On Mon, Oct 18, 2010 at 2:49 PM, Ben Pfaff <b...@nicira.com> wrote: > On Mon, Oct 18, 2010 at 08:44:14AM -0200, Henrique Rodrigues wrote: > > Hi Justin > > > > On Sun, Oct 17, 2010 at 10:36 PM, Justin Pettit <jpet...@nicira.com> > wrote: > > > > > On Oct 17, 2010, at 5:28 PM, Derek Cormier wrote: > > > > > > > When you add ports to a virtual bridge, should you still be able to > ping > > > them? For some reason, I cannot ping the interface on the switch that a > host > > > connects to. Yet, I can ping another host. > > > > > > Do you mean, for example, if you add eth0 to br0, you can't ping the > > > address assigned to eth0? If so, then, no, you cannot ping the > addresses of > > > attached devices. > > > > > > Why not? I also noticed that... > > > > As far as I know, after receiving a packet, the datapath will send a > message > > to the ovs-vswitchd stating that there is no rule matching that packet. > Upon > > receiving this "missed flow message", ovs-vswitchd can check if the paket > > ip_dst matches one of its local interfaces' addresses and is able to > answer > > the question telling the datapath to send the packet to the running > kernel > > IP stack. Is there any rule prohibiting this behavior? > > It doesn't work because it's not the way Linux bridging works, it > doesn't fit well into the Linux network stack, and we didn't see a > good-enough reason to try to fight both of those precedents to implement > it anyway. >
Sorry, I just asked to check if there were a way to do that... -- Henrique Rodrigues
_______________________________________________ discuss mailing list discuss@openvswitch.org http://openvswitch.org/mailman/listinfo/discuss_openvswitch.org
