This document describes how to set up SEBA-in-a-Box (SiaB). SiaB is a functional SEBA pod capable of running E2E tests. It takes about 10 minutes to install on a physical server or VM.

The default configuration of SiaB incorporates an emulated OLT/ONU provided by Ponsim and an emulated AGG switch provided by Mininet. Mininet is also configured with a host that stands in as the BNG and runs a DHCP server. The Ponsim setup installs a single OLT, ONU, and RG. The RG is able to authenticate itself via 802.1x, run dhclient to get an IP address from the DHCP server in Mininet, and finally ping the BNG. This demonstrates end-to-end connectivity between the RG and BNG via the ONU, OLT, and agg switch.

This page describes how to set up SiaB with a physical switch instead of an emulated Mininet topology. An external server running DHCP services connected to the switch acts as the BNG.

Installation procedure


Before installing SiaB, you need a Kubernetes cluster (can be a single node) with the Calico CNI plugin installed. You also need Helm and a few other software packages.

The server or VM on which you are installing SEBA-in-a-Box should have at least two CPU cores, 8GB RAM, and 30GB disk space. Apparmor and SELinux should be disabled.


You need to have Kubernetes with CNI enabled. An easy way to set up a single-node Kubernetes that meets the requirements is with kubeadm. Instructions for installing kubeadm on various platforms can be found here.

NOTE: the setup has not been made to work with minikube; we recommend installing kubeadm instead.

Here’s an example of installing kubeadm on an Ubuntu 16.04 server:

echo "Installing docker..."
sudo apt-get update
sudo apt-get install -y software-properties-common
sudo apt-key adv --keyserver --recv 0EBFCD88
sudo add-apt-repository \
       "deb [arch=amd64] \
       $(lsb_release -cs) \
sudo apt-get update
sudo apt-get install -y "docker-ce=17.03*"

echo "Installing kubeadm..."
sudo apt-get update
sudo apt-get install -y ebtables ethtool apt-transport-https curl
curl -s | sudo apt-key add -
cat <<EOF >/tmp/kubernetes.list
deb kubernetes-xenial main
sudo cp /tmp/kubernetes.list /etc/apt/sources.list.d/kubernetes.list
sudo apt-get update
sudo apt install -y "kubeadm=1.11.3-*" "kubelet=1.11.3-*" "kubectl=1.11.3-*"
sudo swapoff -a
sudo kubeadm init --pod-network-cidr=
mkdir -p $HOME/.kube
sudo cp -i /etc/kubernetes/admin.conf $HOME/.kube/config
sudo chown $(id -u):$(id -g) $HOME/.kube/config

If running on a single node, taint the master node so that we can schedule pods on it:

kubectl taint nodes --all

Calico CNI Plugin

Install the Calico CNI plugin in Kubernetes:

kubectl apply -f \


An example of installing Helm:

echo "Installing helm..."
curl | bash
cat > /tmp/helm.yaml <<EOF
apiVersion: v1
kind: ServiceAccount
  name: helm
  namespace: kube-system
kind: ClusterRoleBinding
  name: helm
  kind: ClusterRole
  name: cluster-admin
  - kind: ServiceAccount
    name: helm
    namespace: kube-system
kubectl create -f /tmp/helm.yaml
helm init --service-account helm
helm repo add incubator

Other prerequisites

Install the http and jq commands. Run: sudo apt install -y httpie jq

Get the Helm charts

Before we can start installing SEBA components, we need to get the charts.

mkdir -p cord
cd cord
git clone

Install Kafka and ONOS

Run these commands:

cd ~/cord/helm-charts
helm repo add incubator
helm install -n cord-kafka -f examples/kafka-single.yaml incubator/kafka
# Wait for Kafka to come up
kubectl wait pod/cord-kafka-0 --for condition=Ready --timeout=180s
helm install -n onos -f configs/onos.yaml onos

Install VOLTHA charts

Run these commands to install VOLTHA:

cd ~/cord/helm-charts
# Install the etcd-operator helm chart:
helm install -n etcd-operator stable/etcd-operator --version 0.8.0
# Allow etcd-operator enough time to create the EtdcCluster
# CustomResourceDefinitions. This should only be a couple of seconds after the
# etcd-operator pods are running. Check the CRD are ready by running the following:
kubectl get crd | grep etcd
# After EtcdCluster CRD is in place
helm dep up voltha
helm install -n voltha -f configs/seba-ponsim.yaml voltha

Before proceeding

Run: kubectl get pod|grep etcd-cluster

You should see the etcd-cluster pod up and running.

