Gabe, I suggest reviewing the output of brctl showmacs brX after attempting communications. For network B are all of respective guests MAC addresses there?
On Jan 29, 2012 2:26 AM, "Gabriel Gunderson" <[email protected]> wrote: > > On Sat, Jan 28, 2012 at 11:03 PM, Tod Hansmann <[email protected]> wrote: > > I'm guessing this has to do with VLAN tagging, and I'm not entirely sure > > how all that works in virtual hosts let alone kvm/libvirt which I've > > never used thus far. The VLAN tagging could be working on Set X > > specifically because it has the 3rd interface. I don't know what that > > 3rd interface translates into on the wire, and on which of the 3 host's > > wires, but it's probably more telling than you're giving it credit for. > > I don't know... when pinging from a virt guest that works to another > that doesn't, the packets shouldn't even hit the switch. Everything is > untagged until then. More ascii flowcharting follows: > > virt-guest-x NIC => virt-host NIC => bridge-x => virt-host NIC virt-guest-x NIC > > So, no CAT5, VLAN or switch in that case. Just the bridge and virtual > interfaces. > > > > Sorry I don't have more specific ideas on this. I'm kind of hoping a > > sounding board effect will be beneficial for you here, as your situation > > is very specific. Call me if you want to keep bouncing ideas back and > > forth. > > You're kind to offer; if it gets bad, I just might :) > > > Best, > Gabe > > /* > PLUG: http://plug.org, #utah on irc.freenode.net > Unsubscribe: http://plug.org/mailman/options/plug > Don't fear the penguin. > */ /* PLUG: http://plug.org, #utah on irc.freenode.net Unsubscribe: http://plug.org/mailman/options/plug Don't fear the penguin. */
