Add new ML2 Driver without needing to release all of neutron

asked 2014-12-11 18:20:32 -0500

mlathe gravatar image

updated 2014-12-12 17:54:07 -0500

smaffulli gravatar image

I'm trying to figure out how to add an new ML2 Driver as an entry point for the neutron-service, without needing to release all of neutron.

There seem to be a few threads, like this:

  • https://ask.openstack.org/en/question/30274/how-to-create-new-ml2-mechanism-driver/ (https://ask.openstack.org/en/question...)
  • https://ask.openstack.org/en/question/50917/how-to-load-entry-point-for-new-ml2-driver-plugin/ (https://ask.openstack.org/en/question...)
  • https://ask.openstack.org/en/question/47190/how-do-i-associate-mechanism_drivers-name-to-the-actual-python-module-to-be-loaded-by-ml2/ (https://ask.openstack.org/en/question...)

These all seem to imply that you need to change the setup.cfg file which will update the entry_point.txt file in: /usr/lib/python2.6/site-packages/neutron-2014.1-py2.6.egg-info/entry_points.txt

My question is, how do you do this without changing something in neutron? Ie if I do this, then don't I need to release neutron too? I suppose i can force my settings into the entry_point.txt file during my install (ie use the openstack-config command to write my properties into that file), but that seems ugly.

Are there any known examples of ML2 Plugins that do this without being known in the main neutron codebase?

edit retag flag offensive close merge delete