[Question] Raven0 traffic is always broadcasted to all vxlan nodes

[suyuee]

What happened: we noticed that all traffic through raven0 is always flooded(broadcast) to all nodes in vxlan. I'm not sure if it's expected, but the bridge fdb command shows the mac addresses are always 0; should they be "learnt" and replaced with actual mac address?

00:00:00:00:00:00 dev raven0 dst <ip1> self permanent
00:00:00:00:00:00 dev raven0 dst <ip2> self permanent

What you expected to happen:

raven0 traffic should flood all vxlan nodes.

How to reproduce it (as minimally and precisely as possible):

Nodes A, B, C are in the same raven vxlan; on node A, pings node B; on Node C, tcpdump -i any icmp, and in our environment, we see the ping requests consistently observed on node C

Anything else we need to know?:

Environment:

• OpenYurt version:
• Kubernetes version (use kubectl version):
• OS (e.g: cat /etc/os-release):
• Kernel (e.g. uname -a):
• Install tools:
• Others:

others
/kind question

[suyuee]

further investigation with ip -d link show raven0 indicates that, raven0 is marked with nolearning (vxlan attributes learning is false by default). My understanding is flooding traffic is expected behavior? changing it from nolearning to learning resolved the flooding behavior. but I'd like to learn how production deployment can work with such setup.

[BSWANG]

@suyuee fixed in openyurtio/raven#165