Recent debugging of some of the CSIT failures indicates missing table miss entries for some of the Tables, which are very crucial for the overall datapath to work. These entries are created during the initial deployment, and without these entries many of the basic functionalities won't work. It does not make much sense to run the whole CSIT when these flows themselves have not made their way to the switch.

It will be easier to add something to detect at deployment and then maybe dump the entire job if basic default flows are not there. Doing after instance creation might be a little trickier, but we can do it.

jluhrsen says 

"I know of a tripleo deployment logic that will actually fail the deploy if it doesn't see flows in the following tables:

flow_tables = [

       18, 19, 20, 22, 23, 24, 43, 45, 48, 50, 51, 60, 80, 81, 90, 210, 211,

       212, 213, 214, 215, 216, 217, 239, 240, 241, 242, 243, 244, 245, 246, 247

     ]

from:

https://git.openstack.org/cgit/openstack/puppet-tripleo/tree/lib/puppet/functions/synchronize_odl_ovs_flows.rb?id=463d3f3a636de7ccb12a207b0407ba3fb7554d7e