$ kubectl get pod|grep etcd-cluster
etcd-cluster-0000                                             1/1       Running     0          20m

Install Ponsim charts

Run these commands to install Ponsim (after installing VOLTHA):

cd ~/cord/helm-charts
helm install -n ponnet ponnet
# Wait for CNI changes
~/cord/helm-charts/scripts/ kube-system
helm install -n ponsimv2 ponsimv2
# Iptables setup
sudo iptables -P FORWARD ACCEPT

Before proceeding

Run: kubectl -n voltha get pod

Make sure that all of the pods in the voltha namespace are in Running state.

$ kubectl -n voltha get pod
NAME                                        READY     STATUS    RESTARTS   AGE
default-http-backend-846b65fb5f-rklfb       1/1       Running   0          6h
freeradius-765c9b486c-6qs7t                 1/1       Running   0          6h
netconf-7d7c96c88b-29cv2                    1/1       Running   0          6h
nginx-ingress-controller-6db99757f7-d9cpk   1/1       Running   0          6h
ofagent-7d7b854cd4-fx6gq                    1/1       Running   0          6h
olt-5455744678-hqbwh                        1/1       Running   0          6h
onu-5df655b9c9-prfjz                        1/1       Running   0          6h
rg-75845c54bc-fjgrf                         1/1       Running   0          6h
vcli-6875544cf-rfdrh                        1/1       Running   0          6h
vcore-0                                     1/1       Running   0          6h
voltha-546cb8fd7f-5n9x4                     1/1       Running   3          6h

If you see the olt pod in CrashLoopBackOff state, try deleting (helm delete --purge) and reinstalling the ponsimv2 chart.

Install Mininet

Run these commands to install Mininet:

cd ~/cord/helm-charts
sudo modprobe openvswitch
helm install -n mininet mininet

After the Mininet pod is running, you can get to the mininet> prompt using:

kubectl attach -ti deployment.apps/mininet

To detach press Ctrl-P Ctrl-Q.

Before proceeding

Run: brctl show

You should see two interfaces on each of the pon0 and pon1 Linux bridges.

$ brctl show
bridge name     bridge id               STP enabled     interfaces
docker0         8000.02429d07b4e2       no
pon0            8000.bec4912b1f6a       no              veth25c1f40b
pon1            8000.0a580a170001       no              veth3cc603fe

Enable pon0 to forward EAPOL packets

This is necessary to enable the RG to authenticate. Run these commands:

echo 8 > /tmp/pon0_group_fwd_mask
sudo cp /tmp/pon0_group_fwd_mask /sys/class/net/pon0/bridge/group_fwd_mask

Install NEM charts

Run these commands:

cd ~/cord/helm-charts
helm dep update xos-core
helm install -n xos-core xos-core
helm dep update xos-profiles/att-workflow
helm install -n att-workflow -f configs/seba-ponsim.yaml xos-profiles/att-workflow

Before proceeding

Run: kubectl get pod

You should see all the NEM pods in Running state, except the att-workflow-tosca-loader pod which should eventually be in Completed state. The latter may go through CrashLoopBackOff state and get restarted a few times first (less than 10). To wait until this occurs you can run:


Load TOSCA into NEM

Run this commands:

helm install -n ponsim-pod xos-profiles/ponsim-pod

Before proceeding

Wait for the ponsim-pod container to reach Completed state, then log into the XOS GUI at http://<hostname>:30001 (credentials: / letmein). You should see an AttWorkflowDriver Service Instance with authentication state AWAITING.

To automatically check for this condition you can run:

http -a GET \ | \
      jq '.items[0].authentication_state' | grep AWAITING

ONOS customizations

Right now it’s necessary to install some custom configuration to ONOS directly. Run this command:

http -a karaf:karaf POST \ defaultVlan=65535

The above command instructs the ONU to exchange untagged packets with the RG, rather than packets tagged with VLAN 0.

