Faster retries for 5xx responses

The first two retries after a server error now fire within 30 seconds, which clears most short outages before customers notice.

Most failed deliveries come from brief restarts or deploys on the receiving side. We tuned the backoff schedule so the first two retries land at 10 and 30 seconds, then fall back to the usual exponential curve.

Do I need to change anything?

No. The new schedule applies to every endpoint automatically. If you rely on the old timing, you can pin it per endpoint in settings.