Open Stack

On 13/06/13 16:06 -0400, Eric Windisch wrote:
>On June 13, 2013 at 15:26:31 , Ray Pekowski (pekowski at gmail.com) wrote:
>On Thu, Jun 13, 2013 at 12:49 PM, Eric Windisch <eric at cloudscaling.com> wrote:
>
>For TCP, I prefer the port-0 approach, but there would need to be a way to determine which port the backdoor is listening on. I think having this represented in logs would be sufficient.
>
>Would you give a +1 to a resubmit of the patch with the backdoor_port option deprecated and replaced it with a boolean option called enable_backdoor?  Or are you of the opinion that the current backdoor_port option is not broken?
>
>
>I'm not sure I'd classify it as 'broken', but I think it is bad at present.
>A boolean would be better. For determining the port, I'd rather it be random. More than that, as before stated, I'd prefer a unix socket.
>To make this a unix socket, you'd have to create the socket yourself and then pass that to eventlet.backdoor.backdoor_server(), rather than using eventlet.backdoor.backdoor()
>

FWIW, I don't consider it broken either and I agree with the support
for unix sockets.

Also, I'd rather use a boolean that *always* pick a random port. It
doesn't make sense to allow port hints if the user will have to "find
out" what port was picked up anyways.

We also have to consider that this is a major change and might break
some deployments that are using it so, we better take a
deprecation-path for the old argument - by warning users that it'll be
deprecated in the Ith release or whatsoever.

Cheers,
FF

-- 
@flaper87
Flavio Percoco