On Fri, May 01, 2020 at 02:40:41PM -0400, Vasu Dasari wrote: > Thanks William for your comments. > > > *Vasu Dasari* > > > On Fri, May 1, 2020 at 11:52 AM William Tu <[email protected]> wrote: > > > On Thu, Apr 30, 2020 at 08:42:04AM -0400, Vasu Dasari wrote: > > > Hi, > > > > > > I am trying to implement a functionality, where in if user specifies port > > > through which a VxLAN encapsulated packet can be sent out, then use that > > > port rather than going through routing procedure. > > > > > > ovs-vsctl add-port br0 at_vxlan_fp1 -- \ > > > set int at_vxlan_fp1 type=vxlan \ > > > options:remote_ip=172.32.2.1 options:local_ip=172.32.2.100 \ > > > options:dst_mac=00:00:00:00:01:02 > > options:src_mac=00:00:00:00:01:01 > > > \ > > > options:out_port=1 > > > > > Why do you need to add dst_mac and src_mac? > > Usually in the OVS kernel datapath case, OVS will consult the Linux > > kernel's > > arp table and get the src/dst mac address. > > What if the manually set dst_mac here is different than the entry in > > kernel? > > > > > The goal is not to use Linux kernel's capabilities to determine which path > the tunnel to take. > > Imagine a set of switches in a fabric controlled by a controller and it is > used primarily for L2 services and some L3 services. If I were to use Linux > kernel to dictate which path to take, it would lead to using one bond or a > physical interface all the time for a particular destination switch. By > doing so, there is no fabric diversity. Hence, as controller has view of > the network, it can dictate which path/topology to take for a particular L2 > or L3 service. By using virtualized the fabric, I do not have to worry > about about Mac pollution and loops (I can detail this separately if > needed). And hence wanted to use VxLAN. > I see your use case, thanks! William
_______________________________________________ dev mailing list [email protected] https://mail.openvswitch.org/mailman/listinfo/ovs-dev
