Jim Wildman wrote -----------------------------
YOu will have to do it on the virbr side, if it does not pick it up automatically. If the bond is set to be your default route, it may just "do the right thing". More surprising things have happened...
On Mon, 24 Oct 2011, Bob Hoffman wrote:
/ Interested in bonding my 2 on board nics along with my add on card NIC
/>/ for a total of 3.// />/ How would this work with the virtual machines? />/ My eths/IPs ---> bond0, bond0:1, etc ---> ? ----> virbr0,virbr0:1 />/ (each machine to have own IP address, one machine may have some virtual />/ sites needing more than one ip, />/ so multiple ips added to mix..)// />/add something to the bond0 file, or just leave it alone and mess with />/ the virbr0 files? />/ I heard something about network manager not liking bonding and bridging... />/ anyone do this kind of thing? /
-----------------------------
Still working on a solution. Apparently the bond<n> files demand an ipaddress, thus there might have to be one for each and every single ip coming into the computer...I guess you would have to do that anyway just like eth0, eth0:0, eth0:1, etc. I think I am going to try to just make a separate eth<n> for each ip, going to their respective bond<n> with the proper ipaddress in them. Then use the bridge as normal, with each bond<n> calling a respective bridge. Not sure how that works with multiple ips going to same machine (as in, can the bridge handle more than one ip, or can the machine look for more than one bridge...?)
Then obviously, the VM would need to go through the whole process too. I can see no other way and there is no information out there going over any of it...will post if it works. (next step after figuring out lvm kvm storage issues.....wheee)