Guides
Dead Letter Queue
What happens to a Delivery Attempt chain once every retry has been exhausted with no success — and how you get it delivered once the underlying problem is fixed.
How a message ends up here
Delivery fails
→
retried: 30s, 5m, 30m, 2h, 8h, 24h
→
10 attempts exhausted
→
Dead Letter Message
Carries a reason
Each Dead Letter Message records why it ended up there — the failure reason from the final attempt — so you're not guessing what went wrong before you fix it and replay.
Expires after 30 days
Unreplayed messages are retained for 30 days, then cleared out automatically. Replay before the window closes if you want another shot at delivery.
Replay creates a new Event
Replaying doesn't resurrect the original delivery attempt — it publishes a new Event with the same payload and runs it through fan-out again. The dead-lettered row is marked
replayed and links to the new Event ID, so you can trace the connection either direction.
Need to clear out a whole batch at once — say, everything that failed during an incident window?
See Bulk Replay.