Pool resources not being killed - clients being kept active forever

[z0mg]

Expected behavior

I'm expecting clients to be disconnected when closing a JedisPool (JedisPool#close).

Actual behavior

Sometimes clients aren't disconnected. I make a new JedisPool instance, use a maxTotal config of 36, call close and repeat. After a while, redis logs report 38... 39... 45... 50... 300 clients. This keeps growing.

Steps to reproduce:

Unfortunately I can't reproduce the issue. I've tried simplifying my mechanism and in synthetic tests it works.

Was this a known problem at some point?