On 06/15/2017 11:05 AM, Jay Pipes wrote:
On 06/15/2017 11:56 AM, Ben Nemec wrote:
Full disclosure: I primarily work on TripleO so I do have a horse in
this race.
On 06/15/2017 10:33 AM, Jay Pipes wrote:
On 06/15/2017 10:35 AM, Jeremy Stanley wrote:
On 2017-06-15 10:48:21 +0200 (+0200), Thierry Carrez wrote:
[...]
I think that, despite the efforts of the Fuel team, Fuel did not
become
what we hoped when we made it official: a universal installer that
would
be used across the board. It was worth a try, I'm happy that we tried,
but I think it's time to stop considering it a part of "OpenStack"
proper and make it a hosted project. It can of course continue its
existence as an unofficial project hosted on OpenStack infrastructure.
Thoughts ?
I agree, it makes sense to be more clear as to the lack of
community-wide support for that effort. Perhaps if its popularity
increases outside one vendor's customer base to the point where
contributions from a broader set of stakeholders emerge, we can once
again evaluate its governance state.
While I personally agree that Fuel should be moved out of the official
projects list, I'd like to point out that Triple-O is virtually entirely
a Red Hat project:
http://stackalytics.com/?module=tripleo-group
http://stackalytics.com/?module=tripleo-group&metric=commits
so the fact that a project is entirely run by a single vendor or "has
popularity outside one vendor's customer base" has not been and
continues not to be a deciding factor on whether something is an
official OpenStack project or not.
I don't believe the single vendor-ness of the project is the reason
it's being proposed for removal. It's the fact that the single vendor
has all but dropped their support for it. If Red Hat suddenly decided
they were pulling out of TripleO I'd expect the same response, but
that is not the case.
Please see Jeremy's paragraph directly above my response. He
specifically mentions single-vendor-ness as a reason for removal.
I'd fully support the removal of all deployment projects from the
"official OpenStack projects list".
I would not. Deployment of OpenStack remains one of the most
difficult to solve problems and I would be highly disappointed in the
community if they essentially washed their hands of it.
This right here is the perfect example of what Thierry is getting at
with the *perceived value* of the term "Big Tent" or "Official OpenStack
project". :(
What about having deployment projects be "non-official" or "ecosystem"
or "community" projects means that "the community ... essentially washed
their hands of it"? :( You are putting words in my mouth and making a
false equivalence between "community project" and "of lesser value". And
that's precisely the problem these terms: people read way too much into
them.
The problem is that no matter what you call it, as long as you have two
groups, one that includes Nova, Neutron, etc. and one that doesn't, the
one that does is always going to be seen as more "important". Even if
it's purely a perception thing (which I won't dispute), it's a
meaningful perception thing and I think moving the deployment projects
would send the wrong message. Deployment tools are a critical part of
the OpenStack ecosystem and their categorization (whatever it ends up
being called) should reflect that.
Maybe part of the problem is the context of this discussion. We moved
from talking about a largely abandoned project to all the deployment
projects, which (intentionally or not) draws some uncomfortable
parallels. It also appears this is a discussion we should probably
table until the big tent terminology one is completed because until then
we're mostly debating our individual interpretation of the governance
model. Once we're done redrawing the lines of demarcation in the
OpenStack community maybe the correct place for deployment projects will
be more obvious. I doubt it, but one can hope :-).
There are considerably more deployment projects than just TripleO and
Fuel, and there is more collaboration going on there than a simple
commits metric would show. For example, see
http://specs.openstack.org/openstack/oslo-specs/specs/pike/machine-readable-sample-config.html
which came out of a cross-deployment project session at the PTG as a
way to solve a problem that all deployment tools have.
Ben, I don't doubt this and as I've said publicly, I 100% support the
joint deployment efforts and collaboration.
We should be encouraging more community involvement in deployment
tools, not sending the message that deployment tools are not important
enough to be official projects.
What better way to encourage *community involvement* by saying all
deployment tools are *community projects*?
Best,
-jay
__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: [email protected]?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
__________________________________________________________________________
OpenStack Development Mailing List (not for usage questions)
Unsubscribe: [email protected]?subject:unsubscribe
http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
