Skip to main content
Delivery of resource lifecycle webhooks typically occurs in near real-time, though occasional delays may occur.For alerts on usage, cost, and credit balance, we evaluate alerts (and therefore issue webhooks) several times a day, but the latency guarantee will be a function of your workload. This can be reduced with provisioning for customers on Growth and Enterprise plans.

​
Retries

Orb guarantees at-least-once delivery for a single webhook event to each of your endpoints. Orb expects a 2xx within five seconds as a successful response to a webhook request. If the request returns an error or otherwise times out, Orb will attempt to retry the event on the following retry schedule, stopping when either retries are exhausted or the request is accepted:
  • 5 seconds
  • 5 minutes
  • 30 minutes
  • 2 hours
  • 5 hours
  • 10 hours
  • 10 hours
In order for your system to handle duplicate notifications, we recommend storing the id of a webhook event temporarily in the webhook consumer and refusing to re-process an event if the same id has already been seen. A retry keeps the payload style its event was created with, even if the endpoint’s style has changed since.

​
Automatic disabling of failing endpoints

Any webhook endpoints that return a 5XX HTTP status codes for more than 99% of deliveries over a 7-day period will be disabled automatically by Orb to preserve system stability. These endpoints can be re-enabled in the Orb UI.