Yes, I know. I do exec() on daemon binary directly.

And that's the problem. If you just use the init scripts both for starting and stopping the daemon, you don't have to care about all this stuff. That's what the init scripts are for.

Is it clear enough?

Yes, the difficulties with your current approach are clear to me. But I don't see why you have to take this approach. When you ask how to start and stop daemons, I tell you how it's usually done. Why can't you use the daemon in a way that the standard techniques apply?