Manage openVswitch via NetworkManager

[Cristian Le]

Debugging network issues on the nodes managed directly via ip tools is unintuitive to newcomers who are more familiar with NetworkManager interface. For example:
After a restart I am affected by [bug#1932143](https://bugs.launchpad.net/tripleo/+bug/1932143), showing `br-int` and similar interfaces down and disconnected. However it is unclear whether this is a normal working state or not.

What are some of the limitations with NetworkManager that would stop this from being implemented?

[Brendan Shephard (bshephar)]

Hey Crisitan,

Yes, this is quite annoying. We are still shipping the old network-scripts, even with CentOS9 precisely because of this limitation. Unfortunately, there just hasn't been much progress on the NetworkManager and OVS integration that would sufficiently cover our use case in TripleO.

It's something we have tried to work around using nmstate configuration of the interfaces and OVS bridges, but so far we haven't had it working reliably enough to ship a solution with TripleO:

Noting:
https://review.opendev.org/c/openstack/os-net-config/+/786323
https://review.opendev.org/c/openstack/os-net-config/+/642975
https://review.opendev.org/c/openstack/os-net-config/+/791409/4

Long story short is that this is a work in progress. There will be some limitations to using nmstate vs the legacy network-scripts, but it's something we're working towards nonetheless.