62 lines
3.3 KiB
Plaintext
62 lines
3.3 KiB
Plaintext
Hi everyone,
|
|
|
|
This is my self-nomination to continue running as Nova PTL for the Pike cycle.
|
|
|
|
If elected, this would be a third term for me as Nova PTL. In Ocata I thought
|
|
that I did a better job of keeping on top of a broader set of efforts than I
|
|
was able to in Newton, including several non-priority vendor-specific
|
|
blueprints.
|
|
|
|
I have also tried to make regular communication a priority. The topics vary,
|
|
but in general I try to keep people informed about the release schedule,
|
|
upcoming deadlines, areas that need attention, and recaps of smaller group
|
|
discussions back to the wider team. We have a widely distributed team and a lot
|
|
of groups are impacted by decisions made within Nova so it's important to
|
|
continue with that communication. Despite my best efforts I have also learned
|
|
in Ocata that we need to get earlier feedback on changes which impact
|
|
deployment tooling, and make documentation of such changes a high priority
|
|
earlier in the development of new features so that people working on tooling
|
|
are not left in the dark.
|
|
|
|
Ocata has been a tough release, and I think we knew that was going to be the
|
|
case going in. It was a shorter cycle but still had some very high-priority
|
|
and high-visibility work items such as integrating the placement service with
|
|
the scheduler and further integrating support for cells v2, along with making
|
|
both of those required in a Nova deployment for Ocata. We also had to deal
|
|
with losing some key people and filling those spots. But people have stepped up
|
|
and we still made some incredible progress in Ocata despite the difficulties.
|
|
|
|
For Pike I want to focus on the following:
|
|
|
|
* Continue integration of the placement service into making scheduling
|
|
decisions, including working with Neutron routed networks and work on
|
|
defining traits for resource providers so we can model the qualitative
|
|
aspects of resources in making placement decisions.
|
|
* Continue working on cells v2 for multi-cell support including investigating
|
|
the concept of auto-registration of compute nodes to simplify deployment
|
|
automation, and also focus on multi-cell testing and Searchlight
|
|
integration.
|
|
* Work on volume multi-attach support with the new Cinder v3 APIs introduced in
|
|
Ocata for creating and deleting volume attachments. I think we are finally at
|
|
a place where we can make some solid progress on the Nova side with improved
|
|
understanding between the Nova and Cinder teams.
|
|
* There are going to be several efforts going on across several projects in
|
|
the Pike release, including modeling capabilities in the REST API, and Nova
|
|
is going to have to be a part of those efforts. We also need to get teams
|
|
together to figure out what are the issues with hierarchical quotas and what
|
|
progress can be made there since that is a high priority item that lots of
|
|
operators have been requesting for a long time.
|
|
|
|
In general, we are going to have to improve our review throughput, especially
|
|
given the change in resources we experienced in Ocata. To me, a lot of this
|
|
will have to do with mentoring people that are newer to Nova but are stepping
|
|
into leadership positions, and having a shorter feedback loop on "leveling up".
|
|
|
|
To summarize, I aim to be of service to those using and contributing to Nova
|
|
and want to continue doing that in the PTL role for the project in the Pike
|
|
release if you will have me for another round.
|
|
|
|
Thank you for your consideration,
|
|
|
|
Matt
|