On Mon, Jun 12, 2023 at 3:51 AM Dmitry Tantsur <dtantsur@protonmail.com> wrote:


On 6/8/23 12:53, smooney@redhat.com wrote:
> On Wed, 2023-06-07 at 15:44 -0700, Dan Sneddon wrote:
>> On Wed, Jun 7, 2023 at 12:49 PM Jeremy Stanley <fungi@yuggoth.org> wrote:
>>
>>> On 2023-06-07 12:07:59 -0700 (-0700), Dan Sneddon wrote:
>>>> On Wed, Jun 7, 2023 at 9:00 AM Clark Boylan <cboylan@sapwetik.org>
>>> wrote:
>>> [...]
>>>>> Is there some reason the existing repository won't work?
>>>>
>>>> Currently os-net-config is a part of the TripleO project. Since
>>>> TripleO is retiring and being replaced by something hosted on
>>>> GitHub, we are no longer maintaining the Master/Zed branches.
>>> [...]
>>>
>>> Maybe you misunderstood. To restate: Is there any reason the people
>>> who want to use and maintain the openstack/os-net-config master and
>>> stable/zed (or other) branches can't just adopt the project? It's
>>> well within the TC's power to grant control of that repository to
>>> another project team who isn't TripleO.
>>>
>>> Which use cases specifically (outside of Red Hat's lingering
>>> interest in the stable/wallaby branches of TripleO repositories) are
>>> you referring to?
>>> --
>>> Jeremy Stanley
>>
>>
>>>
>> The people who want to continue to use os-net-config are developing the
>> replacement for TripleO, but they have moved to GutHub. That’s an option
>> for os-net-config, but not my first preference.
>> I have over the years heard of companies using os-net-config for various
>> use cases. It’s possible that posting to openstack-discuss won’t reach any
>> of those users, but it doesn’t hurt to ask here.
> i will need to reread the thread but i thought that there was reference to baremetal
> use cases i had assumed that meant ironic?
>
> just because we are using GitHub for the replacement for tripleo is not a reason to move
> things to github by default. movign to github for the code review as im sure you are aware
> is a much worse code review interface. we would be loosing the release tooling and ablity to
> publish to pypi. the ci would need to be ported amoung other things.
>
> There is also an open question of if/how much of os-net-config will continue to be used for the
> operator based installer. alternatives are being considered although we have known gaps.
> no desicssion has been made on if we will continue with os-net-config or replace it with nmstate
> or a hybird of the two.
>
> a quick search https://codesearch.opendev.org/?q=os-net-config&i=nope&literal=nope&files=&excludeFiles=&repos=
> seams to indicate that os-net-config is used by:
> - openstack-virtual-baremetal
> - networking-bigswitch
> - possibly starlingx
>
> so i would suggest moving it to ironic governance and keeping the repo as is to suppot the virtual baremetal usecase.

None of these repos are part of the baremetal project. I cannot speak
for the PTL or for the team, but I highly doubt we'll be in a position
to adopt os-net-config.

Dmitry

+1

I don't think any Ironic documentation refers to using os-net-config -- it's a single option for how to configure your images post-boot out of a large number (many of which remain supported). I agree with Dmitry that it's hard to see this fitting well in the baremetal program. 

Thanks,
Jay Faulkner
Ironic PTL
TC Vice-Chair