Hi,

     Looking further into this defect, it appears that,

1)      There is an exception in getDHCPServerPort, the return of neutronPortCache.getPort(neutronPortId) in not validated. Hence the removelocalflows does not complete its action. 

2)      We have two test-beds and this happens consistently in one test bed only. 

3)      Where the issue is happening, the order of the port delete logs ( when the VM is deleted is ), 

a.      Ovsdbupdate – 

b.      EnqueueEvent ( SouthBound Event ( Port Delete ) – 

c.      EnqueueEvent ( NorthboundEvent Neutron Port Delete )  )

d.      OF13Provider : Delete intf <tap-id> 

e.      Exception.

4)      When the issue is not happening, the order of the port delete is slightly different. 

a.      Ovsdbupdate – 

b.      EnqueueEvent ( SouthBound Event ( Port Delete ) – 

c.      OF13Provider : Delete intf <tap-id> 

d.      EnqueueEvent ( NorthboundEvent Neutron Port Delete )  )

Does this indicate any sort of race condition between southbound and northbound events.

Please provide any pointers/suggestions if you have seen the same.  

Thanks

Badri