Open Stack

On 10/31/2014 05:17 PM, Matt Joyce wrote:
> On one hand, I agree a member of the TC should be a very active member
> of the development community.  Something I have not been, much to my shame.
>
> However, there are obviously some fundamental issues in how the TC has been
> governing OpenStack in the past few releases.  Very serious issues in the
> project have been largely ignored.  Foremost in my mind, among them, is
> the lack of an upgradability path.  I remember there being large discussion
> and agreement to address this at folsom, and further back.  I have seen no
> meaningful effort made to address a functionality requirement that has been
> requested repeatedly and emphatically since as far back as austin.
There has been quite a lot of meaningful effort in this area. The
offline upgrade testing on every patch was added in Grizzly (grenade),
enforced early in Havana. It was made an integrated project requirement
in Icehouse (as soon as we got a fully directly elected TC), and now
most integrated projects participate in it (after TC gap analysis). We
held Ironic graduation a cycle largely due to the lack of an upgrade
path. The Nova team has been working on something better than that,
which is mixed version services, which we also test on every commit.
With the import of nova versioned objects into oslo, that should create
the framework for other projects to do the same kind of thing.

There are other decoupling that are happening for libraries consumed by
servers which should make this better as well (just pushed another patch
for that this morning).

And realistically part of my motivation for advocating a smaller ring 0
in OpenStack is to provide the focus on just this class of issues within
that.
>
> I can raise other issues that continue to plague usership, such as neutron
> failing to take over for nova-network now two releases after it's planned
> obsolescence.  My concern, is that the TC comprised entirely of active 
> developers ( most of whom are full time on the open source side of this
> project ), is trapped in something of an echo chamber.  I have no real
> reason to suggest this is the case, beyond the obvious failure by the 
> project to address concerns that have been paramount in the eyes of users
> for years now.  But, the concern lingers.  
>
> I fear that the TC is beholden entirely to the voice of the development
> community and largely ignorant of the concerns of others.  Certainly,
> the incentives promote that.  The problem of course, is that the TC is
> responsible for driving purogratives in development that reflect more
> than the development communities desires.
I think this was far more the case when the TC was largely made up of
PTLs. I also think we need to admit that the TC has limited direct
authority, and what those of us on the TC get done in these areas is as
much about our personal authority, i.e. we're useful humans that
contribute broadly, so when run in one direction people give us the
benefit of the doubt and tend to help out.

I agree that I'd love to have these problems fully solved already. They
definitely are not. But there is meaningful effort going into them.

    -Sean

-- 
Sean Dague
http://dague.net


-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 465 bytes
Desc: OpenPGP digital signature
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20141102/d1bc2b01/attachment.pgp>