Help Center/Admin Operations/Review failed jobs and the dead-letter queue, and retry safely
Admin Operations•Last updated: 2026-10-20•3 min read

Review failed jobs and the dead-letter queue, and retry safely

Failed Jobs lists every failed run; the Dead-letter queue lists work that exhausted its attempts (jobs and emails). A super administrator can retry only jobs whose handler is safe to run again.

Have questions or need step-by-step guidance on this topic?

Ask AI Copilot

Step-by-Step Procedure

1

Open Failed Jobs

Open "System Operations" and click "Failed Jobs" (address /admin/system/failed-jobs). Each row is a failed run with "Job", "Trigger", "Attempt", "Started", "Duration (ms)", "Retryable now" and the "Error".

2

Retry

On a row with "Retryable now: yes" click the row button "Retry job" and confirm. The job runs again immediately through its registered handler and the new run is recorded. A job already succeeding, or one with no safe handler, is refused with the reason.

💡 Pro Tip: Retryable jobs: price refresh, email outbox flush and the retention purge (a retry of the purge is always a dry run).
3

Open the dead-letter queue

Click "Dead-letter queue" (address /admin/system/dlq). A job is dead-lettered after failing its maximum attempts in a row. Emails the outbox gave up on appear too, with the recipient address partly hidden.

4

Replay

Click "Replay" on a row and confirm. A job runs again; an email goes back into the outbox for the next drain. Nothing is deleted.

Expected Operational Outcome

Failed work is visible with its error; a retry or replay really runs it again and the outcome is recorded, or you are told why it was refused.

Troubleshooting & Failure Modes

Symptom: "has no handler that is safe to run again"
Likely Cause: That job is operator-driven (for example ending stale meetings).
Resolution: Run it from its own tool in "Platform Operations".
Symptom: "ran again and failed"
Likely Cause: The retry itself failed; the failure is recorded.
Resolution: Read the error, fix the cause, then retry again.

Was this guide helpful?

Help us refine our platform documentation

Related Ecosystem Documentation