Delivery and retries
Each delivery is a POST of one event to your endpoint. Answer any 2xx within 10 seconds to acknowledge it.
- Answer
2xxwithin 10 seconds. Store or queue the event, answer, then do the slow work. - Never redirect. Mountaya does not follow a
3xx, so it counts as a failure. Register the final URL. - Refuse what does not verify. Answer
400to a bad signature, and process nothing.
Only Mountaya-Signature is signed. The other Mountaya- headers restate the body to help you route it, so read the event from the verified body.
Any answer but a 2xx fails the attempt, and so does no answer within 10 seconds. A failed attempt is retried on this schedule, about 72 hours in all:
To ask for a pause, answer with a Retry-After header. Mountaya waits as long as you ask, within that schedule.
Mountaya delivers each event at least once. A retry after a lost answer delivers it again, with the same id. Record the ids you processed, and skip one you have seen.
Events arrive in no promised order. To order one resource's events, compare the version its data.object carries, or read the resource itself when order matters.
Every failed attempt names its kind and whose it is. A receiver failure is yours: a timeout, a refused connection, a TLS error, a redirect or a non-2xx status. A Mountaya failure is ours, and never counts against your endpoint.
When every delivery to an endpoint has failed for 24 hours, Mountaya emails the organization's owners and admins. At 72 hours, it disables the endpoint. It does neither while it detects a delivery problem of its own.
The Mountaya app shows the same report under Organization → Webhooks.