Failed Webhook Delivery Recovery

Hookdeck queues incoming webhooks and records whether each delivery to a destination succeeded or failed. A Beam agent reads failed deliveries and applies the customer's approved retry rule, such as retrying up to five times over an hour for a known flaky destination, then confirming once a retry succeeds. It writes the outcome back against the event so the team has a record without checking Hookdeck by hand. Deliveries that keep failing past the retry limit, or that return an error the rule does not cover, are routed to an engineer to investigate the destination directly.

Webhook Routing Rule Updates

Hookdeck lets a team route webhooks from one source to multiple destinations based on rules like event type or payload content. A Beam agent watches for a new event type that does not match an existing routing rule and, where the connector allows it, applies the closest approved pattern the team has defined, such as sending unmatched Stripe events to a general queue instead of dropping them. It notes which events were routed this way. New event types with no comparable existing rule are left unrouted and flagged for an engineer to configure properly.

Delivery Backlog Alerts

A backlog builds in Hookdeck when a destination is slow or down and cannot keep up with incoming webhook volume. A Beam agent checks queue depth against a threshold the customer sets, for example more than five hundred pending events for one destination, and notifies the owning engineer once it is crossed. Where the rule allows, it also pauses lower priority destinations first to protect capacity for critical ones. Backlogs tied to a destination outage of unknown cause, or that cross multiple destinations at once, are escalated to a person rather than resolved by pausing alone.