Saw this and thought I'd try to correct a possible terminology confusion involving terms bridging and bonding as related to teaming. Rodney Baker wrote:
Bridged interfaces (also known as teamed interfaces) like br0 are generally used to provide redundant connections via multiple physical interfaces - the IP setttings are configured on the "bridge" interface rather than the component interfaces.
--- Bridging != teaming. Bridging makes what would normally be 2 (or more) physically separate *subnets* into 1 subnet. Teaming takes 2 separate interfaces to the *same* subnet and uses them as 1 interface -- allowing one to achieve one of multiple goals: 1) redundancy based on fail-over; 2) load-balancing, 3) and aggregation. Perhaps there is confusion with the term "bonding". Bonding is, essentially, the same as teaming. In linux, bonding is implemented in the kernel. Teaming is implemented as a driver but is controlled with a user-space daemon. I have only used bridging (teaming wasn't available when I used bridging) and used it for aggregation (using 2 links in a 'RAID0' type manner) to achieve up to 2X the speed of a single link (depending on traffic). I don't know if the original poster got the answer they wanted, but you usually DO want your bridge to have an address. If you have 2 interfaces to two separate subnets (presumably both already have addresses), and you want them to act like 1, then the interfaces end up being declared as HW interfaces on the bridge and will not have distinct IP addresses. Since they are merged into 1 virtual interface (the bridge), it is the virtual interface that will end up having an address. For optimal transparency, you'd likely take the ip addr off 1 of the network interfaces and give it to the bridge. Example: assume eth0 and eth5 are on server 'S'. They connect to different physical networks -- eth0 connects to a 1Gbps net @ 192.168.3.0 (w/host addr of 192.168.3.1) and eth5 connects to net 192.168.4.0 (w/host addr of 19.2.168.4.1). Only one other host is on the 192.168.4.0 net (at 192.168.4.12). So will get rid of the ".4.0" net and move ".4.12" -> ".3.12". To combine the nets, I create a "bridge device", "br0", on the server. eth0 and eth5 become interfaces of the bridge. # brctl show bridge name bridge id STP enabled interfaces br0 8000.001517bfbeb2 no eth0 eth5 The bridge becomes the server's new interface to net 192.168.3.0 and gets the server's previous address on the 3.0 net (192.168.3.1). Since the bridge is a "network device", it usually needs an IP address (unless it is an interface or slave to some other network device). Hope that made sense. It was complicated enough for me that I wrote a script for it and added it as an option to my network start script. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org To contact the owner, e-mail: opensuse+owner@opensuse.org