[nova][neutron][cyborg] Bandwidth (and accel) providers are broken if CONF.host is set

Matt Riedemann mriedemos at gmail.com
Mon Nov 25 13:51:02 UTC 2019

On 11/22/2019 10:14 AM, Bal√°zs Gibizer wrote:
> Opened a blueprint[1], pushed up a small spec[2], and a WIP
> implementation[3] for the new API.
> Cheers,
> gibi
> [1]
> https://blueprints.launchpad.net/nova/+spec/filter-hypervisors-by-service-host
> [2]https://review.opendev.org/#/c/695716

The contentious point on the spec seems to be that it currently proposes 
to keep the same (incorrect) behavior where if filtering by 
hypevisor_hostname_pattern yields no results the API returns a 404 error 
today when it should just return an empty list. The question in the spec 
is then if when we add service_host filtering, if there are no results 
do we:

1. 404 (that's what the spec proposes to be consistent with existing 
[odd] behavior)

2. Return an empty list and in the same microversion change that 404 -> 
empty response behavior even if not filtering by service_host (so just 
filtering by hypervisor_hostname_pattern).

 From the review it sounds like most people (myself, Alex and Chris) 
think we should go with the latter and not perpetuate the former sin.




More information about the openstack-discuss mailing list