The inventory section in the developer docs could do with a bit more detail about how the dynamic inventory generates the container entries.
From IRC today:
[13:26:22] odyssey4me: wadeholler the dynamic inventory generates container names and IP addresses based on input from the files in /etc/openstack_deploy/env.d/
[13:27:03] odyssey4me: wadeholler if a system is marked with the flag 'is_metal: true' then that service is ignored as it's expected that the service will run on a host, not in a container
[13:27:55] odyssey4me: wadeholler where the containers are located depends on what is specified in /etc/openstack_deploy/openstack_user_config.yml and/or the files in /etc/openstack_deploy/conf.d/
[13:29:34] odyssey4me: effectively if you specify that the host for the 'identity_hosts' group is on my-identity-server1 and my-identity-server2 - similar to https://github.com/openstack/openstack-ansible/blob/master/etc/openstack_deploy/openstack_user_config.yml.aio#L99
[13:29:52] wadeholler: got it. thank you odyssey4me! thats (env.d) is the piece I was missing
[13:30:12] odyssey4me: then the dynamic inventory finds the group called identity_hosts in the environment configuration, eg: https://github.com/openstack/openstack-ansible/blob/master/etc/openstack_deploy/env.d/keystone.yml#L38
[13:31:17] odyssey4me: and seeing as the service doesn't have the is_metal flag set, it'll default to 'is_metal: false', which means that the service is assumed to be running in a container
[13:31:46] odyssey4me: so the dynamic inventory will then create a host entry and ip address for that service container to be located on those hosts
Fix proposed in: https:/