# Retries and turning off

When Mestra tries again after a failed delivery, and when it switches an endpoint off.

## What counts as delivered

Any `2xx` answer means the event is delivered. Anything else is a failure: another status, a timeout after 10 seconds, a connection error, or a redirect. Mestra doesn't follow redirects.

## Retries

After a failure, Mestra tries again after about:

| Attempt    | Wait after the previous one |
| ---------- | --------------------------- |
| 2          | 1 minute                    |
| 3          | 5 minutes                   |
| 4          | 30 minutes                  |
| 5          | 2 hours                     |
| 6          | 6 hours                     |
| After that | 12 hours                    |

It stops three days after the first attempt. The waits vary a little, so don't rely on exact times.

Every attempt carries the same event and the same `id`, with a fresh [signature](https://withmestra.com/docs/webhooks/signatures/). Use the id to [ignore repeats](https://withmestra.com/docs/webhooks/best-practices/).

## When an endpoint is switched off

If an endpoint has had no successful delivery for 3 days, and at least one failure in that time, Mestra switches it off and emails the organisation's owners. One success starts the count again. An event that is still failing after about 3 days is given up on, and if nothing has succeeded since it was first sent, the endpoint is switched off then.

A switched-off endpoint gets no more events, and its waiting retries are dropped.

## Turning it back on

Fix the receiver, then switch the endpoint on under **Webhooks**. Mestra doesn't replay what it missed. To catch up:

1. Open the **Events** tab and find the events from the gap.
2. Choose **Resend** on each, and pick the endpoint. Resend works for events up to 15 days old.

Or list the events from the [API](https://withmestra.com/docs/webhooks/testing/#list-events-from-the-api) and fetch what you need.
