[occm] Replace call to Nova os-interfaces with direct Neutron call

[mdbooth]

What this PR does / why we need it:
os-interfaces is a proxy API for neutron's ports list. This has 2 significant disadvantages to us:

• It adds the overhead of an extra API call every time it is used
• It doesn't contain all the information returned by Neutron

This is a non-functional change, although is does reduce the load on nova-api without reducing functionality.

The primary motivation for this is to enable additional filtering using information from ports, e.g. filtering by port tag or the addition of subports as in #2228.

Release note:

NONE

[mdbooth]

/cc @stephenfin @dulek

[k8s-ci-robot]

@mdbooth: GitHub didn't allow me to request PR reviews from the following users: stephenfin.

Note that only kubernetes members and repo collaborators can review this PR, and authors cannot review their own PRs.

In response to this:

/cc @stephenfin @dulek

Instructions for interacting with me using PR comments are available here. If you have questions or suggestions related to my behavior, please file an issue against the kubernetes/test-infra repository.

[stephenfin]

To the best of my knowledge, the only thing exposed by the nova /server/{server}/os-interface API that is not exposed by the /ports?device_id={server} API is the device tags. If you don't need those, this should be a no-op.

Edit: The code agrees.

[mdbooth]

To the best of my knowledge, the only thing exposed by the nova /server/{server}/os-interface API that is not exposed by the /ports?device_id={server} API is the device tags. If you don't need those, this should be a no-op.

Edit: The code agrees.

To be clear, this is a tag on the attachment, which is independent of the port. We still get port tags. I'm guessing only the former is available through the metadata service? For the purpose I had in mind I don't think it makes any difference which one we use.

[jichenjc]

/lgtm

[mdbooth]

/hold for @dulek's review comments.

[mdbooth]

/hold cancel

[kayrus]

@mdbooth I don't think it's a good idea. We need to keep the interfaces order, listing neutron ports doesn't provide the exact order. See #863

[mdbooth]

You're thinking of server.Addresses, which preserves order. This change doesn't affect that. The API it replaces also doesn't guarantee order.

For confirmation, you can check the code which was linked above: https://github.com/openstack/nova/blob/27.1.0/nova/api/openstack/compute/attach_interfaces.py#L74-L106

Nova is simply proxying this neutron call, and returns results in the same order as neutron.

server.Addresses additionally uses Nova's internal network_info cache, which is how it preserves order.

[kayrus]

@mdbooth interesting. so it looks like the neutron default order is a timestamp, right? When an interface is attached to a node, neutron creates a port and the port gets the recent timestamp. If this order corresponds to the os-interfaces order, then I'm fine with a change ;)

[dulek]

/lgtm

[mdbooth]

@zetaab @jichenjc Could we merge this?

[zetaab]

wow this looks interesting. Thank you!

/approve

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is APPROVED

This pull-request has been approved by: zetaab

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• pkg/openstack/OWNERS [zetaab]

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment