CSIT 3node failures - External Network Suite - FIP connectivity failures

https://logs.opendaylight.org/releng/jenkins092/netvirt-csit-3node-openstack-ocata-upstream-stateful-nitrogen/148/log.html.gz#s1-s1-s3-t13

These failures are not there in our 1node (non-HA) job, but are the
same tests. It should also be noted that these failures are happening
before any HA related activities (e.g., taking a controller in/out of
the cluster) are done.

My first suggestion would be to compare flow table entries of the
control/compute nodes in this job vs a passing example in the 1node
job:

https://logs.opendaylight.org/releng/jenkins092/netvirt-csit-1node-openstack-ocata-upstream-stateful-nitrogen/152/