[openstack-dev] [nova] [neutron] the nova network facade that isn't

Sean Dague sean at dague.net
Wed Apr 20 00:58:09 UTC 2016


On 04/19/2016 07:11 PM, Matt Riedemann wrote:
<snip>
> I've always personally considered the network proxy to neutron in the
> API as getting more of a pass since nova has a network service in it, it
> doesn't have it's own image or volume service since those were
> completely split out. But since the adoption rate for neutron has taken
> off, it does make less sense to put effort into filling the API gaps
> with new proxy code.

I think we're working on the general principles as follows:

1) Do not make the proxies any better then they are. We don't want to
encourage anyone to actually use them.

2) Deprecate (then delete) interactions with the proxies from novaclient
(volumes just got deleted, images are now deprecated, networks can be
deprecated once nova-net goes deprecated).

3) Deprecate the API proxies themselves. This still begs the question
how we can signal that as it's a more discreet path to drop.

I think on #3 the work the api-ref transition into RST (and fixing the
multitude of doc issues) is making it a bit easier to see the whole API
again and some other things which are weird enough that we want to
outright deprecate towards remove, and not phase out via microversions.

I've also kind of been wondering if we would start that process by
making it easy for administrators to disable the proxies through
configuration, so they could start flushing out their users early and
get them doing it the right way (and back it off if the screaming was
too loud).

	-Sean

-- 
Sean Dague
http://dague.net

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 811 bytes
Desc: OpenPGP digital signature
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160419/6ac5de75/attachment.pgp>


More information about the OpenStack-dev mailing list