[openstack-dev] Planning for the Pike PTG

Thierry Carrez thierry at openstack.org
Wed Jan 4 10:16:36 UTC 2017


Matt Riedemann wrote:
> I haven't been living under a rock but I'm not aware of any major
> announcements regarding session planning for the PTG - has that happened
> somewhere and I'm just in the dark?
> 
> I'm mostly wondering about the Monday and Tuesday cross-project sessions
> - are those time-boxed sessions like at the summit and will have a
> schedule? Or are we just playing fast and loose and hoping someone will
> lead us out of a hallway and into a room for Major Synergy (tm)?

There are no "cross-project sessions" on Monday-Tuesday. There are a
number of horizontal team meetings (Infrastructure, QA, Documentation,
Security, Oslo...), transverse team meetings (Horizon, Kolla,
OpenStackClient, AppCatalog, RPM packaging...), and workgroup meetings
(Architecture WG, Stewardship WG, Interop WG...). All of those are full
days (or full 2-days) in a room owned by a given team (and PTL or chair)
and they are free to organize in whatever way they see fit (there are no
time-boxed sessions, so we expect most teams to use an etherpad-based
open agenda).

We'll also have a room (or two) dedicated to the Pike goals (currently
under discussion) -- whoever wants to meet and make quick progress on
the Pike goals during the PTG should be able to find facilitators there.
We are still waiting on the final list of goals to formally make
progress on that front.

Additionally from Monday to Thursday we'll have one openly-scheduled
fishbowl room, in case we need to have specific discussions. Think a
Cinder/Nova/os-brick discussion outside of the Nova and Cinder-specific
rooms, but for which you'd rather not all stand in the hallway. For that
room I thought we could set up an etherpad with time slots and let
people schedule topics there on the spot... But I'm happy to take
suggestions.

> I see project teams are working on getting etherpads together for
> topics, including myself, which got me thinking about how to plan the
> Wed-Friday sessions which for a midcycle meetup would normally be a list
> of topics that we'd go through in order (or by priority) but not
> time-boxed or scheduled. But then I got thinking about how the PTG is
> right before we start working on Pike, so I'm now thinking we need more
> structure than what we did at the midcycles, and more like what we do at
> the design summit with respect to scheduled discussions about things
> that are going to be worked on in the upcoming release, figuring out
> goals, determining review priorities, etc - which is actually a lot more
> work to plan and schedule ahead of time, especially when we consider
> (vertical) cross-project sessions like between nova/cinder or nova/neutron.

One of the goals of splitting the Design Summit into PTG & Forum is to
separate the "feedback/requirements gathering" phase (at the Forum) from
the "let's plan and bootstrap the actual work" phase (at the PTG). The
Pike PTG is arguably a transition PTG, since we won't have had a "Forum"
in the months before. The PTG is still happening at a point where most
people already started working on Pike though (rather than "right before
we start working on Pike"), and ideally should be focused on
implementation plans, review priorities and getting things done, without
the constraints of time-boxed slots.

That said, you should definitely take advantage of having everyone
around (and with less scheduling constraints compared to Summit) to
discuss inter-project questions (think Nova/Neutron or Nova/Cinder). You
can hold those within your room if you think all team members should
follow them, or take advantage of the aforementioned extra fishbowl room
to hold those.

> In other words, the fact I haven't had anxiety yet about planning the
> PTG makes me anxious that I'm falling way behind already.

I don't think you are way behind. Now is a good time to brainstorm on an
etherpad what your team needs to discuss and do during those days. If
you identify inter-project discussions, there is still time to reach out
to those other teams to make sure it's on their radar as well, and
arrange a common time for the discussion. I like to think we can achieve
that without the stress and constraints of strict centralized
scheduling, using a more peer-to-peer/unconference approach to magically
make the best use of our time there. If the feedback at the end of the
week is that it was a big bowl of mess, we'll revisit for the next one :)

Cheers,

-- 
Thierry Carrez (ttx)



More information about the OpenStack-dev mailing list