For consistency with beakerd, we should use python-daemon (instead of kobo.process.daemonize) in the lab controller daemons as well. See also bug 812687.
James wrote a patch for this: http://gerrit.beaker-project.org/#/c/1399/ I will just need to double-check that everything works nicely with gevent.
The main problem that I can find with this is that we daemonise before we try to login to the scheduler, and if the login fails we do not kill the process. If time permits I think we should change this.
(In reply to comment #4) > The main problem that I can find with this is that we daemonise before we > try to login to the scheduler, and if the login fails we do not kill the > process. > > If time permits I think we should change this. I've already tried to fix it, a long time ago: http://gerrit.beaker-project.org/824 but it's not possible due to kobo bug 753006. I believe this patch for switching to python-daemon hasn't made the situation any worse than it was before.
Sorry I forgot about that old bug. And no this doesn't make it anyworse than before.
Beaker 0.11.0 has been released.