On Sat, Mar 15, 2014 at 8:31 AM, Stuart Henderson <st...@openbsd.org> wrote: >> > >Description: >> > Configuring trunk with LACP is broken. Cannot issue ICMP from >> > one trunk to another. This is done is a sandbox network. using to qemu >> > hosted openBSD. Strange FACT the id of the trunk stay 0000 Each >> > interface are direclty bridged to each other on the host, the bridge >> > is not shared, theres is one bridge per interface; >> > vio0(bsd1) <---> tapXXX <---> br0 <---> tapYYY <---> vio0(bsd2) >> > vio1(bsd1) <---> tapXXX <---> br0 <---> tapYYY <---> vio1(bsd2) > > Oh! I've just thought of something else. You have no switch sending the > LACP frames out! So of course it won't work. > > Maybe you could look at openvswitch or something if you need to model > a LACP trunk setup in a VM. (If you don't need to do that, then I don't > see all that much point doing this in the first place). >
Ok what i get is the bridge itself who announce the working link ? But if the switch is sending LACP what is this outgoing LACP packet (the one not forwaded by tun nor by a bridge) ? So lets say two machines got a bunch of nics and want to have a direct connection between them (no switch ) To get Failover(cable is long and may be unplugged) + Load balancing (need the bandwith) over the links, trunk0(LACP) is not an option . What else ? I guess VLT Force10 answered that a few years ago. -- --------------------------------------------------------------------------------------------------------------------- () ascii ribbon campaign - against html e-mail /\
