ovn-kubernetes: Connectivity problems with overlay network on windows

I have a connectivity problem with one of my windows nodes and the error messages are not helping me to diagnose the problem.

The following steps are currently working:

• The ovn-kubernetes-nodes service is running.
• The ovn-controller service is running.
• The ovsdb-server and ovs-vswitchd service are also running.
• The kubelet service is running as well.

Kubectl shows the the node has joined the cluster and it is in status ready. I can also deploy pods to the node. The pods are however not being started.

The integration bridge k8s-*** has been created and has been assigned an overlay IP (e.g. 10.254.10.2).

I can ping the following IPs:

• 10.254.10.2 (integration bridge)
• 10.254.10.1 (gateway ip)
• 192.168.254.9 (host network IP)
• 192.168.254.1 (default gateway and next hop IP for ovn-kubernetes-node)
• IP of kubernetes master
• Host IP of other Windows and Linux k8s nodes.

I can also ping the gateway ip of other nodes for example 10.245.0.1, 10.245.1.1, but the ping times are less than a 1ms which is impossibly fast so I do not think the ping is actually being relayed to these nodes.

I cannot ping the integration bridges of other linux/windows nodes (e.g. 10.245.0.2 or 10.245.1.2) and there is no connectivity to pods on other nodes.

The ovn-kubernetes-node.log is full of failing ADD/DEL operations and I constantly see these messages (https://gist.github.com/lanoxx/d3e9090187eda8b193665de6b3119f7b).

I have also already disabled the windows firewall.