inconsistency in documenting the use of eth12

[Jonathan Rosser]

This describes eth12 as being used to support one flat neutron network, via the br-vlan bridge https://github.com/openstack/openstack-ansible/blob/master/etc/openstack_deploy/openstack_user_config.yml.example#L232-L233

This seems to be contradicted by the text from the example config https://docs.openstack.org/openstack-ansible/latest/user/test/example.html

# For tenant vlan support, create a veth pair to be used when the neutron
# agent is not containerized on the compute hosts. 'eth12' is the value used on
# the host_bind_override parameter of the br-vlan network section of the
# openstack_user_config example file. The veth peer name must match the value
# specified on the host_bind_override parameter.

which on the face of it says that eth12 is integral to tenant vlan support.

The distinction between vlan and flat networking is unclear in the documentation, made more complicated by them sharing the same bridge.

The purpose of eth12 is unclear and mismatched between the example config and the explanatory text.

[Alexander Neck (anymorrow)]

What is the correct solution? I try to roll out an Openstack-Ansible rocky Production Ceph envoirement(10hosts - one is a seperated deployment-host). So far just problems :(

i tryed to follow the instructions closely, but can't manage to get it running.

When i leave out the part about the veth-pairing in network config, i can execute the playbooks successfully(all 3). But in Horizon i cannot manage the networks... - the neutron-linuxbridge-agenst crashes and reboots every 5s seconds on the infra-nodes... error is a missing interface eth12

if i uncomment the part about veth-pairs the interface eth12 gets created and the neutron-linuxbridge-agent starts at least and keeps running. But in this configuration i cannot execute the playbooks succsessfully anymore... it crashes because the containers cannot update the apt-cache anymore :(

can anybody explain to us, what is the correct config? Is the veth-pair needed on infrastructure-nodes?
i followed this: https://docs.openstack.org/openstack-ansible/rocky/user/ceph/full-deploy.html

[Filippo DiNoto (fdinoto)]

I think the part of your config that is causing the problems is:

host_bind_override: "eth12"

If your metal hosts have a br-vlan, then you can just remove that. If you opted not to make the br-vlan bridge on your non-container hosts, then you would change that value to the physical interface that would equate to br-vlan on your container hosts.

The veth interface name can be set to anything sensible, there is no OSA standard to use 'eth12' for any specific purpose.

You don't need to use both 'flat' and 'vlan' provider networks. For simplicity, you may just want to remove the flat provider network definition. The difference between them is just that one allows you to configure the segment ID (VLAN ID) either from within OpenStack ('vlan') or not from within OpenStack ('flat'), which would require you to configure all the VLAN sub-interfaces on the linux host manually.