On Thu, Dec 2, 2010 at 11:56 PM, Grzegorz Nosek <r...@localdomain.pl> wrote: > W dniu 03.12.2010 01:22, Roger Hoover pisze: >> >> What if the existing reference counter both closes the socket and >> removes the reference? I think that would make the code much simpler. > > I'm not sure it will work either, though I don't know if the problem may > ever occur in supervisord in real life. If sockets are only closed just > before destruction, it probably isn't an issue and would indeed be the > simplest. > > If you get a ManagedSocket to close the socket and unregister itself, you > may end up with two ManagedSockets for the same url -- one that is kept > alive by a SocketManager but not present in SocketManager.sockets and a new > one, created by _another_ SocketManager and put into .sockets. > > Does that make sense or am I just overly paranoid?
I think it makes sense. We want to keep a registry (the SocketManager class) of all managed sockets. This same registry can be used to do error checking to make sure that a config doesn't have duplicate sockets without the reuse_socket flag. > > Best regards, > Grzegorz Nosek > _______________________________________________ Supervisor-users mailing list Supervisor-users@lists.supervisord.org http://lists.supervisord.org/mailman/listinfo/supervisor-users
