This is an archived copy of the Xen.org mailing list, which we have preserved to ensure that existing links to archives are not broken. The live archive, which contains the latest emails, can be found at http://lists.xen.org/
Home Products Support Community News


Re: [Xen-users] XEN and NICS

To: Hans Pfeil <HPfeil@xxxxxxx>
Subject: Re: [Xen-users] XEN and NICS
From: Mike Lovell <mike@xxxxxxxxxxxx>
Date: Tue, 17 Feb 2009 19:06:43 -0700
Cc: xen-users@xxxxxxxxxxxxxxxxxxx
Delivery-date: Tue, 17 Feb 2009 18:07:40 -0800
Envelope-to: www-data@xxxxxxxxxxxxxxxxxxx
In-reply-to: <499ADD8D.6010.00BD.0@xxxxxxx>
List-help: <mailto:xen-users-request@lists.xensource.com?subject=help>
List-id: Xen user discussion <xen-users.lists.xensource.com>
List-post: <mailto:xen-users@lists.xensource.com>
List-subscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=subscribe>
List-unsubscribe: <http://lists.xensource.com/mailman/listinfo/xen-users>, <mailto:xen-users-request@lists.xensource.com?subject=unsubscribe>
References: <499ADD8D.6010.00BD.0@xxxxxxx>
Sender: xen-users-bounces@xxxxxxxxxxxxxxxxxxx
User-agent: Thunderbird (Windows/20081209)
Hans Pfeil wrote:
Hey all, running sles10sp2 with Xen running from Novell.  My goal is to get 
each virtual server its own nic.  Here is what I have:

-   eth0 (onboard broadcom) is the nic dom0 is using, this has an IP address
-   eth1 (onboard broadcom) does not have an IP address, nothing connected
-   eth2-9 (two quad-port intel nics) these do not have an IP address.
-   All nics will be on the same subnet.

So, with the above shouldn't my bridge script be as follows:

dir=$(dirname $0)
"${dir}/network-bridge" "$@" vifnum=2 netdev=eth2 bridge=eth2
"${dir}/network-bridge" "$@" vifnum=3 netdev=eth3 bridge=eth3

and so on.....

Wouldn't adding eth0 to the script mess things up?  Shouldn't I leave eth0 and 
eth1 out of the script?
If you aren't going to use eth0 for guest vms, then I would leave it out of your wrapper script. It will just turn into one more thing to debug when there are problems. I would say the same for eth1. Will putting them in mess things up? It kind of depends on the rest of the configuration. But the way you are showing seems like it should work.

Here is another approach to this though. If each nic is going to be dedicated to individual virtual machines, you could do a pci passthru on them and give each nic to the associated virtual machine. This can take a bit of voodoo to get working but as long are you are using a paravirtualized guest and the nics are presented to the host as individual pci devices, it should be possible to get working. Section 5.3 of the Xen 3.0 users manual might help. Basically, you hide the pci device from the host and a pci option to the configuration for the device to pass thru. I don't remember all of the particulars since it has been a while. But that might be able to get you started on it.

Finally, why have individual nics for guests if all of them are going to be on the same subnet? I am mainly curious here. For me, the existing network setups and documentation work well. So I would like to see why you are doing this.

Hopefully I have been of some help instead of just confusing you.


Xen-users mailing list

<Prev in Thread] Current Thread [Next in Thread>