Comment 60 for bug 1794991

We see the Problem in our infrastructure like zhhuabi after updating to following versions:
Compute Nodes:
  neutron-common/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed]
  neutron-l3-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed]
  neutron-metadata-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed,automatic]
  neutron-openvswitch-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed]
Network Nodes:
  neutron-common/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed,automatic]
  neutron-dhcp-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed]
  neutron-l3-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed]
  neutron-lbaas-common/bionic,now 2:12.0.0-0ubuntu1 all [installed,automatic]
  neutron-lbaasv2-agent/bionic,now 2:12.0.0-0ubuntu1 all [installed]
  neutron-metadata-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed,automatic]
  neutron-metering-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed]
  neutron-openvswitch-agent/bionic-updates,now 2:12.1.0-0ubuntu1 all [installed]

Here we have a infrastructure with three network and six compute Nodes and gre tunnels between them.

At the moment we are manipulating the flow entrys in br-tun table=22 with a script to output the packets to every compute/network nodes.

I think, if the patch is included in this packet versions, it would not completely solve the issues described above and seen here.

We will now check the agent_boot_time quick fix mentioned from Bruca Lock in post #41