Neutron MTU Path Discovery Fails to Detect Correct MTU in VXLAN-Production Deployments

Goal: Verify that Neutron's mtu_path_discovery feature can automatically determine the correct MTU for tenant networks that use VXLAN encapsulation in a production environment, ensuring that guest instances do not experience packet fragmentation or TCP hangs.

Constraints: The feature's behavior varies across different Linux kernel releases and virtual switch implementations (OVS vs Linux bridge), and path MTU discovery packets may be blocked by intermediate firewalls, making automated detection unreliable. Manual MTU configuration risks silent drops if the underlying fabric does not support jumbo frames.

Questions: Does enabling mtu_path_discovery guarantee that the reported MTU matches the smallest MTU along the encapsulated path for all kernel/vSwitch combinations? If not, what additional verification or fallback mechanisms are required to prevent TCP connection stalls during large-data transfers?