Choose node network interface

Like other most CNI implement, hybridnet need an actual network interface of Node to take the responsibility of delivery of packets.

It's important to confirm that which Node NIC need to be used before deploying, or you might have to face the risk of rebooting the machine.

Hybridnet-daemon Pods controls the data plane configuration on every Node, e.g., iptables rule, policy routes, and are deployed by a DaemonSet. We can specify the actual Node network interface by --prefer-vlan-interfaces and --prefer-vxlan-interfaces parameters of hybridnet-daemon DaemonSet. For example:

spec:
  template:
    spec:
      containers:
      - name: cni-daemon
        command:
        - sh
        - /hybridnet/start-daemon.sh
        args:
        - --prefer-vlan-interfaces=eth1,ens22
        - --prefer-vxlan-interfaces=eth0,ens23

Both of the two parameter support to be passed as a string of interface name list. Hybridnet-daemon will check if every interface name in the list (from left to right) exist on node and pick the first exists one as the target network interface.

If you started with Getting Started, or you are using the hybridnet online helm chart, use these commands to make the same configuration as above:

# Change default network type to Underlay
helm upgrade hybridnet hybridnet/hybridnet -n kube-system --set daemon.preferVlanInterfaces='eth1\,ens22' --set daemon.preferVxlanInterfaces='eth0\,ens23' --set daemon.preferVxlanInterfaces='eth0\,ens23'