[Openstack-operators] Flat network with linux bridge plugin
Sam Morrison
sorrison at gmail.com
Wed Apr 8 23:14:09 UTC 2015
I reset neutron (cleared DB etc.) and rebooted the network node and it worked fine the second time around.
I think the first time something went wrong talking to the DB initially. I fixed up the config but then it was impossible for it to fix itself. eg. restarting the neutron agents did nothing.
Sam
> On 9 Apr 2015, at 1:47 am, Daniele Venzano <Daniele.Venzano at eurecom.fr> wrote:
>
> I am 99% sure I configured the linuxbridge agent on the network node the same way as on the compute nodes, but it was doing nothing.
> But I did it a while ago, so I could be wrong. Anyway having the agent constantly running just to create a bridge at boot is bit of a waste.
> The next maintenance window I will try again, just to understand.
>
>
> From: Kris G. Lindgren [mailto:klindgren at godaddy.com]
> Sent: Wednesday 08 April 2015 17:01
> To: Daniele Venzano; 'Daniel Comnea'
> Cc: openstack-operators at lists.openstack.org
> Subject: Re: [Openstack-operators] Flat network with linux bridge plugin
>
> We run this exact configuration with the exception that we are using OVS instead of linux bridge agent. On your "Network nodes" (those running metadata/dhcp) you need to configure them exactly like you do you compute services from the standpoint of the L2 agent. Once we did that when the l2 agent starts it creates the bridges it cares about and the dhcp agent then gets plugged into those bridges. We didn't have to specifically create any bridges or manually plug vifs into it to get everything to work.
>
> I would be highly surprised if the linuxbridge agent acted any differently. Mainly because the dhcp agent consumes an IP/port on the network, no different than a vm would. So the L2 agent should plug it for you automatically.
> ____________________________________________
>
> Kris Lindgren
> Senior Linux Systems Engineer
> GoDaddy, LLC.
>
> From: Daniele Venzano <daniele.venzano at eurecom.fr <mailto:daniele.venzano at eurecom.fr>>
> Organization: EURECOM
> Date: Wednesday, April 8, 2015 at 4:21 AM
> To: 'Daniel Comnea' <comnea.dani at gmail.com <mailto:comnea.dani at gmail.com>>
> Cc: "openstack-operators at lists.openstack.org <mailto:openstack-operators at lists.openstack.org>" <openstack-operators at lists.openstack.org <mailto:openstack-operators at lists.openstack.org>>
> Subject: Re: [Openstack-operators] Flat network with linux bridge plugin
>
> Juno (from ubuntu cloud), on Ubuntu 14.04
>
> From: daniel.comnea at gmail.com <mailto:daniel.comnea at gmail.com> [mailto:daniel.comnea at gmail.com <mailto:daniel.comnea at gmail.com>] On Behalf Of Daniel Comnea
> Sent: Wednesday 08 April 2015 11:29
> To: Daniele Venzano
> Cc: Sam Morrison; openstack-operators at lists.openstack.org <mailto:openstack-operators at lists.openstack.org>
> Subject: Re: [Openstack-operators] Flat network with linux bridge plugin
>
> Which release are you using it, on which OS ?
>
>
> On Wed, Apr 8, 2015 at 9:00 AM, Daniele Venzano <daniele.venzano at eurecom.fr <mailto:daniele.venzano at eurecom.fr>> wrote:
> Well, I found a way to make it work.
> Yes, you need a bridge (brctl addbr ...).
> You need to create it by hand and add the interfaces (physical and dnsmasq namespace) to it.
> The linuxbridge agent installed on the network node does not do anything.
>
> The problem with this is that the interface for the namespace is created after an arbitrary amount of time by one of the neutron daemons, so you cannot simply put the bridge creation in one of the boot scripts, but you have to wait for the interface to appear.
>
>
> From: Sam Morrison [mailto:sorrison at gmail.com <mailto:sorrison at gmail.com>]
> Sent: Wednesday 08 April 2015 05:46
> To: Daniele Venzano
> Cc: openstack-operators at lists.openstack.org <mailto:openstack-operators at lists.openstack.org>
> Subject: Re: [Openstack-operators] Flat network with linux bridge plugin
>
> Hi Daniele,
>
> I’ve started playing with neutron too and have the exact same issue. Did you find a solution?
>
> Cheers,
> Sam
>
>
>
>> On 18 Feb 2015, at 8:47 pm, Daniele Venzano <Daniele.Venzano at eurecom.fr <mailto:Daniele.Venzano at eurecom.fr>> wrote:
>>
>> Hello,
>>
>> I’m trying to configure a very simple Neutron setup.
>>
>> On the compute nodes I want a linux bridge connected to a physical interface on one side and the VMs on the other side. This I have, by using the linux bridge agent and a physnet1:em1 mapping in the config file.
>>
>> On the controller side I need the dhcp and metadata agents. I installed and configured them. They start, no errors in logs. I see a namespace with a ns-* interface in it for dhcp. Outside the namespace I see a tap* interface without IP address, not connected to anything.
>> I installed the linux bridge agent also on the controller node, hoping it would create the bridge between the physnet interface and the dhcp namespace tap interface, but it just sits there and does nothing.
>>
>> So: I have VMs sending DHCP requests. I see the requests on the controller node, but the dhcp namespace is not connected to anything.
>> I can provide logs and config files, but probably I just need a hint in the right direction.
>>
>> On the network controller:
>> Do I need a bridge to connect the namespace to the physical interface?
>> Should this bridge be created by me by hand, or by the linuxbridge agent? Should I run the linuxbridge agent on the network controller?
>>
>> I do not want/have a l3 agent. I want to have just one shared network for all tenants, very simple.
>>
>> Thanks,
>> Daniele
>> _______________________________________________
>> OpenStack-operators mailing list
>> OpenStack-operators at lists.openstack.org <mailto:OpenStack-operators at lists.openstack.org>
>> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators <http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators>
>
>
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org <mailto:OpenStack-operators at lists.openstack.org>
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators <http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators>
>
> _______________________________________________
> OpenStack-operators mailing list
> OpenStack-operators at lists.openstack.org
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-operators
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-operators/attachments/20150409/854b826e/attachment.html>
More information about the OpenStack-operators
mailing list