At this point the system should be fully installed and functional.

Validating the install

Authenticate the RG

Enter the RG pod in the voltha namespace:

RG_POD=$( kubectl -n voltha get pod -l "app=rg" -o jsonpath='{.items[0]}' )
kubectl -n voltha exec -ti $RG_POD bash

Inside the pod, run this command:

wpa_supplicant -i eth0 -Dwired -c /etc/wpa_supplicant/wpa_supplicant.conf

You should see output like the following:

$ wpa_supplicant -i eth0 -Dwired -c /etc/wpa_supplicant/wpa_supplicant.conf
Successfully initialized wpa_supplicant
eth0: Associated with 01:80:c2:00:00:03
WMM AC: Missing IEs
eth0: CTRL-EVENT-EAP-STARTED EAP authentication started
eth0: CTRL-EVENT-EAP-PROPOSED-METHOD vendor=0 method=4
eth0: CTRL-EVENT-EAP-METHOD EAP vendor 0 method 4 (MD5) selected
eth0: CTRL-EVENT-EAP-SUCCESS EAP authentication completed successfully

Hit Ctrl-C after this point to get back to the shell prompt.

Before proceeding

In the XOS GUI, the AttDriverWorkflow Service Instance should now be in APPROVED state. You can check for this by running:

http -a GET \ | \
      jq '.items[0].authentication_state' | grep APPROVED

The FabricCrossconnect Service Instance should have a check in the Backend status column. You can check for this by running:

http -a GET \ | \
      jq '.items[0].backend_status'|grep OK

Obtain an IP address for the RG

Run the following commands inside the RG pod.

ifconfig eth0

You should see output like the following:

$ dhclient
mv: cannot move '/etc/resolv.conf.dhclient-new.46' to '/etc/resolv.conf': Device or resource busy

You can ignore the Device or resource busy errors. The issue is that /etc/resolv.conf is mounted into the RG container by Kubernetes and dhclient wants to overwrite it.

Before proceeding

Make sure that eth0 inside the RG container has an IP address on the subnet:

$ ifconfig eth0
eth0      Link encap:Ethernet  HWaddr 0a:58:0a:16:00:06
          inet addr:  Bcast:  Mask:
          RX packets:600 errors:0 dropped:559 overruns:0 frame:0
          TX packets:15 errors:0 dropped:0 overruns:0 carrier:0
          collisions:0 txqueuelen:0
          RX bytes:57517 (57.5 KB)  TX bytes:3042 (3.0 KB)

Ping the emulated BNG

The emulated BNG has an IP address of After successfully running dhclient you should be able to ping it from the RG.

$ ping -c 3
PING ( 56(84) bytes of data.
64 bytes from icmp_seq=1 ttl=64 time=34.9 ms
64 bytes from icmp_seq=2 ttl=64 time=39.6 ms
64 bytes from icmp_seq=3 ttl=64 time=37.4 ms

--- ping statistics ---

3 packets transmitted, 3 received, 0% packet loss, time 2002ms

rtt min/avg/max/mdev = 34.940/37.343/39.615/1.917 ms

That’s it. Currently it’s not possible to send traffic to destinations on the Internet.

Convenience script

A Makefile can be used to install SEBA-in-a-Box in an automated manner on an Ubuntu 16.04 system. To try it out:

mkdir -p ~/cord
cd ~/cord
git clone
cd automation-tools/seba-in-a-box

To build a “stable” version of SiaB that uses the service versions specified in the Helm charts:

make stable    # or just ‘make’

To build a SiaB that uses the latest development code:

make latest

After a successful install, you will see the message:

SEBA-in-a-Box installation finished!

If the install fails for some reason, you can re-run the make command and the install will try to resume where it left off.

You can optionally install the logging and nem-monitoring charts during the installation by passing one or both of them (space delimited) via the INFRA_CHARTS variable. E.g.:

make INFRA_CHARTS='logging nem-monitoring' stable

To run the E2E ping test described above, as well as tests of other SEBA functionality, you can run:

make run-tests

Report any problems to acb on the CORD Slack channel.

results matching ""

    No results matching ""