[openstack-dev] [nova] Stuck on VIF plugin script?

Jay Pipes jaypipes at gmail.com
Mon Jun 22 13:39:37 UTC 2015


On 06/22/2015 06:59 AM, Neil Jerram wrote:
> On 22/06/15 11:28, Neil.Jerram at metaswitch.com wrote:
>> ‎It seems like we're possibly stuck now on the VIF plugin script spec
>> [1]; there being core comments in apparently conflicting directions. I
>> wonder if it's still feasible for a version of this to land during
>> Liberty?
>>
>> [1] https://review.openstack.org/#/c/162468/
>>
>> I plan to reread everyone's comments and see if ‎I can propose a way
>> forward - but thought I'd circulate this email first in case anyone
>> wants to say that it's already too late...
>
> I'm sorry, actually Dan PB has already done this with [2], so I'll
> review that.
>
> [2] https://review.openstack.org/#/c/193668/

Yes, and actually this weekend, I pushed a start of an os_vif library to 
GitHub:

https://github.com/jaypipes/os_vif

I still need to port/reconcile the unit tests and add functional tests, 
but all the basics have been ported from the Nova libvirt/vif.py file, 
along with supporting code from nova/network/linux_net.py.

The proposed library above adds object models for the VIF, Network, and 
Subnet objects, using oslo.versionedobjects, that can be used to 
construct the VIF requests.

Each vendor/VIF type has its own plugin, loaded via Stevedore and 
standard Python setuptools entry points, making it trivial to add new 
plugins and use this library for coding instead of the Nova codebase 
itself, which should solve the problem of development velocity.

The plugins need only implement plug() and unplug(). I left the libvirt 
XML configuration pieces in Nova's libvirt/vif.py file for a couple reasons:

a) libvirt knows best how to configure this stuff

b) os_vif should really be a resource for more than just libvirt

Anyway, it's at least something that we can use as a starting point for 
the discussion on Dan B's spec above...

Best,
-jay

> (For some reason Gerrit has stopped sending me new comments on [1]. Does
> anyone know why that happens?)
>
> Regards,
>      Neil
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev



More information about the OpenStack-dev mailing list