Comment 49 for bug 1815989

@sean-k-mooney

First of all, thank you for the work you've put into this issue and for clarifying the actual cause/effect here.

I don't know that I can speak for everyone using the hybrid driver, but in our testing, we kept an eye on the OVS bridge during live migrations and found that migrations in which the Neutron ports attached to br-int had VLAN tagging applied in < 4 seconds were successful 100% of the time (no drops). If the vlan tag application took > 5 seconds, and especially > 7 seconds, the migration would result in ~5-40 seconds of connectivity loss to the VM.

Although the comment references ovs firewall, driver, https://bugs.launchpad.net/neutron/+bug/1414559/comments/49 calls out the same behavior with:

"When the VM is migrated to the destination, and the VM send the rarp packets, but because it's too fast, the openflow and tag haven't been configed in br-int, then the rarp packets will be drop."

In our case, it would seem the VM is already sending RARPs prior to the VLAN tag being set on the port.

Could it be 'live_migration_wait_for_vif_plug=true' is really what we're after, here?