[openstack-dev] [nova] Announcing my candidacy for PTL of the Ocata cycle
mriedem at linux.vnet.ibm.com
Wed Sep 14 21:38:51 UTC 2016
This is my self-nomination to continue running as Nova PTL for the Ocata
Looking back at Newton, the Nova team accomplished a lot. Random things that
stick out for me:
* Cells v2 in a single-cell deployment plus CI jobs.
* Placement API for tracking quantitative resources on compute nodes.
* Dedicated voting live migration CI job.
* os-vif integration for libvirt (linuxbridge and OVS backends).
* Glance v2 integration.
* nova-network deprecation (for real this time!)
* Server tags support (v2.26).
* Virtual device tags for libvirt and hyper-v (v2.32).
* Proxy API deprecation (v2.36).
* Get me a network (v2.37).
* API policy defaults in code.
* api-ref docs in tree and massively cleaned up.
* Continued cleanup and documentation of the configuration options.
* Ironic multiple compute host support.
* Ironic multiple tenant networking support.
* Libvirt live migration post-copy mode.
* Vendordata (version 2) API for the metadata service.
* Several notifications are now versioned.
* os-brick + oslo.privsep support.
* Improved third party CI for NFV (Intel and Mellanox).
* Deferred fixed IP allocation for Neutron ports.
* gate-tempest-dsvm-lvm job in the experimental queue.
* plus a lot more (100 approved blueprints, 64 complete or partially
I think we have some momentum going into Ocata, which is a shorter
continue some of the work from Newton. My personal priorities for Ocata are:
* Continue work on cells v2, specifically around multi-cell support and
cells v2 required in Nova deployments for the Ocata release.
* Continue work on the placement API, including making it required for Nova
deployments by the Ocata release. There will also be work on modeling
capabilities for the placement service.
* The libvirt imagebackend refactor which is a prerequisite for implementing
libvirt storage pools is going to need focus in Ocata. This was the one
priority from Newton that did not really "make it" and I want to come
a plan to push that forward, which probably includes dedicated review
improved test coverage (non-Tempest integration testing), and
make sure we are staying on track.
* Discoverable API support which builds off the policy in code and
* There were several blueprints which were close in Newton but missed
due to the non-priority feature freeze and I think we can get those
* Python 3 support since that's a cross-project OpenStack effort for Ocata.
* We will need to start taking a deeper look at improving workflows between
Nova/Cinder and Nova/Neutron. For Cinder this means the simplified
that John Griffith has been POCing which is a prerequisite for volume
multiattach support. For Neutron this means improved coordination of
like during live migration. John Garbutt has already started working
Neutron team on this.
Like in Newton, I would like to restrict spec reviews mainly to re-approvals
early so that we can close out things that are nearly already ready to go
before accepting new work, especially with the shorter cycle.
I will continue to foster the subteams that are working and holding
Nova. I think that has been going well and helps focus efforts which do not
have a dedicated core involved on a daily basis. I want to also continue
regular status checks in the mailing list and recaps of events, be those
summit sessions or even hangout sessions, so that the entire team is on the
same page and we have clear communication.
This is definitely not a one person show and I owe a lot to the people
in Nova every day. As I said in my Newton nomination I want to be PTL to
manage the project forward in Ocata and keep the development team focused on
getting work done, but I definitely can not and will not do that alone.
Thanks for your consideration.
More information about the OpenStack-dev