Neutron does not work properly L3 HA VRRP

[Braian]

I'm using Openstack Kilo in Redhat 7.
Version of Neutron package in RDO: openstack-neutron-2015.1.1-1.el7.noarch

I have two nodes of Neutron in which I running neutron-l3-agent. I've configured HA with VRRP

# neutron l3-agent-list-hosting-router 6a69d00a-3e42-46da-b38a-1bbedc873c92
+--------------------------------------+--------------------------+----------------+-------+----------+
| id | host | admin_state_up | alive | ha_state |
+--------------------------------------+--------------------------+----------------+-------+----------+
| f16f472a-93b1-44bf-9eda-ecf40866795f | prod-neutron01 | True | :-) | standby |
| bb35d988-7009-497c-a791-92e057ae7679 | prod-neutron02 | True | :-) | active |
+--------------------------------------+--------------------------+----------------+-------+----------+

When I reboot the active node, in this case neutron02 (shutdown -r now). I still watching this:

# neutron l3-agent-list-hosting-router 6a69d00a-3e42-46da-b38a-1bbedc873c92
+--------------------------------------+--------------------------+----------------+-------+----------+
| id | host | admin_state_up | alive | ha_state |
+--------------------------------------+--------------------------+----------------+-------+----------+
| f16f472a-93b1-44bf-9eda-ecf40866795f | prod-neutron01 | True | :-) | active |
| bb35d988-7009-497c-a791-92e057ae7679 | prod-neutron02 | True | xxx | active |
+--------------------------------------+--------------------------+----------------+-------+----------+

In this moment, the HA works properly and I still have connections in my virtual machines.

The problem appears two minutes later, when the neutron02 is up again

# neutron l3-agent-list-hosting-router 6a69d00a-3e42-46da-b38a-1bbedc873c92
+--------------------------------------+--------------------------+----------------+-------+----------+
| id | host | admin_state_up | alive | ha_state |
+--------------------------------------+--------------------------+----------------+-------+----------+
| f16f472a-93b1-44bf-9eda-ecf40866795f | prod-neutron01 | True | :-) | active |
| bb35d988-7009-497c-a791-92e057ae7679 | prod-neutron02 | True | :-) | standby |
+--------------------------------------+--------------------------+----------------+-------+----------+

And then

 neutron l3-agent-list-hosting-router 6a69d00a-3e42-46da-b38a-1bbedc873c92
+--------------------------------------+--------------------------+----------------+-------+----------+
| id | host | admin_state_up | alive | ha_state |
+--------------------------------------+--------------------------+----------------+-------+----------+
| f16f472a-93b1-44bf-9eda-ecf40866795f | prod-neutron01 | True | :-) | standby |
| bb35d988-7009-497c-a791-92e057ae7679 | prod-neutron02 | True | :-) | active |
+--------------------------------------+--------------------------+----------------+-------+----------+
[root@prod-oscontroller01 ~(keystone_admin)]#

The neutron02 become itself in Master again, so I lost connection with my virtual machines.

If I see the state of FlotingIP's port in the router:

# neutron port-show 54fef3f2-3d13-4575-9bb2-4491ae4896d7
+-----------------------+-------------------------------------------------------------------------------------+
| Field | Value |
+-----------------------+-------------------------------------------------------------------------------------+
| admin_state_up | True |
| allowed_address_pairs | |
| binding:host_id | prod-neutron02 |
| binding:profile | {} |
| binding:vif_details | {} |
| binding:vif_type | binding_failed |
| binding:vnic_type | normal |
| device_id | 6a69d00a-3e42-46da-b38a-1bbedc873c92 |
| device_owner | network:router_gateway |
| extra_dhcp_opts | |
| fixed_ips | {"subnet_id": "415b44b8-a2a5-4a43-a87d-6aee9b558063", "ip_address": "10.100.44.87"} |
| id | 54fef3f2-3d13-4575-9bb2-4491ae4896d7 |
| mac_address | fa:16:3e:df:e5:45 |
| name | |
| network_id | 5bb1eb0f-19ad-4a62-bf6e-19ac6dacff15 |
| security_groups | |
| status | ACTIVE |
| tenant_id | |
+-----------------------+-------------------------------------------------------------------------------------+
[root@prod-oscontroller01 ~(keystone_admin)]#

I see 'binding_failed'

To solve this problem, I've restarted the services 'systemctl restart neutron-l3-agent neuton-openvswitch-agent' in both neutrons. I've found this workaround in this ticket https://bugs.launchpad.net/neutron/+bug/1488619

In my nodes selinux and Iptables are disabled.

I don't know what is the problem?. Greetings!

[Ann Taraday (akamyshnikova)]

I think the issue that described here is similar to https://bugs.launchpad.net/neutron/+bug/1597461