Pool cannot be used when connection is broken without rst due to DB hang

[choijunwoo]

Bug Report

Versions

• Driver: 8.0.31
• Database: mysql 8.0.28
• Java: 17
• OS: rocky8

Current Behavior

• If the DB hangs for some reason, the application cannot receive rst and the connections will remain in the established state. I guess if there are max size of connections in the pool and a new acquire connection occurs, the request will fail because the connection cannot be newly created.

• 
• 

Steps to reproduce

1. use long sleep statement
2. packet drop use iptable on mysql server
3. restart mysql and wait close fin-wait socket
4. remove iptable rule

Expected behavior/code

Possible Solution

• I think there should be a similar function to connection timeout or query timeout

Additional context

• It seems like a similar case to the issues below
  • Add support for disconnect on timeout to recover early from no RST packet failures redis/lettuce#2082
  • https://blog.cloudflare.com/when-tcp-sockets-refuse-to-die/