Do you also need to clear states when a fail over pool recovers?

[Perry]

My feature request took a turn and the solution has just move the problem for me (might even made it bigger). Never the less I need to know how guilty I should feel so vote away.

[cmb]

As I said, not something a lot of people need.  ;D My last comment in #8 explains why it won't be done for 2.0. You can use sticky connections to avoid the breakage, and alternatively you can easily modify the source to kill all states on every status change which sounds like probably what you want. Or put any command in there you want, kill only states for specific IPs, lots of possibilities.