Thu, May 05, 2016 at 02:26:46PM CEST, olichtne@redhat.com wrote:
From: Ondrej Lichtner olichtne@redhat.com
Hi,
what follows is a very large patch set with a lot of stuff happening in it. Normally I would have split it up into at least 2 logical parts, but unfortunately most of the changes introduced in the patch set is dependent on each other. The reasoning behind all of the code should be nicely explained in the commit messages, if you have any additional questions don't hesitate to ask.
One thing that I want to explain is the creation of a whole new concept of the Controller Device Database, mirroring the one on the Slave so that I could work on testing OvS vxlans. For now the implementation is just that and nothing more. At the moment this makes the Controller look very ugly seeing as there's both an Interface object and a Device object representing the same NIC but with different information.
In the future I want to merge these two classes, mostly by replacing the current Interface objects with the new Device objects extended with setters and other features required for proper configuration. This however is a large restructuring of the code and it would either depend or heavily conflict with the implementation of python recipes which is why I chose to just do a very quick and ugly implementation of a parallel concept that I can use for now while we wait to merge in the Python recipes implementation.
v2: fixed a traceback when configuring veth devices - reported by jtluka, the 3 updated commits are: InterfaceManager: fix rescan_devices method InterfaceManager: extend Device::if_data method InterfaceManager: add if_deleted update messages
I've checked my recipes with veth devices. The issue has been resolved. Ack to series.
Acked-by: Jan Tluka jtluka@redhat.com