Comment 48 for bug 1841189

@Wendy, @Peng, I am reproducing this bug, and can you please do the following test:
1. Create a subnet in a network (for example tenant1-mgmt-net has only one subnet of tenant1-mgmt0-subnet0).
2. Under this network, create two instances, vm1 and vm2.
3. Check if natbox ping vm1 is successful?
4. Check if natbox ping vm2 is successful?
5. Check if vm1 ping vm2 is successful?(If steps 3, 4, and 5 are successful, proceed to the next step.)
6. Live migration of vm1, vm2 remains unchanged.
7. When natbox can't ping to vm1, repeat steps 4, 5.
8. At this time,can you successfully ping vm1 and vm2 from the router's namespace?(The command is as follows)
   $ ip netns
   $ sudo ip netns exec "router's name" bash
   $ ping "vm1's ip"
   $ ping "vm2's ip"
9. Check the status of ovs at this time(The command is as follows)
   $: systemctl status ovs-vswitchd