Bug 1234837

Description of problem:

I followed the following documentation to deploy an allinone instance of Kilo:

https://www.rdoproject.org/Neutron_with_existing_external_network

No errors were present during the deployment and it reports the status as OK.

I then started an instance of Cirros. The instance starts correctly and I see a floating IP assigned.

I then go into the console and view the output of 'ifconfig' for the cirrus instance. It is assigned a ipv6 address and no the floating address.

Under /var/log/messages of the CirrOS node I also see 'no IPv6 routers present'

I also found:

avahi-daemon[745]: Registering new address record for fe80::fc16:3eff:fe80:c58 on tapb179fe4e-b8.*.

This all happens despite having use_ipv6=false set in /etc/nova/nova.conf. I also restarted the whole node to insure it was set.

More details can be seen on ask.openstack:

https://ask.openstack.org/en/question/67724/kilo-assigning-ipv6-even-though-its-disabled/

Version-Release number of selected component (if applicable):

CentOS 7 - 3.10.0-229.4.2.el7.x86_64

openstack-packstack-2015.1-0.1.dev1537.gba5183c.el7.noarch


How reproducible:

Follow the installation guide on the RDO site and then view a cirros instance on the console. 

Steps to Reproduce:
1. Deploy RDO following the steps outlined here https://www.rdoproject.org/Neutron_with_existing_external_network

2. Insure ipv6 is disabled 'use_ipv6=false'
3. Start a cirrus instance, which should be automatically assigned a floating address (ipv4)
4. Open a vnc console to the cirrus instance and check the IP assignment using 'ifconfig'
5. Check dmesg or /var/log/messages for ipv6 messages (as above)

Actual results:

The cirros instance is assigned an ipv6 address, even though ipv6 is disabled, and no ipv6 router is present.

Expected results:

The cirros instance should be assigned an ipv4 address

Additional info:

Available on request