Pain point · Reliability & bugs · Causes churn
Event persister stops flushing after database cancellation
Part of the theme “Event persister stops reliably processing events after errors” · 2 reports · see them all
1 source thread · first seen 2026-10
Summary
A database outage can cause the event persister's periodic flush task to exit permanently. Events may no longer be stored and event-based automations stop firing until the server is restarted; a batch already removed from the queue can also be lost.
- Affects
- Operators running Prefect Server with PostgreSQL
- Workaround
- Run a timer that posts a probe event, checks whether it was stored, and restarts the server if it was not.
Evidence
Excerpts are copied word for word from the source; follow the link to read it in full.
“no events are written to the database until the server is restarted”