[openstack-dev] [elections][shade] PTL candidacy for Queens
mordred at inaugust.com
Mon Aug 7 14:24:44 UTC 2017
I would like to be the PTL of the shade project for another cycle.
Over the past few years the shade project has grown from a utility
library containing some logic needed by the Infrastructure team and
Ansible for interacting with OpenStack clouds to a rich SDK that serves
the general needs of humans who want to use OpenStack.
This past cycle we (mostly) accomplished two main goals that were not
exciting features but which are immensely important nonetheless. We
ported our unittests to use requests-mock instead of mocking out
individual python client libraries, and we replaced our use of the
python client libraries with direct REST calls based on keystoneauth.
This was rather mind-numbing and I fear we may have eaten the brains of
a few wonderful contributors - but it means that our dependency chain is
much slimmer so inclusion in distros and alongside apps should be less
We still need to finish converting the last keystoneclient calls and to
transition the Ironic calls. TheJulia did WAY too good of a job writing
tests when implemeting Ironic support originally, so it may take a
little while to finish that one.
Over the next cycle we need to update our REST layer to incorporate the
recent changes made in keystoneauth related to service and version
discovery. This largely means removing code, which is always pleasant.
Once that's done we'll be in a position to start using microversions as
Glance has landed experimental support for the new Image upload process,
and we need to add detection and implementation of that to our Image
upload system so that we'll use it when it's available.
There is some very important work we need to do deep in the guts related
to caching and batching of calls. We have a system for this that is
designed to ensure that large-scale systems such as Nodepool can operate
effectively, but it's proving costly for smaller users and needs to be
I recently suggested merging the work of the python-openstacksdk team
into shade. If we decide as a larger OpenStack community that this is a
direction we want to go, we'll need to do a decent amount of plumbing to
ensure that's as smooth as possible for both sets of users.
Finally, I think if we don't make some forward progress on the oaktree
gRPC federation API that samueldmq will be very unhappy with me. We've
been explicitly putting off working on that to ensure we get the
RESTification work done. This next cycle I expect to make real progress
on that so that we can at least have a proof point to look at so we can
discuss a real thing rather than a theory.
I'm proud of what we've done so far, and I think there is a ton we can
continue to do to serve OpenStack End Users well.
Thank you for your consideration,
More information about the OpenStack-dev