Description of problem:
Let's say you send a signal to the engine service (python script).
Then, you restart the engine.
The engine isn't stopped, as the service is down, so the result is two engines.
Only one can get requests, of course, but there are many internal processes that are running.

Version-Release number of selected component (if applicable):
3.6.4

How reproducible:
Happened on local rhev environment once. Didn't try to reproduce.

Steps to Reproduce:
Longer description - steps to reproduce can be easily derived from here.
* The engine was started on Mar 20, at 12:40:12, with PID 22749 for the
engine and PID 22747 for the service script.

* On Mar 21, around 13:01, someone tried to make a thread dump of the
engine, maybe because it wasn't working correctly. It tried to use "kill
-3", but used the wrong PID:

  # kill -3 22747

This killed the service script, but left the engine running.

* On Mar 22, around 13:33, someone restarted the engine, but this didn't
stop the engine that was already running, because the service script was
already gone. As a result a new engine was started. This second engine
failed to listen to ports, as they were in use by the old one, but
anyhow it deployed the applications, connected to the database, and
started to try to manage the hosts. Note that the old engine was still
running, and servicing UI and API requests.

Actual results:
The signal wasn't propagated.

Expected results:
The signal should propagate. IMO on -3 it should keep the service running as well.