Open Stack

On 3/15/2018 3:30 PM, melanie witt wrote:
>      * We don't need to block bandwidth-based scheduling support for 
> doing port creation in conductor (it's not trivial), however, if nova 
> creates a port on a network with a QoS policy, nova is going to have to 
> munge the allocations and update placement (from nova-compute) ... so 
> maybe we should block this on moving port creation to conductor after all

This is not the current direction in the spec. The spec is *large* and 
detailed, and this is one of the things being discussed in there. For 
the latest on all of it, gonna need to get caught up on the spec. But it 
won't be updated for awhile because Brother Gib is on vacation.

>    * On routed provider networks:
>      * On the Neutron side, this is already done: 

I still don't know how this works because the summit videos talk about 
needing port creation happening in conductor which we never got done in 
nova.

I suggested to Miguel the other day that we/someone should setup a 
multi-node CI job which does all of the configuration required here to 
make sure this stuff actually works, because it's quite complicated 
given there are three services involved in making it all work.

Hopefully anyone that's already using this successfully in production, 
or is thinking about using it, would help out with setting up the CI 
configuration for testing this out - we could do it in the nova-next job 
if we wanted to make that multinode. Maybe jroll can be guilted into 
helping since he was the one asking about this at the PTG I think.

-- 

Thanks,

Matt