As far as I see the force kill of a command is triggered 5 seconds after the graceful kill if the process is not already gone.

We have some integration tests which need more time than 5 seconds to really shutdown everything.

Correct me if I'm wrong.

If I'm right please make the wait time configurable. This would avoid hanging processes which are not really in the process tree.
Basically the thing is that we start webservers which start workers. These workers are not in the process tree. Only the monitoring process. And when the monitoring process is killed before the workers were shut down they remain active.