@Marvin I am using prefect 3.x and seeing a large events and event_resources table. I am seeing the error message

Event %r (%s) for %r has exceeded the maximum recursion depth of %s

which seems to be related to https://github.com/PrefectHQ/prefect/issues/15399. We have observed that a cancelled deployed parent flow with deployed child flows continues to grow the event database with the child flow being stuck in the CANCELLING state. This seems to strain our prefect server as it pretty quickly tries to run

periodic evaluation

upon startup. Given the events tables being so large this causes long running queries which cause health check requests to fail. Any suggestions on how to clean up the events table?