How to reliably terminate processes?

[nipos]

I have a problem with dead processes that don't want to go away on my servers.
I searched countless hours already without a suitable solution.
I'm experimenting with a ASP.NET based webapp that doesn't correctly work currently and the amount of mono processes keeps increasing,as I just can't kill them.
I tried many different kill signals already,up to -126 (which is the most aggressive,I think?).
Now htop shows the processes with a red T,what I think stands for terminating.
Still the processes haven't gone away for hours,and the network port is still blocked by them.
I don't want to change the port (and the Nginx reverse proxy config) for every try,or do I really have to?
Is there any way to get rid of those processes?
Or is there a way that I can at least detach the port from them,so that I can listen to it in another process again?
You can see the htop output in the attached screenshot.

The same thing already happened to me in the past with other programs,for example some Node.js applications.
They crash sometimes and show up as Zombie in the process list (and I learnt that I can't kill something that is already dead,right?) but they still block the port where the application should run,so that I can't start them again.
When that happened on a development server,it wasn't that much of a problem as I could simply reboot,but I can't reboot a production server with dozens of public services every time a Zombie process appears...
Please help,I need the ports to be usable again at least

[J65nko]

As root user sending signal nr 9 (KILL) should do it. Signal nr 9 is not catchable by the process itself.