For a lab application, I have
A single VM located in Europe-West with an application listening to port 4242 on TCP and UDP, responding with a dummy resposne
A regional load balancer located in Europe-West, that has rules for port 4242 on TCP and UDP and that single node as a backend.
A global load balancer, also with rules for port 4242 on TCP and UDP, with the regoinal LB as backend
Connecting to the VM itself works on both TCP and UDP. I can send packages and I receive the answers. The same goes for the regional LB, to which I can connect just fine via TCP and via UDP and everything works.
However, to the global Load Balancer, I can connect to TCP just fine and I get the answers. However, on UDP, the traffic seems to get lost. From what I can tell, the traffic doesn't even arrive inside the VM - I don't see any incoming packets when looking at a tcpdump port 4242 on a VM - while I'm able to see the TCP packages from the global LB just fine, and I'm also able to observe the TCP and UDP packets from the regional LB.
The introduction YouTube video (https://www.youtube.com/watch?v=3awUwUIv950) explicitly says the global LB works with both TCP and UDP. However, since the rules for TCP and UDP are pretty much identical in my setup, I wonder if that's actually the case? Is this supposed to work?