A webhook system is only as useful as its worst day — the day your server is down, or ours is degraded, or the network between us drops half the traffic. We designed Mereb's webhook delivery around that assumption from day one.
Every event is retried with exponential backoff for up to 72 hours, and every delivery attempt (successful or not) is logged and replayable from the Mereb Hub. Events are also idempotent — each carries a unique event ID so you can safely process a redelivered event without double-counting it.
If you're building a webhook consumer, our recommendation is simple: acknowledge the event immediately with a 200, then process it asynchronously. Anything else risks timeouts turning a transient hiccup into a much longer retry storm.