. We do not publish a numeric uptime percentage or an uptime guarantee.
- Planned maintenance, a box failure, or an upstream INAPI outage can make the service unavailable. Recovery is operator-driven and not bounded by a published recovery-time commitment.
- Data durability is covered by off-box encrypted backups; this is a durability measure, not an availability guarantee.
If your integration requires guaranteed uptime or failover, treat IRIS as a best-effort source and design your own retry/caching accordingly.
Freshness cadence
IRIS keeps its neutral corpus current by running its INAPI ingestion pipelines on a managed cron. The cadence follows how INAPI itself publishes:
| Source | INAPI publication cadence | IRIS sync |
|---|---|---|
| Sheets | Daily | Managed cron, daily |
| Estado-Diario | Monday–Friday (Santiago) | Managed cron, on publication days |
| Buscador | Continuous | Managed cron |
All cadences are stated in America/Santiago time. Because syncing is best-effort on a single
box, a given run can be late or skipped (e.g. during a box outage); the per-source freshness at
any moment is reported by GET /v1/freshness (scope insights:read) — see the
status page. Treat GET /v1/freshness, not this document, as the live truth for
how fresh the corpus is right now.
Support response
Support is best-effort, handled by the operator during normal working hours (America/Santiago):
- There is no 24/7 support and no contractually guaranteed response time.
- For issues, contact your operator with the
requestIdfrom the error envelope (it correlates your report with the server log). - Breaking-change and deprecation notices are published in the changelog with the migration window stated in the versioning policy.
Summary
| Aspect | Commitment |
|---|---|
| Uptime | Best-effort, single-box (SPOF accepted). No numeric guarantee. |
| Recovery | Operator-driven, best-effort. No published recovery-time objective. |
| Freshness | Sheets daily; Estado-Diario Mon–Fri (Santiago); via managed cron. |
| Live freshness | GET /v1/freshness is authoritative for current staleness. |
| Support | Best-effort, operator, working hours. No guaranteed response time. |