Open Stack

Hi George,

there a lot of deployment options for OpenStack and Networking services.
First i need to clarify that OpenStack can be run in many different 
ways. That is running OpenStack services in different machines according 
to someone's needs.
This means that controller and compute nodes can have several setups 
depending each deployment case.
In your case a controller and a compute node are described in the manual 
as in the following link
http://docs.openstack.org/trunk/install-guide/install/yum/content/ch_overview.html#overview-architecture
and nova-network is used in a multi-host setup.
Multi-host setup for the networking means that every computing host will 
deploy a nova-network service, avoiding SPOF in comparison with a single 
node setup.
Further, take a note that in a single node setup you could collocate 
nova-network in the "controller"node.
All these trying to explain you some of the many different setups.

Now, for your case br100 should be on compute node only. This can be 
done either by running the command in controller or in compute, usually 
controller is used because nova-clients live there.

Hope i didn't grow your confusion :)
Thanassis

Thanassis Parathyras
StackMasters - The European OpenStack Integration Company
www.stackmasters.eu

On 13/1/2014 12:54 πμ, Georgios Dimitrakakis wrote:
>
> Thank you all for your suggestions!
>
> I was (and still am) confused since the manual says that the 
> nova-network create command should be run on the controller node. That 
> is obviously failing because no br100 is defined. So do I have to put 
> br100 on the controller as well or should I just create the network on 
> the compute node?
> What are the differences in these setups?
>
> Best,
>
> G.
>
> On Thu, 09 Jan 2014 10:05:54 -0600, Dimitri Maziuk wrote:
>> On 1/9/2014 4:11 AM, Georgios Dimitrakakis wrote:
>>> Hello again!
>>>
>>> No the br100 was not created automatically unfortunately! There is also
>>> this bug report: 
>>> https://bugs.launchpad.net/openstack-manuals/+bug/1241331
>>
>> I'm not 100% sure but I think in my case it did get created
>> automatically when I tried to fire up the cirros instance the 2nd time
>> around... However, I may have created the bridge manually at some
>> point while fighting with it, I forget.
>>
>>
>> https://access.redhat.com/site/documentation/en-US/Red_Hat_Enterprise_Linux/5/html/Virtualization/sect-Virtualization-Network_Configuration-Bridged_networking_with_libvirt.html 
>>
>>
>> (from "disable network manager" down to "restart networking or reboot")
>>
>> Also check that vhost_net module is loaded (lsmod).
>>
>>> Since it wasn't created automatically I am asking if it has to be on
>>> both nodes (controller + compute) in order for the network to work
>>> correctly? Furthermore, I would like to know how should it be 
>>> bridged in
>>> order to achieve floating (public) IPs?
>>
>> Compute only: this is the interface VMs use to talk to the world.
>> Floating IPs are a separate story really -- but if I uderstand the
>> question correctly, your br100 should be bridged to the subnet where
>> your floating IPs are.
>>
>> I.e. if your eth0 is on 1.2.3.0/24 and your eth1 is on 10.0.0.0/24,
>> and you want floating IPs in 1.2.3.0/24 range, then you want to bridge
>> br100 to eth0.
>>
>> Dima
>>
>>
>> _______________________________________________
>> Mailing list: 
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>> Post to : openstack at lists.openstack.org
>> Unsubscribe : 
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack
>