[virt-tools-list] Bridged Network, Fedora 16 and Network Manager

[Laine Stump]

On 01/19/2012 11:58 PM, Eric Mesa wrote:
> I'm using:
>  virt-manager 0.9.0
> Fedora 16
> KDE 4.7.4
> network manager 0.9.2-1
>
> I want to have bridged connections so my VMs can be on the same 
> network segment as the host.  I used the instructions on here: 
> http://www.linux-kvm.com/content/bridged-networking-virt-manager-083 
>  When I did that, it made my ethernet connection go away.  It seemed 
> to freeze up, so I stopped the process.  After that I had no ethernet 
> connection so I had to go into network manager and add it back in. 
>  When I tried it again, I once again lost my ethernet connection.


Once you put an ethernet into a bridge, NetworkManager will never see it 
again. NM in F16 is smart enough to *ignore* bridge devices, and 
ethernets attached to bridges, but that's as far as it goes.


(Of course, the reason that you were no longer able to communicate out 
the ethernet after the change could be because of the absence of DEVICE 
in your ifcfg-br0 file. I'm not sure what the bridge device will be 
called if you don't specify it, but if it doesn't match what's in the 
BRIDGE= line of ifcfg-eth0, definitely your networking will be notworking.)

> So I looked around on Linuxquestions.org and found instructions to do 
> the following:
>
> create /etc/sysconfig/network-scripts/ifcfg-br0 by copying it from 
> /etc/sysconfig/network-scripts/ifcfg-eth0 then edit both files to 
> match what you need, something like I have listed below
>
>
> /etc/sysconfig/network-scripts/ifcfg-br0
> Code:
> DEVICE="br0"
> ONBOOT=yes
> TYPE=Bridge
> BOOTPROTO=static
> IPADDR=192.168.10.40
> GATEWAY=192.168.10.1
> DELAY=0

You might want to add "STP=on" to that (otherwise DELAY=0 has no 
meaning). Unless your ethernet is connected to a managed switch that 
forbids STP packets coming from your port (some admins put a rule on the 
ports that shuts them down if they detect STP packets. grumble grumble).

>
> /etc/sysconfig/network-scripts/ifcfg-eth0
> Code:
> DEVICE="eth0"
> ONBOOT=yes
> TYPE=Ethernet
> HWADDR=00:12:13:14:AB:CD
> BRIDGE=br0
> Then do 'service network restart'
>
> On my computer instead of DEVICE it has NAME.  So I set it up like this:

NAME and DEVICE are not the same thing. NAME is just a descriptive bit 
of text, while DEVICE is the actual name of the device in the kernel.

>
> cat ifcfg-br0
> TYPE=Bridge
> BOOTPROTO=none
> IPADDR0=192.168.1.17
> PREFIX0=24
> GATEWAY0=192.168.1.1
> DNS1=192.168.1.7
> DEFROUTE=yes
> IPV4_FAILURE_FATAL=yes
> IPV6INIT=no
> NAME=br0
> ONBOOT=yes
>
>
> When I put the BRIDGE=br0 into my eth0, it stopped working correctly. 
>  What do I have to do to get this set up correctly?
>
>
> One other question - would I need to set up a brX for each VM?

No. As a matter of fact you *can't* do that. For the type of networking 
you're using (called "bridged mode" in most documentation), the bridge 
device needs to have a physical device directly attached to it, and each 
physical device can only be attached to a single bridge, so you would 
need an extra physical device for each bridge. Anyway, a bridge device 
can have 1024 devices attached to it; counting the one physical device, 
you could attach 1023 guest interfaces to a single bridge (not that 
you'd get stellar performance, but that's the limit).