dead. Find them, read why, fix the receiver, re-enable it, and replay exactly the events it missed — nothing lost, nothing double-run if your consumer is idempotent.
1
List what failed
dead means the retry schedule (8 attempts, ~32h) exhausted; failed is an attempt that errored. Filter to both.2
Read the actual error
get shows the status code and error string from the last attempt — a 500, a TLS failure, a timeout.3
Fix, then re-enable the destination
After 20 dead deliveries a destination auto-disables. Clearing it resumes owed deliveries on the next wake — so re-enable only once the receiver is actually fixed.
4
Replay the events it missed
Replay each captured event to the destination. The server delivers the stored bytes with a fresh idempotency key per call, so a retried replay is safe.
Replaying to a destination re-delivers the exact signed bytes — no
--edit, unlike a --forward
replay to localhost. That’s what keeps the signature valid for your receiver.