Trunk bridges aren't removed

[Slawek Kaplonski]

Recently I found out on the devstack that when I have VM with trunk port and some subports connected to it, trunk bridge isn't deleted when VM is migrated to the another host.
I think that the same thing will happen when instance will be simply deleted as vif object in case of trunk port (when ML2/ovs is used on Neutron's side) is objects.vif.VIFOpenVSwitch and _unplug method for this type of vif object don't tries to delete bridge at all.

[Rodolfo Alonso (rodolfo-alonso-hernandez)]

Hello:

As commented in IRC, Nova needs to know if a port belongs to a trunk (actually if this port is the trunk parent port). Neutron provides this info in the port GET result. From the result dictionary in the HTTP request: http://paste.openstack.org/show/807221/

"port.trunk_details" contains the trunk ID and the subports info.

Sean, do you need something else?

Regards.

[Bence Romsics (bence-romsics)]

Hi Slawek,

In my ml2/ovs based multihost devstack I could not reproduce this at first. What are we doing differently?

$ openstack network create net0
$ openstack network create net1
$ openstack network create net2

$ openstack subnet create --network net0 --subnet-range 10.0.4.0/24 subnet0
$ openstack subnet create --network net1 --subnet-range 10.0.5.0/24 subnet1
$ openstack subnet create --network net2 --subnet-range 10.0.6.0/24 subnet2

$ openstack port create --network net0 port0
$ parent_mac="$( openstack port show port0 -f value -c mac_address )"
$ openstack port create --network net1 --mac-address "$parent_mac" port1
$ openstack port create --network net2 --mac-address "$parent_mac" port2

$ openstack network trunk create --parent-port port0 trunk0
$ openstack network trunk set --subport port=port1,segmentation-type=vlan,segmentation-id=101 trunk0
$ openstack network trunk set --subport port=port2,segmentation-type=vlan,segmentation-id=102 trunk0

$ openstack server create --image cirros-0.5.2-x86_64-disk --flavor cirros256 --nic port-id=port0 --availability-zone :devstack0 vm0 --wait
$ sudo ovs-vsctl list-br | egrep tbr
tbr-801806cc-3 # on devstack0

$ openstack server show vm0 -f value -c OS-EXT-SRV-ATTR:hypervisor_hostname
devstack0
$ openstack server migrate vm0 --wait
$ openstack server resize confirm vm0
$ openstack server show vm0 -f value -c OS-EXT-SRV-ATTR:hypervisor_hostname
devstack0a # migration was successful

$ sudo ovs-vsctl list-br | egrep tbr
[no tbr left behind]

# cleanup
$ openstack server delete vm0 --wait
$ openstack network trunk delete trunk0
$ openstack port delete port2 port1 port0
$ openstack network list -f value -c ID -c Name | awk '/ net[012]$/ { print $1 }' | xargs -r openstack network delete

Also don't we delete trunk bridges on the neutron side instead of os-vif?

https://opendev.org/openstack/neutron/src/commit/2acb96c37468120ba79de7d8de48ffb942e78a2b/neutron/services/trunk/drivers/openvswitch/agent/ovsdb_handler.py#L236

[Slawek Kaplonski (slaweq)]

@Bence - I think that I was doing live migration but I will verify that and will update this LP.

[Slawek Kaplonski (slaweq)]

@Bence: I just confirmed it. After live migration trunk bridge isn't removed from the src node.
When vm is deleted, bridge is deleted properly.
I'm not sure if it's os-vif or neutron issue but there is for sure something wrong there.

[Slawek Kaplonski (slaweq)]

Now I can't reproduce it and it seems that tbr bridge is deleted after short deplay. So I'm going to close this bug.