vl3-scale-from-zero

Test automatic scale from zero

This example shows that vL3-NSEs can be created on the fly on NSC requests. This allows effective scaling for endpoints. The requested endpoint will be automatically spawned on the same node as NSC, allowing the best performance for connectivity.

Run

Deploy NSC and supplier:

kubectl apply -k https://github.com/networkservicemesh/deployments-k8s/examples/features/vl3-scale-from-zero?ref=c91be29099fab1f8376d9ff90c858efd829de35e

Wait for applications ready:

kubectl wait -n ns-vl3-scale-from-zero --for=condition=ready --timeout=1m pod -l app=nse-supplier-k8s

kubectl wait -n ns-vl3-scale-from-zero --for=condition=ready --timeout=1m pod -l app=alpine

kubectl wait -n ns-vl3-scale-from-zero --for=condition=ready --timeout=1m pod -l app=nse-vl3-vpp

Find all nscs:

nscs=$(kubectl  get pods -l app=alpine -o go-template --template="{{range .items}}{{.metadata.name}} {{end}}" -n ns-vl3-scale-from-zero)
[[ ! -z $nscs ]]

Ping each client by each client:

(
for nsc in $nscs 
do
    ipAddr=$(kubectl exec -n ns-vl3-scale-from-zero $nsc -- ifconfig nsm-1) || exit
    ipAddr=$(echo $ipAddr | grep -Eo 'inet addr:[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}\.[0-9]{1,3}'| cut -c 11-)
    for pinger in $nscs
    do
        echo $pinger pings $ipAddr
        kubectl exec $pinger -n ns-vl3-scale-from-zero -- ping -c2 -i 0.5 $ipAddr || exit
    done
done
)

Ping each vl3-nse by each client. Note: By default ipam prefix is 172.16.0.0/16 and client prefix len is 24. We also have two vl3 nses in this example. So we expect to have two vl3 addresses: 172.16.0.0 and 172.16.1.0 that should be accessible by each client.

(
for nsc in $nscs 
do
    echo $nsc pings nses
    kubectl exec -n ns-vl3-scale-from-zero $nsc -- ping 172.16.0.0 -c2 -i 0.5 || exit
    kubectl exec -n ns-vl3-scale-from-zero $nsc -- ping 172.16.1.0 -c2 -i 0.5 || exit
done
)

Cleanup

Delete namespace:

kubectl delete ns ns-vl3-scale-from-zero