On Tue, 2012-01-03 at 18:35 +0000, Daniel P. Berrange wrote: > On Tue, Jan 03, 2012 at 11:29:40AM -0700, Steven Dake wrote: > > On 01/03/2012 05:32 AM, Mark McLoughlin wrote: > > > On Fri, 2011-12-23 at 13:16 -0700, Steven Dake wrote: > > >> Killing the first VM that was started (which has a MAC address equal to > > >> the br0) will cause deletion of vnet0 and cause the bridge driver to > > >> flush the arp table and choose a new bridge interface. This results in > > >> disconnection of the remaining VM for 30-60 seconds. > > > > > > What exactly is happening during this time? The guest's routing table is > > > configured with the host as its gateway and it has a stale ARP cache > > > entry for the host? > > > > > > Cheers, > > > Mark. > > > > > > > br0 adopts the mac address of the first interface added to it. If the > > first interface added to it is removed, br0 finds a new mac address from > > the "next" interface in the bridge. When the mac address is removed, > > the arp table is flushed for 30-60 seconds. You mean the host kernel arp table is empty for 30-60 seconds? or ...? A stale arp cache in the guest, I can understand. Anything else smells like a bug. > Aside from the connectivity blackout, this mac address also causes > pain for recent Windows guests, which will think they have been moved > to a new physical network everytime the MAC addr changes, causing them > to popup annoying dialog boxes on every login Yep, I understood that from the libvirt commit message - makes perfect sense. I'm just curious what exactly is going on in this example. Cheers, Mark. _______________________________________________ cloud mailing list cloud@xxxxxxxxxxxxxxxxxxxxxxx https://admin.fedoraproject.org/mailman/listinfo/cloud
