----- Original Message ----- > > > Hi, folks, > > > > My usecase follows: > > 1. create two vms A and B by using the ports that have been created. > > 2. vm A can ping vm B > > 3. Delete one port of A or B > > 4. vm A can still ping vm B > > > > IMO, ping should not be ok when vm's port have been deleted. > > > > Two alternative solution: > > 1. do more restriction in def prevent_l3_port_deletion(self, context, > port_id), we should not allow to delete the ports in use. > > 2.permit to delete the port but notify ovs-agent to unbind the port. >
I think option 2 is better. Implementing option 1 within Neutron would imply getting the device_id, and sending an API call to Nova to see if it's in use or not. Introducing more Neutron --> Nova interactions seems like a really bad idea. Additionally, I'm trying to lead an effort of cleaning up tenant resources when one is deleted. Here's the spec for Neutron: https://review.openstack.org/#/c/98097/ If and when Nova implement something similar, prevention of deletion of in-use ports would imply a dependency/order between projects when deleting a tenant. > > > Any thoughts? Looking forward to your response. > > > > Cheers, > > XuRong Yang > > > > _______________________________________________ > OpenStack-dev mailing list > OpenStack-dev@lists.openstack.org > http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev > _______________________________________________ OpenStack-dev mailing list OpenStack-dev@lists.openstack.org http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
