Re: [Xen-users] [XEN][Ubuntu's Linux bridge replaced by OVS problem using XEN]

[lists+xen@xxxxxxxxxxxxxx]

On Fri, 29 Jun 2012 10:21:40 +0500
Waqar Ahmed <waqar.ahmed@xxxxxxxxxxxxx> wrote:

> Thank you Fajar, No I have never used OVS on any configuration.
> I am following the INSTALL.LINUX and INSTALL.BRIDGE to insert the OVS
> as a kernel module in XEN booted kernel on Ubuntu 11.10. I do remove
> the Linux bridge to get OVS working.
> I am still working on it and will reply once i get something
> significant


I have been using this for quite some time.

In xen-network-common.sh you can find the add_to_bridge function. Here
you find the line 'brctl addif ${bridge} ${dev}'. Simply replace this
with 'ovs-vsctl add-port ${bridge} ${dev}'.

If you want to assign the device to a VLAN from your VM config you can
do that too. Openvswitch can create "fake bridges" that are VLAN tagged
and you can add interfaces to that fake bridge.

Example:

# Create virtual bridge
ovs-vsctl add-br vbr0

# Create a fake bridge connected to vbr0 with tag 10
ovs-vsctl add-br vbr0 vlan10 10

# Add a device that gets tagged with tag 10
ovs-vsctl add-port vlan10 vm1


So, in your VM's vif line this translates to:
script=vif-bridge,vifname=vm_foo0,mac=02:16:3e:0a:00:01,bridge=vlan10,model=e1000

As an aside, I always advice people to not let Xen interfere with their
networking setup - if you use xend this means to have the second line
of network-bridge be 'exit 0'. That way you can just setup your own
networking without the odd routine of network-bridge.

Basically the Xen scripts are scripts you can customise to your needs
in whatever way you want. Many people seem to be confused about this
but in the end it is all very easy to customise.

That being said, one feature that is currently missing is that I would
really like to be able to pass arguments to the scripts, for example
'vtag=10' or 'vtag=trunk'. It seems impossible, or at least I could not
find ways to do that without hacking other bits and pieces of Xen which
are not as easy to maintain when you upgrade Xen.

HTH,
Mark

_______________________________________________
Xen-users mailing list
Xen-users@xxxxxxxxxxxxx
http://lists.xen.org/xen-users