Kernel2KernelVLAN

Test kernel to kernel connection over VLAN Trunking at NSE side

This example shows that NS Clients and NSE on the one node can find each other.

NS Clients and NSE are using the kernel mechanism to connect to its local ovs forwarder. The NS Client connections are multiplexed over single veth pair interface on the NSE side.

Requires

Make sure that you have completed steps from ovs setup. There is more consumption of heap memory by NSE pod due to vpp process when host is configured with hugepage, so in this case NSE pod should be created with memory limit > 2.2 GB.

Run

Deploy NSC and NSE:

kubectl apply -k https://github.com/networkservicemesh/deployments-k8s/examples/use-cases/Kernel2KernelVLAN?ref=b5a6aa9adf11b838bc0684a0a9e17ad8690dfb52

Wait for applications ready:

kubectl wait --for=condition=ready --timeout=1m pod -l app=nsc-kernel -n ns-kernel2kernel-vlan

kubectl wait --for=condition=ready --timeout=1m pod -l app=nse-kernel -n ns-kernel2kernel-vlan

Choose one ns client pod and nse pod by labels:

NSC=$((kubectl get pods -l app=nsc-kernel -n ns-kernel2kernel-vlan --template '{{range .items}}{{.metadata.name}}{{" "}}{{end}}') | cut -d' ' -f1)
TARGET_IP=$(kubectl exec -ti ${NSC} -n ns-kernel2kernel-vlan -- ip route show | grep 172.16 | cut -d' ' -f1)

NSE=$(kubectl get pods -l app=nse-kernel -n ns-kernel2kernel-vlan --template '{{range .items}}{{.metadata.name}}{{"\n"}}{{end}}')

Ping from NSC to NSE:

kubectl exec ${NSC} -n ns-kernel2kernel-vlan -- ping -c 4 ${TARGET_IP}

Cleanup

Delete ns:

kubectl delete ns ns-kernel2kernel-vlan