Open Stack

Excerpts from Jean-Philippe Evrard's message of 2016-12-15 09:11:12 +0000:
> Hello,
> 
> Maybe this will sound dumb …
> 
> I received this email on openstack-dev mailing list. I don’t know if it was sent to any other place, because it’s basically agreeing on development to be done, which makes sense to me.
> So openstack-dev people (called further “devs”) will push their company agenda on these goals based on what they know in their company. I see the work done together there, and I find it great, but…
> 
> Wouldn’t that be better if we open this discussion to the general population (openstack users, operators, and devs) instead of just devs?
> I submit this question because what I see on https://etherpad.openstack.org/p/community-goals is not only tech debt items that we have to fix, but also ideas of improvement on the long run for our users.

Excellent point.

Collecting enough information to be confident that we are choosing
goals that are important to the whole project while still being
actionable is going to be a challenge. I expect us to get better
at it over time, but this is only the second time we've tried.

Emilien has also started talking with the Product Working group
[1], which is chartered by the board to collect this sort of feedback.

As far as a "company agenda" goes, I hope we will be able to identify
those items as either too narrowly scoped (into one project, or a
minor feature), or of narrow interest (something one customer wants)
to avoid them. On the other hand, just because an idea starts out
with one supporter doesn't mean it's necessarily bad. It may just
mean that it takes more time to build up enough support to become
a community goal. That's one benefit of keeping the backlog: We can see
what has been proposed in the past, until older ideas become more
relevant or mature into something that we can implement.

[1] http://lists.openstack.org/pipermail/product-wg/2016-December/001372.html

> It makes sense to me to keep the tech debt items as a devs only topic (I don’t see why a user cares _at least at first_ about using oslo.privsep vs oslo.rootwrap), and it makes also sense to me to align “community goals” with the broad community.
> 
> What do you think? Should we remove the non tech-debt items in this dev-community-goals etherpad?

It's up to us to ensure that operators and users understand the
importance of those technical debt items so we can all prioritize them
together.

> Should we have another set of community goals that could serve as a basis for the OpenStack user survey?

I'm not sure a simple survey is a useful way to collect this sort
of information. I would expect everyone to answer yes to features
and no to anything else. A tiny text box also doesn't give enough
space to describe goals in sufficient detail to make them actionable.
Engagement through discussion allows for more collaborative decision
making, because we can argue the relative merits of proposals. It
also allows us to have more detailed writeup, which will help
identify things we are ready to finish versus things that need more
experimentation, examples, or refinement.

> Or should we keep these goals merged together, with the risk of having tech-debt items having lower priority the user requirements? (For that, the TC would be a good judge for final cycle prioritization)

Yes, that's why the TC makes the final call, after considering all of
the input. Sometimes we will need to take the timing of a goal into
account, and pay down debt before we add a feature.

> I think having community goals is great for openstack, and I’d be happy to understand how we’ll adapt http://governance.openstack.org/goals/index.html into real life work usable for everyone.
> 
> Thanks for your clarifications.

Thanks for these questions!

Doug

> 
> Best regards,
> Jean-Philippe Evrard
> 
> 
> On 12/12/2016, 12:19, "Emilien Macchi" <emilien at redhat.com> wrote:
> 
>     On Tue, Nov 29, 2016 at 7:39 PM, Emilien Macchi <emilien at redhat.com> wrote:
>     > A few months ago, our community started to find and work on
>     > OpenStack-wide goals to "achieve visible common changes, push for
>     > basic levels of consistency and user experience, and efficiently
>     > improve certain areas where technical debt payments have become too
>     > high – across all OpenStack projects".
>     >
>     > http://governance.openstack.org/goals/index.html
>     >
>     > We started to define a first Goal in Ocata (Remove Copies of Incubated
>     > Oslo Code) and we would like to move forward in Pike.
>     > I see 3 actions we could take now:
>     >
>     > 1) Collect feedback of our first iteration of Community Goals in
>     > OpenStack during Ocata. What went well? What was more challenging?
>     >
>     > Some examples:
>     > - should we move the goal documents into a separate repo to allow a
>     > shorter review time, where we could just have 2 TC members approve
>     > them instead of waiting a week?
>     > -  we expected all teams to respond to all goals, even if they have no
>     > work to do. Should we continue that way?
>     > - should we improve the guidance to achieve Goals?
>     >
>     > I created an etherpad if folks want to give feedback:
>     > https://etherpad.openstack.org/p/community-goals-ocata-feedback
>     >
>     > 2) Goals backlog - https://etherpad.openstack.org/p/community-goals
>     > - new Goals are highly welcome.
>     > - each Goal would be achievable in one cycle, if not I think we need
>     > to break it down into separated Goals (with connections).
>     > - some Goals already have a team (ex: Python 3) but some haven't.
>     > Maybe could we dress a list of people able to step-up and volunteer to
>     > help on these ones.
>     > - some Goals might require some documentation for how to achieve it.
>     >
>     > I think for now 2) can be discussed on the etherpad, though feel free
>     > to propose another channel.
>     >
>     > 3) Choose Goals for Pike.
>     > Some of us already did, but we might want to start looking at what
>     > Goals we would like to achieve during Pike cycle.
>     > I was thinking at giving a score to the Goals, that could be
>     > calculated by its priority (I know it's vague but we know what is
>     > really urgent for us versus what can wait 6 months); but also the
>     > number of people who are interested to contribute on a Goal (if this
>     > Goal doesn't have a team yet).
>     > For now, openstack/governance is the repository for Goals, please
>     > propose them here.
>     >
>     >
>     > Please give feedback, we're doing iterations here, and hopefully we'll
>     > improve our Community Goals over the next cycles.
>     > Thanks for your time,
> 
>     Two weeks happened, here's a digest version of the etherpad:
> 
>     - Most of projects achieved the goal for Ocata, and we saw strong
>     interest to do it on time
>     - Some confusion between the ACK'ing of a goal, and actually doing the work.
>     - Some projects were slow on the uptake (of starting the work) and
>     even reviewing the patches.
>     - For now, keep using openstack/governance repo for documenting Goals.
>     - Improve guidance on what projects are expected to do when updating
>     the status of the Goal.
>     - For each Goal, document who the "guides" are and how to find them
>     when help is needed.
>     - It seems like achieving multiple Goals in a single cycle wouldn't be
>     possible for all teams, we could prioritize them to let teams achieve
>     more than one Goal within a cycle.
> 
>     What's next?
>     https://etherpad.openstack.org/p/community-goals
>     Now that we have a good set of Goals that are proposed in this
>     etherpad, we might want to rank them by priority (1 is the most
>     important). Feel free to do it in the etherpad, by putting a rank in
>     "Priority rank".
> 
>     Also, I've noticed some Goals might be too big to be achievable within
>     a single cycle and might need to be split (Rolling upgrades for
>     example). If you're author of one these goals, please do so.
>     I hope we can start defining Pike Goals by next week, so we can start
>     documenting what we would expect and the guidance to achieve it/them.
> 
>     Any feedback is welcome,
>     --
>     Emilien Macchi
>