Failed to load VM in multi-node setup

[Yuli]

After creating a VM in command line it was not working
and I found an exception in the log file:

[-] Dragonflow application 'DHCPApp' failed in add_local_port
Traceback (most recent call last):
   File "/opt/stack/dragonflow/dragonflow/controller/dispatcher.py", line 46, in dispatch
     handler(*args, **kwargs)
   File "/opt/stack/dragonflow/dragonflow/controller/dhcp_app.py", line 419, in add_local_port
     if not self._is_dhcp_enabled_for_port(lport):
   File "/opt/stack/dragonflow/dragonflow/controller/dhcp_app.py", line 354, in _is_dhcp_enabled_for_port
     subnet = self._get_subnet_by_port(lport)
   File "/opt/stack/dragonflow/dragonflow/controller/dhcp_app.py", line 332, in _get_subnet_by_port
     subnets = l_switch.get_subnets()
 AttributeError: 'NoneType' object has no attribute 'get_subnets'

[Yuli (stremovsky)]

I was researching this bug and found the following.

I was creating VM as an admin user and admin group (tenant_a).
When creating the VM I assigned it to the "private" network
(tenant_b).

So, when a local port for VM is created, it is assigned to tenant_a.
While, it's switch is assigned to tenant_b.

In log I see that the system is subscribed to tenant_a topic
(and not to tenant_b).
Subscribe topic: tenant_a by lport

As a result, nothing is returned for: self.db_store.get_lswitch(l_switch_id)

[Yuli (stremovsky)]

When running the same command as a demo tenant everything works.

[Yuli (stremovsky)]

When creating a VM as an Admin user and Admin tenant,
I can not select the private network as it assigned
to the demo tenant.

To prevent the above bug, I suppose we need to prevent
creating of this port at the higher level API.