Webhook Events Your Endpoint Does Not Accept Are Sent Again
When your webhook endpoint does not accept an event, Straumur now keeps that event waiting and sends it again until your endpoint accepts it, instead of retrying it on its own and giving up after about seven hours. Every other event is still sent as normal.
What changes
A failed event is tried once more a minute later. If your endpoint does not accept that attempt either — any answer that is not a 2xx, a 3xx, 4xx or 5xx, or no answer at all — the event waits and shows as Waiting in the delivery log, and the webhook shows as Failing. New events are still sent and retried as normal while it is failing.
Straumur keeps sending the oldest waiting event, starting after about 10 minutes and backing off to every 8 hours. As soon as your endpoint accepts it with a 2xx, the webhook is no longer failing and every other waiting event is sent.
Your endpoint must accept every event with a 2xx.
Redirects are no longer followed: a 3xx answer counts as a failed attempt, so make sure the webhook URL is the final one. New webhook URLs must use a public domain name; IP addresses, localhost and names that resolve only to private network addresses are refused.
The longer retry schedule (10 minutes, 1 hour, 6 hours) is gone: it rarely delivered anything the 1-minute retry had not, and kept calling endpoints that were down.
Emails and Retry now
The users who manage the webhook get an email when it is still failing after the first check, about 20 to 30 minutes in, when it accepts the waiting event again, and when it is about to be disabled. In the Partner Portal the webhook shows as Failing with the number of events waiting, and Retry now sends the oldest waiting event straight away once you have fixed your endpoint.
Disabled after 30 days
A webhook that is still failing after 30 days — its endpoint has not accepted the waiting event in all that time — is disabled and its waiting events are marked as failed. Events it accepted meanwhile stay delivered. Once the endpoint is fixed, choose Enable on the webhook; the failed events can then be re-sent one by one. Disabling a failing webhook yourself also marks its waiting events as failed, so fix the endpoint at the same URL rather than replacing the webhook.