Fix my database connection management. Find where the app creates its database connection or client and check how connections are pooled and bounded. Configure: a connection pool sized for the actual environment (a small pool like 5-10 for a single serverless-ish or small server, not one connection per request); a maximum total so concurrent requests wait instead of opening unbounded connections; connection acquisition timeouts so a request fails fast when the pool is exhausted rather than hanging forever; query timeouts so a runaway query can't hold a connection hostage; and idle connection cleanup so stale connections get recycled. Pay attention to how the runtime actually works: if this runs on serverless functions, one long-lived global pool per warm instance is the pattern, not a pool per invocation, and the pool size must be small because many instances multiply connections. Don't change any query logic. Don't add an external pooler service unless the numbers justify it; note it as an option instead. Deliver: the connection configuration changes, the pool size and timeout values with the reasoning, and verification: show the maximum connections the app can open and confirm it stays under the database's limit.