Open Stack

On 08/21/2012 10:19 AM, Mark McLoughlin wrote:
> Hey,
> 
> On Tue, 2012-08-21 at 09:45 -0600, Monty Taylor wrote:
>> Hey all!
>>
>> I'd like to request a FFE for the following:
>>
>> https://review.openstack.org/#/c/11027/
>>
>> The patch adds the ability to register and load virt drivers via
>> entrypoints. There are a couple of virt drivers that I've been involved
>> with or that I'm becoming involved with, such as the openvz driver and
>> the bare-metal work, that aren't going to land this cycle. But there's
>> no real reason why it needs to be quite so wonky to have a virt driver
>> out of tree. (patch was driven by trying to figure out ways to have CI
>> test these things _before_ we had them landed or ever proposed)
>>
>> The way the patch works is that it registers an entrypoint for virt
>> drivers and also registers the current in-tree virt drivers with that
>> entrypoint. Because we have people using the direct class-loading
>> interface that is currently there, the driver loader falls back to
>> class-path loading as currently works.
>>
>> So while it adds a potential code loading path, it does not alter the
>> behavior of the current system - and I think it should be safe enough to
>> land, which should give us an easier mechanism moving forward in this
>> next cycle for testing and vetting new work in the virt driver layer.
> 
> I guess one of the most important things with an FFE is the benefit to
> users. You say in the commit:
> 
>   With this, things like "pip install nova-virt-hyper-v" could be
>   expected to work.
> 
> With the patch, will that actually work in Folsom, or is it just laying
> the groundwork for it to work in Grizzly?

The mechanism with work in folsom. (pip install might be a bad example
here) Hyper-V landed, so that probably won't so much work. but a better
example could be:

pip install nova-virt-openvz

or

pip install nova-virt-ntt-baremetal

And we could TOTALLY make that work for folsom.

Monty