Does tcp wait recycle function on disabled pool nodes?

Hi,

 

I have some app devs who are taking down their nodes automatically when deploying new code. The idea is they take a node down, verify no connections to the server, deploy code and bring the node up. Rinse and repeat for each node in the LTM pool.

 

They're running into prolonged connections after disabling the pool node. We've tried setting action on service down to "Reselect", but that doesn't seem to work. No persistence is used on the VIP but connections continue to persist to the disabled node.

 

The one thing I'm assuming but haven't been able to backup with credible confirmation is that the TCP profile using tcp wait recycle isn't a culprit. I'm assuming that the F5 won't recycle connections in a time-wait state if the node is in a disabled state, but I don't see any documentation stating this is the case. Am I correct in my assumption? Or can the tcp wait recycle need to be disabled for this to work properly so that the F5 is keeping a connection open allowing new connections to reuse a connection that was in time-wait?

 

I'm assuming the client or server is keeping this connection open, but want to make sure this isn't the F5 somehow responsible for keeping the connection open.