Kilo Project Priorities¶
List of priorities (in the form of use cases) the nova development team is prioritizing in Kilo.
For more information see: http://docs.openstack.org/developer/nova/devref/kilo.blueprints.html#project-priorities
Although the current Cells code is used in production by several large deployments, the code is difficult to maintain and the implementation is missing major features. The goal of this effort is to produce a replacement for the current cells models, so cells become a first class Nova citizen.
Moving to objects makes the code easier to read and more maintainable for developers and paves the way for online database migrations, one of the main goals in No downtime DB upgrades.
This paves the way to pulling out the scheduler allowing for faster scheduler development while reducing the scope of nova to help with nova’s growth challenges
Pave the way for a better user experience by moving towards API microversions.
Nova currently has unit tests and integration testing but practically no functional testing. This should make it easier to test and debug nova race conditions and edge cases.
Finish making neutron the preferred networking model so nova can deprecate nova-network and start the timer to removal. Reduces the scope of nova.
No downtime DB upgrades¶
Operators tell us one of the biggest pain points in upgrading is running the database migrations, so we are fixing that with online database migrations.
Nova is doing a bad job of managing bugs, a key way users provide feedback to the developers.
Test coverage, and specifically third party CI coverage, is always an issue. We frequently say we require it but we don’t do a great job of checking on status, i.e. is a job responding quick enough and is accurate, should it be voting or not, etc. Also, we ask for third party CI on new features but let things get merged without enforcing the third party CI, and once the code is in it’s hard to remove it.