- Newest
- Most votes
- Most comments
Aurora MySQL Events are managed by a dedicated event scheduler thread that must be explicitly enabled. The global event_scheduler parameter controls whether the event thread runs and defaults to OFF, so it needs to be turned on for events to execute.
For your issue with events suddenly changing from a per-minute to an hourly schedule, there are a few potential causes to investigate:
-
Event Scheduler Status: Verify that the
event_schedulerglobal parameter remains enabled (ON) on your Aurora MySQL instance. If it gets disabled or the instance restarts without the parameter being persistent, events won't run as scheduled. -
Event Metadata: Check the event definitions in the
INFORMATION_SCHEMA.EVENTSview to confirm that both events still have their correct schedule definitions (EVERY 1 MINUTE). It's possible that one event's schedule was inadvertently modified. -
Event Errors: Review the Aurora MySQL error log for any event-related errors. Event errors are written to the error log and could indicate why an event isn't executing as expected.
-
Resource Constraints: On a db.t4g.large instance, if the database is experiencing high load or resource constraints, the event scheduler thread might be delayed in executing events, though this typically wouldn't cause such a consistent shift to hourly execution.
I recommend checking the current event definitions, verifying the event_scheduler status, and reviewing the error logs to identify what might be causing one event to deviate from its intended schedule.
Sources
SQL Server Agent and MySQL Agent - SQL Server to Aurora MySQL Migration Playbook
Relevant content
- asked 4 years ago
- asked 4 years ago
This server was restored from a snapshot and had existing events that were scheduled hourly. The event that was not running correctly was running on the same schedule as one of these existing events. I dropped these events that were originally created on the source server of the snapshot and the schedules are now working as expected. Another issue I was having was running "ALTER EVENT myEvent ON SCHEDULE AT UTC_TIMESTAMP + INTERVAL 1 MINUTE ENABLE;" was also not running until the hourly schedule. So it seems that when restoring from a snapshot events need to be dropped and re-created. I suspect also performing a global failover will also results in events needing to be dropped and re-created.