Rudder

The issue is that systemd does not have proper inital track of the state of traditionnal init services. It means that mixing service/etc/init.d calls with systemctl cause trouble.

For example, if systemd thinks a service is not running (for example if it has been started usint the init script directly, wich is the case on CentOS7 using the service command), when asking for a restart, it will not stop the service but only try to start it, which will break...

It may break running systems when updating ncf, so we should use the init script when it is there.