[openstack-community] [upstream] Vancouver upstream training session debrief

Stefano Maffulli stefano at openstack.org
Fri May 29 17:48:04 UTC 2015


Hello folks,

first of all thank you to all instructors and volunteer staff and
mentors that helped in Vancouver: this program would not be possible
without the kind donations of your time and effort.

I'd like to collect notes and comments on what went well and what should
be changed for the next events. I'll start (quite long):

Slides & content
================

- I think the rst format and common repository worked well. We miss one
(set of) slides to illustrate the whole 2 days program. Maybe copying
the content over from the wiki Info page[1] to the index.rst page would
be a good thing.

- the index.rst is not formatted properly. Christian Berendt is working
on this and may need a help from sphinx/hieroglyph gurus
https://review.openstack.org/#/c/185946/

- the draft site[2] is great. I think that keeping the slides in /draft/
is good enough since the status of those slides may not be always up to
date with the authoritative source of information like governance.o.o or
infra/manual

- with the slides in .rst format and in a git repo, I think they could
be hooked to a translation system. I'd consider hooking up and translate
the material in other languages. Japanese would be the most likely
candidate, and people from China have expressed also interest.

- suggestions and commentts from the participants:

   + Maybe work through some more demos - ie, run through committing a
change to sandbox doing it "live" from the presenter desk
   + Point out the differences between docs and code reviews -
environment prerequisites
   + Talk about gating checks - ie, what the automated checks are
actually checking for.
   + It might be worth considering recording it and making it like a
Coursera course or a Kahn Academy course.
   + The slideshow can improved. It should contain more graphics (ex:
apc/atc relation), clickable links (ex: stackforge).
   + It would be great to collect a list of bugs which are really "low
hanging" fruits, so that they can be solved during the two days.
   + Teach about the source code. How it's organized? Why is it so?
After rpm/deb packaging, where do files go on the root filesystem? How
do I run tests? How do I write my own tests? What if I need to introduce
a new file? How are the python packages arranged? How do subprograms
communicate to each other?
   + consider separating software developers and technical writers

One more comment led me to believe that we have not done a good job at
communicating that the training is 2 days live + mentoring sessions
afterwards. See below for ideas on how to fix this.

Administrative process
======================

- Planning attendance is hard. Of the 85+ people who originally signed
up, about 45 showed up. I noticed that about 10 people removed
themselves when we started sending notifications in preparation to the
live session. I think that the barrier to sign up is too low and we can
raise it so that only highly motivated people sign up to attend. The low
barrier also makes it hard for people to read and understand that the
training is 2 days live + online mentoring afterwards.

  + To set the bar higher, I think we need to find a way to scale the
method used by Loic. He engages with every applicant immediately, as
they sign up, and if they don't respond to his solicitations he removes
them from the list. That's very intense and time consuming but ends up
getting more motivated people. I think this can be scaled by having a
larger pool of mentors that gets to solicit applicants. A form of light
automation using Google Forms and Trello (with glue provided by
zapier.com) may be enough? Needs experimenting.

  + To make it clear that the training is 2 days + live sessions I think
we should have people engage with mentors and set dates for meetings
*before* the training starts.

- I think the Trello board is a good step forward: it allows staff and
management to follow progress of each student.


Format
======

- I'm starting to think that to days is probably too much for some
people. The comments I have received from the survey are all extremely
positive but a couple of written comments make me believe that trying to
shorten the class to 1 day and 1 extra day for custom deep dives would
probably give more value. This deserves a thread on its own.


Venue & logistics
=================

- Participants and staff should have their names printed to facilitate
social interactions
- Renting the Legos was the best idea ever. Definitely worth considering
for the future.

Thanks again,
stef

[1] https://wiki.openstack.org/wiki/OpenStack_Upstream_Training/Info
[2] http://docs.openstack.org/draft/upstream-training/#1



More information about the Community mailing list