Open Stack

> On 21 Jun 2016, at 08:47, Armando M. <armamig at gmail.com> wrote:
> 
> 
> 
> On 20 June 2016 at 18:41, Carl Baldwin <carl at ecbaldwin.net> wrote:
> Somehow, this thread hid from me for a couple of weeks.  I just
> reviewed something relevant to this here [1].  It proposes adding
> tenant id to segment.  But, it also enforces that tenant id is the
> same as that of the network owning the segment.  So, I say why store
> it at all?
> 
> I would argue that subnet should also not have a tenant_id and should
> just inherit it from the network.
> 
> It seems it may potentially limit the ability to describe ownership. Virtually all Neutron models have it. Not sure I see the value in its absence.

In QoS, we actually removed the attribute from QosRule based models in the early phase of feature development, for reasons similar to Carl’s. We now have an issue updating rules due to the way controller behaves. The issue is not fixed just yet, but I think this discussion in gerrit is very relevant to the topic:

https://review.openstack.org/#/c/244680/1/neutron/api/v2/base.py

Ihar