We found a significant bug in our error handling. Essentially a slow query will log but it logs using a query which if under load causes the slow query logger to log again which starts an infinite loop of self logged slow queries. I am refactoring our entire error handling system with Opus 4.1 right now. Hope I don't regret it.