[openstack-dev] [nova] API work for ocata summit session recap
mriedem at linux.vnet.ibm.com
Fri Nov 4 20:07:39 UTC 2016
John Garbutt led a session at the summit focusing on some of the API
work items for Ocata. The full etherpad is here:
We talked a bit about the cross-project discoverable API spec. There was
also a xp session on that earlier in the week so John was reporting the
highlights of that session. There is some general consensus on the
approach, but the details are going to be hashed out in the API WG spec:
The goal for Ocata is just finding agreement in the API WG spec and
getting that merged so we can start focusing on implementation in Pike.
We deprecated a lot of the proxy APIs in the 2.36 microversion in
Newton, but there are still other proxy APIs left in Nova. Some are
straight-forward and we've already approved specs to clean them up, like
the image metadata proxy extension. But others like multinic and
floating IP actions and security group handling in server create
requests are more complicated so we've elected to defer those topics for
Ocata as we have higher priorities.
API extension folding
sdague started doing some of this in Newton and while it's fairly
mechanical it's not trivial work to do correctly or review, so it's
probably the type of thing that's going to be deferred to Pike.
We'll continue working the api-ref cleanups. We now have docs generated
for the sample policy file in the nova devref. The policy.yaml docs
could use descriptions for each policy, but we said we wouldn't start on
that until after the config option cleanup effort is done.
General loose ends and new ideas
There are several random API impacting specs up for review. We didn't
discuss many of these in great detail. The thing to take away from this
is we have several approved blueprints already that impact the REST API
and will require microversions, so we already have plenty to do (and
review). I'll also make a side note that some of the approved blueprints
don't have code up for review yet which is concerning.
Some new things that we wanted to focus on for Ocata was json schema
validation of query parameters, which is something we need to restrict
server sort/filter parameters. Today you can sort on anything including
joined tables to the instances table. We need to restrict this in
general as it's a security issue, but also because of it's impact on
cells v2 and how we're going to be handling querying instances across
multiple cells. Alex Xu is working on the spec for the json schema
validation and Kevin Zheng is working on the fixes.
Not really discussed in the session, but this is the rough list if
you're keeping track at home:
- query parameter validation
- restricting server sort/filter parameters
- agree on and get the API WG capabilities spec approved
More information about the OpenStack-dev