vxlan-6784 dropped packets, no fastdp

[FredoAF]

What you expected to happen?

Fastdp connection to be used

What happened?

logs show that within a second, fastdp connections revert (or are never fully established) to sleeve

Anything else we need to know?

Kubeadm cluster, on prem

$ weave version
2.8.1
$ uname -a
5.17.4
$ kubectl version
1.19

I'm sorry I have none of the logs or other information to hand, but I've been looking into this problem for a while now. On two environments, our weave network, sitting across 12 nodes, tries to start in fastdp and then instantly reverts to weave with no error listed in the logs. Looking at /var/log/messages it looks like the vxlan-6784 interface is created but then goes into a disconnected state. if I run

netstat -i

then I see that the vxlan-6784 interface as dropped transmission packets and no other activity.

What could be causing the interface drop packets like that?