Ask Your Question

"connection to neutron failed" on network node (VirtualBox VM)

asked 2014-12-15 04:52:59 -0500

122014 gravatar image

updated 2014-12-16 09:50:49 -0500

neutron agent-list works on the controller node but fails on the network node.


Despite following the guide on ( there was no connection between the Network and Controller nodes while using "neutron agent-list" on the network node. Error was simply "connection to neutron failed".

The nodes were run on VirtualBox. The internal network 10.0.0.xx and tunneling 10.0.1.xx were host-only networks and the external network 10.0.3.xx was on NAT for internet access.

The solution to this was to change NAT to Bridged Network in VirtualBox.


Management Network ( is a host-only network.

VM Traffic Network ( is a host-only network.

Public Network ( is a bridged network. (IP begins with 192.168 but may end differently depending on your setup. The automatically generated IPs and default gateway were manually entered into /etc/network/interfaces/.)

Solved problem:

I am on the Network Node=>Check Neutron agents step of this guide: (

ConnectionFailed: Connection to neutron failed:
DEBUG: clean_up ListAgent
DEBUG: Got an error: Connection to neutron failed:

Nodes are Ubuntu 14.04 VMs on VirtualBox. Eth0 is a host-only network on 10.0.0.xx. Eth1 is a NAT network for internet access with fixed IPs and port forwarding, 10.0.3.xx. Eth2 is tunneling for the network node on 10.0.1.xx. Neutron server and RabbitMQ are configured and running.

Network node can ping the controller at (host-only) but not at (NAT). is the publicurl of neutron. The same happens with telnet. can be pinged by controller.

Pings from network to controller, internet and compute node through and work. Controller can ping network at and but not (br-ex; network's internet connection).

tcpdump of ping to (the br-ex IP of network node) from controller node
ARP, Request who-has tell, length 28
"Destination Host Unreachable. 100% packet loss."

After entering neutron agent-list, tcpdump on network's eth1 reports: ARP, Request who-has tell, length 28

Pinging from network node results in Destination Host Unreachable.

No bridges show up in brctl .

Is this a problem with the NAT setup on 10.0.3.xx or bridging (br-ex is on eth1 on

[Included logs were removed due to being unnecessary for solution]

edit retag flag offensive close merge delete


thanks for sharing the solution... it would have been better to add it as an 'answer' instead of removing the question, as you did. Check how to use this site on and if you have time, consider restoring the question and put the answer below.

smaffulli gravatar imagesmaffulli ( 2014-12-16 12:02:34 -0500 )edit

The question is below the blockquote in italics, I simply removed the 5 or so logs because they were irrelevant, as the problem wasn't related to /etc/network/interfaces, OpenStack or OpenVSwitch of the nodes, rather the VirtualBox setup. If it would be helpful I could reinclude the logs.

122014 gravatar image122014 ( 2014-12-17 05:12:26 -0500 )edit

1 answer

Sort by ยป oldest newest most voted

answered 2014-12-16 03:53:16 -0500

122014 gravatar image

updated 2014-12-16 09:52:29 -0500

Update: the neutron/network node connectivity problem was fixed by changing the external network setup in VirtualBox from NAT to Bridged Interface, and updating the IPs in /etc/network/interfaces, config files and in the keystone database accordingly.

edit flag offensive delete link more

Get to know Ask OpenStack

Resources for moderators

Question Tools

1 follower


Asked: 2014-12-15 04:52:59 -0500

Seen: 482 times

Last updated: Dec 16 '14