Dev.to
7/15/2026

The original title is about a technical pattern called "The Reconciler Pattern" dealing with queued jobs that never run.
Original: The Reconciler Pattern: When a Queued Job Simply Never Runs
Short summary
When a queued job is dispatched once at row-creation time and that dispatch is silently lost, the row sits pending forever with no error or failed_jobs entry. The fix is a reconciler: a scheduled sweep that finds stale, unclaimed rows and re-dispatches them. The database is the source of truth, not the queue. Key implementation details include touching rows before re-dispatch, capping re-dispatch volume, and ensuring job idempotency.
- •Lost job dispatches leave rows pending with no error — invisible failures
- •Reconciler pattern: scheduled sweep for stale unclaimed rows that re-dispatches them
- •Database is source of truth; queue is just a delivery hint
Generated with AI, which can make mistakes.
Is this a good recommendation for you?



