Go back to see the policy.json and the behavior in my environment make sense.

My environment works out tenant A can use tenant B's network NB. I don't think it's reasonable.

Here's a senario,

Tenant A has network "NA". Tenant B has network "NB".

When booting vm with OS_TENANT_NAME "A" , can network NB used for vm creation under this scenario? My assumption is no.

Thanks ZhiQiang Fan, that solved my question.

I don't see any neutron function using slave. But in neutron, there's method for get_session in https://github.com/openstack/neutron/blob/master/neutron/openstack/common/db/sqlalchemy/session.py (https://github.com/openstack/neutron/...)

That's why I got confused here.

From your reference[0], other projects should get session from oslo get_session(), right? I looked at cinder,nova and neutron, all of them define and consume their own get_session function instead of oslo's. This is potential code refactor work here?

Thanks for explanation!

In neutron.conf, for section [database], there's two connection as below: ===================[database]================================ [database]

This line MUST be changed to actually run the plugin.

Example:

connection = mysql://root:pass@127.0.0.1:3306/neutron

Replace 127.0.0.1 above with the IP address of the database used by the

main neutron server. (Leave it as is if the database runs on this host.)

connection = sqlite://

The SQLAlchemy connection string used to connect to the slave database

# slave_connection =

What's the role of slave_connection ?