> ## Documentation Index
> Fetch the complete documentation index at: https://docs.telzino.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Delivery, retries and health

> What happens when your receiver is slow or down

* Delivery is usually within seconds of the event being recorded. A few events are held briefly before their message is built, so their details are complete:
  * about 60 seconds: `partner.created`, `account.signed_up`;
  * about 90 seconds: `agent.created`, `account.first_agent_created`, `org.first_agent_created`;
  * about 2 minutes: `account.first_call_received`, `account.first_real_call_received`, `org.first_call_received`, `org.first_real_call_received`, `agent.first_call_received`.
* Telzino is alerted internally if delivery lags more than 5 minutes.
* Any non-`2xx`, timeout (10 s) or connection error is retried with exponential backoff (1 minute doubling to 1 hour, with jitter) for **24 hours**. `429` and `503` with `Retry-After` wait as long as you ask, between 1 minute and 1 hour.
* Redirects are not followed. URLs must be `https` on a public host.
* Messages can arrive out of order; use `occurred_at`.
* [Monitoring events](https://docs.telzino.com/event-webhooks/events.md#monitoring) are sent only to the endpoint whose event settings they met, not to every endpoint that would otherwise receive them.

## Endpoint health

| State | Meaning |
| - | - |
| Healthy | Last delivery succeeded. |
| Failing | Deliveries are failing. Telzino sends one probe at a time until one succeeds, so a down receiver is not flooded. After 1 hour the endpoint's notify email gets a warning. |
| Disabled automatically | Failing continuously for 24 hours. New events are **held**, not dropped (kept for the retention period, 30 days by default). The notify email is told, and a `webhook.disabled` event goes to endpoints subscribed to it at that level, and at higher levels that include descendants. The disabled endpoint itself never receives it. |
| Disabled | Switched off by a person. New events are not queued; waiting deliveries are held. |

A short outage never disables an endpoint. When you enable an endpoint that has held deliveries, choose **resume** (send them) or **discard**.

Delivery logs are kept for the retention period Telzino sets for the platform, 30 days by default.


This documentation is built and hosted on [Mintlify](https://mintlify.com), a developer documentation platform.