Comment 40 for bug 1794991

An interesting observation is when use neutron-ovs-cleanup tool on the controller, flows get repopulated quickly after reboot of neutron-ovs-agent. Somehow another problem occurs where I'm not able to do icmp on remote host. This could be due to the fact that I may reboot the compute.

So from a quick extrapolation this could either due to the amount of resources that need to be repopulated during reboot and somehow either timeout (which I didn't see in the logs) or dangling resources that corrupt the repopulation.

Anyone aware of this situation or knows whether neutron-ovs-cleanup is still required.