Database Connection Pooling

Description

Configure, size and review PostgreSQL connection pooling - one app-level pool per process (pg-pool, SQLAlchemy), PgBouncer session/transaction/statement modes and what breaks in transaction mode (session SET, LISTEN/NOTIFY, advisory locks, named prepared statements), serverless and edge connections through a pooler (Supavisor, Neon, PgBouncer), sizing from the database server's cores, timeouts, and finding connection leaks in pg_stat_activity. Use whenever the user deploys a Postgres app to production, hits "too many connections" or "remaining connection slots are reserved", configures PgBouncer or a cloud pooler, asks what pool size to use, connects from Lambda/Vercel/Cloud Run/edge functions, sees "prepared statement does not exist", asks why the app is slow or hangs under load, or sees connections piling up in pg_stat_activity.