Ask Your Question

Network and Controller on same node [closed]

asked 2014-11-12 13:00:27 -0500

Campos gravatar image

Hey y'all. I've been stuck on a problem here for a while, and I finally decided to come here and ask for help. Here's my situation: On this lab I work, there are 2 nodes.

Machine1: Two network interfaces, p10p1 and p4p1 (I don't know why it isn't eth0 and eth1, when I arrived it was already like this). p10p1 is connected to the wall, and gets the internet. p4p1 is connected to a router, where my other node is linked (in a private lan).

Machine2: One network interface, p10p1.

On Machine 1, I'm supposed to install the Controller and Network Nodes, and on Machine 2, the Compute Node. I'm following this installation guide: (

There, it says the Network Node has to have 3 network interfaces, but that's impossible in the current infracstructure of the lab, so I have to work around that. Here lies my question: I only have 2 network interfaces, so which one of them is going to be the External Interface? According to the guide, this interface has to be configured in a special way that has no ip attached, and I never really understood that. Appreciate all the help!

edit retag flag offensive reopen merge delete

Closed for the following reason question is not relevant or outdated by Campos
close date 2016-09-20 12:44:58.626821

2 answers

Sort by ยป oldest newest most voted

answered 2014-11-12 14:36:31 -0500

dbaxps gravatar image

updated 2014-11-12 15:01:07 -0500

This is a working Fedora 20 Two Node Cluster && ready and tested solution ( one more p4p1 has to be plugged in Compute Node - not expensive procedure). View RDO Setup Two Real Node (Controller+Compute) IceHouse Neutron ML2&OVS&GRE Cluster on F20
External and Management Network are physically the same. Another similar architecture ( requires CentOS 7 from scratch setup on both boxes )
RDO Setup Two Real Node (Controller+Compute) IceHouse Neutron ML2&OVS&VXLAN Cluster on CentOS 7. Minimal experience needed to convert answer file from CentOS 7 Cluster to Fedora 20 Cluster ( Two Node), but again p4p1 (Realtek 8169) has to be plugged in Compute Node as endpoint for VXLAN tunnel ( in previous design GRE tunnel)
It's RDO Multi node deployment technology matches your request. Answer files are provided ready to go in documentation
The most recent similar solution for RDO Juno RDO Juno Set up Two Real Node (Controller+Compute) Gluster 3.5.2 Cluster ML2&OVS&VXLAN on CentOS 7
Once again conversion answer file from CentOS7 to Fedora 20 requires minimal experience
Names of Ethernet interfaces p10p1 && p4p1 are specific for Fedora 20.

edit flag offensive delete link more

answered 2014-11-27 06:53:38 -0500

Campos gravatar image

Thanks for the reply! Actually, I still didn't quite understand. I'm using Ubuntu 14.04. My question is: If External and Management are the physically the same, how would I configure them? On the installation guide, the External is supposed to be configured without IPs. Do I have to create a virtual interface network? And on the compute node, if really two NICs are required, where would they connect? I use a small D-link router between my Controller and Compute nodes.

edit flag offensive delete link more


Samples are for RDO deployment and answer-files provided are ready to go. Please, view links.

dbaxps gravatar imagedbaxps ( 2014-11-27 07:04:05 -0500 )edit

Get to know Ask OpenStack

Resources for moderators

Question Tools



Asked: 2014-11-12 13:00:27 -0500

Seen: 290 times

Last updated: Nov 27 '14