Skip to content
Dashboard
Theme
Theme

Retries

Handle bounded webhook retries, duplicates and endpoint disablement.

View Markdown

Failed webhook deliveries receive up to nine attempts. Retry delays after successive failures are 1 minute, 5 minutes, 30 minutes, 2 hours, 5 hours, 10 hours, 24 hours and 30 hours, each with ±10% jitter. Do not use those times as a scheduling guarantee.

After 72 hours of failures without a success, the endpoint is disabled and account owners/admins are emailed. Test events are sent once and are not retried. Monitor disabled endpoints and failed deliveries in your integration operations.

Return a successful response after durably accepting a verified event. Keep the HTTP handler short: verify raw bytes, validate the envelope, deduplicate and commit or enqueue the work. A timeout or failure can cause a retry even if your first attempt performed some effects.

Webhook attempts can arrive out of order. Event IDs deduplicate business effects; directory sequence numbers support ordered processing. Do not mark an unapplied event as completed simply to suppress retries.

The public delivery API exposes delivery status, attempt count and response status. It is not an archive of complete request or response bodies. Use replay and catch-up for recovery.