Events that are never lost

Every save and every business event is written to the event log and handed to workflows, agent triggers and webhooks from there, so a restart loses nothing. Platform, Automation, Event Delivery shows what could not be handed over.

A save writes its event in the same moment as the record. Workflows on Record created or Record updated, workflows on an event, agent triggers on an event or a record, and outbound webhooks that name the event read it from there, each exactly once, even when the server restarts in between.

If one of them cannot take an event, it is tried again after 30 seconds, then 1, 2, 4 minutes, and after five attempts it is set aside on Event Delivery with the reason. Send again tries it once more when the cause is fixed.

An outbound webhook delivery is always signed. A webhook with no signing secret has its deliveries failed with the reason (Unsigned — no secret set) rather than sent; a delivery whose receiver could not be looked up (a DNS failure) is retried up to the webhook's retry limit. See Signing webhook deliveries.