[openstack-dev] [glance] PTL candidacy

Brian Rosmaita brian.rosmaita at RACKSPACE.COM
Mon Sep 12 18:26:59 UTC 2016

I've just put up https://review.openstack.org/#/c/368995/, declaring my
candidacy for Glance PTL.

Here's the content of that patch for your convenience:

Hello everyone,

I would like to announce my candidacy for the role of PTL of Glance for
the Ocata cycle.

Current State of Glance

This is a good time to be PTL of Glance.  The project has recently
served as the incubator for a successful project (Searchlight) and a
hopefully soon-to-be successful project (Glare), and at this point we
can focus on being Glance [0].  In Newton, the Images v1 API is being
deprecated [1] and Nova will be using the v2 API by default [2], which
allows us to focus even further.

Glance has some important efforts in progress that need to be seen
through to completion, so given the short upcoming cycle, I anticipate
quick agreement in determining our Ocata priorities.  In short, I see
myself as a "continuity" candidate, building on the good work done by
Flavio and Nikhil in the previous cycles.

Glance Priorities

This is how I see things shaping up for Ocata.

1. Image Import

The Glance image import refactoring proposal [3] has been discussed
thoroughly through two cycles by all OpenStack groups and individuals
interested in a discoverable, interoperable means of getting end-user
created images into OpenStack clouds of various types (public/private,
large/small).  There are some patches up, and Nikhil has indicated
that image import is a priority for post RC-1 time; I want us to
continue that work and deliver image import in Ocata.

2. Community Images

Implementation of the Community Images spec [4], approved for Newton,
has been delayed, in large part due to reviewer bandwidth issues (which
I'll address below).  It's a much-requested feature, has several
patches up, and is a priority to deliver early in Ocata.

3. Rolling Upgrades

The OpenStack Innovation Center has been promoting various operator
friendly enhancements to OpenStack, one of which is zero-downtime
control plane upgrades.  Several Glance developers associated with OSIC
have been working out a strategy (and some alternatives) to make this
happen in Glance [5,6].  I expect we'll be able to converge on a
strategy during (or shortly after) the Barcelona Summit with the goal
of having a zero-downtime Newton-to-Ocata upgrade available in Ocata.

(These three items are high-impact for operators and end users, so are
appropriately our top three Ocata priorities.  I have serious concerns
about Glance being able to deliver more than this in Ocata, so please
notice the caveats attached to the following items.)

4. Community Priorities

The TC recently adopted "community goals" for each cycle [7], and as a
good community citizen, the Glance project will want to adopt these as
priorities for Ocata.  The goal "remove copies of incubated Oslo code"
[8] primarily affects the python-glanceclient.  The goal "support
python 3.5" [9] is still under discussion in the wider community, with
some projects reporting that it will be a stretch to implement in
Ocata.  It will also be a stretch for Glance given our resource
constraints.  (I think the major impact for Glance will be in getting
the functional tests to pass under Python 3.5, but I haven't done a
complete analysis of this.)

5. Other

As noted earlier, this is a short cycle. The above is quite a list,
though that's mitigated a bit by some of the items already being in
progress.  There are some other candidates for Glance priorities that
we can discuss carefully at the Summit [10], in particular, the Glance
Store Refactor and Hierarchical Image Access.  Personally, I would like
to see us at least (a) come to a consensus on, and (b) articulate
clearly the envisioned usage of the glance_store library during the
Summit as that will be helpful in guiding the refactoring of its API
and guide its future development (whether or not that development
happens in Ocata).  Hierarchical Image Access would complete the Glance
image sharing use cases, and shouldn't be too bad to implement given
the changes made for the Community Images implementation, but we need
to make a realistic assessment of this at the Summit.

Finally, the glance-specs repo contains some approved but not
implemented specs and lite-specs.  These are useful items (or they
wouldn't have been approved in the first place!).  They will also have
to be the subject of a realistic assessment during the Summit.

Glance Community

I said earlier that this is a good time to be Glance PTL; it's also a
good time to join the Glance community, particularly for ambitious
developers who would like to achieve core status on one of the major
OpenStack projects.  We need more reviewers as, right now, most of the
core reviewers for Glance have additional commitments.

Although we need more reviewers, I don't want to "clean house" because
our core reviewers represent a cross-section of the community and bring
a lot of expertise in varying realms to the project [11].  Neither do I
want to introduce any kind of "fast track" to becoming a Glance core
contributor.  So what's this great opportunity I'm talking about?

What I would like to do is request that core reviewers concentrate on
patches associated with declared Glance priorities, and not review
other patches unless they already have been +1'd.  That way, our
priorities get the attention they need, and people working toward core
can demonstrate the quality of their code reviews independently of a
core reviewer having already looked at the code.  Further, a thoughtful
-1 with good suggestions is probably the most useful kind of review, so
I'd like to allow noncores the first crack at demonstrating their
competency by posting some thoughtful -1 reviews along the way to a
patch getting a +1.  Additionally, not all the cross-project liasons
need to be core reviewers, so where appropriate, serving as a
cross-project liason is another way for developers who want to
eventually become cores to step up and take more responsibility for the
project.  (We'll be reviewing the cross-project liason expectations and
commitments [12] very early in the Ocata cycle.)

(Basically what I'm saying is that Glance needs some new core
reviewers, and we're arranging things to help facilitate motivated
developers in this regard.)

As far as process goes, I don't anticipate any major changes.  The spec
and spec-lite process for Glance has evolved over the past few cycles
and seems to be working OK.

About Me

I've been an ATC to OpenStack since Folsom.  I'm currently a core for
Glance and Searchlight.  I've worked with Glance both from the
perspective of a product manager and (currently) as a software
developer.  I've been around the project for a while and have
represented Glance in interactions with various teams, for example, the
API WG, the Product WG, the documentation team, OSIC, and DefCore.  My
"real life" cloud experience has been working with large scale clouds,
namely the Rackspace First Generation Cloud and the current OpenStack
based Rackspace Public Cloud.  I've spoken with my managers and, if
elected, I will be able to work upstream full time.

I have a real passion for working on OpenStack, especially Glance,
which I know some people think is crazy, but there it is.  Glance is an
important part of the OpenStack ecosystem and it has a community of
high-quality contributors who I've always enjoyed working with.  I'm
asking you for the opportunity to work with the Glance community in a
new way as Glance PTL for the Ocata release.

Thanks for reading this far.  Before I shut up, I'd like to thank all
the Stackers I've spoken with over the past few weeks, as a lot of the
ideas presented above come from those conversations.

Thank you for your consideration,
Brian Rosmaita

 [5] https://review.openstack.org/#/c/331489/
 [6] https://review.openstack.org/#/c/331740/
 [9] https://review.openstack.org/#/c/349069/
[10] https://etherpad.openstack.org/p/ocata-glance-summit-planning
[12] https://wiki.openstack.org/wiki/CrossProjectLiaisons

More information about the OpenStack-dev mailing list