Configuring a lab setup

This document is not strictly related to SEBA, but will be helpful in the case you are trying to setup a POD in you lab.

Overview

In a LAB setup you won't (most likely) use either a real RG or a real BNG, but they are not strictly needed to test a SEBA pod as:

  • the RG can be emulated by an lxc container (from now on client)
  • the BNG can be emulated by an ubuntu server

SEBA Lab Setup

The image above represents the data plane connections in a LAB setup. It does not include the kubernetes cluster for simplicity.

What you'll need to emulate E2E traffic is:

  • 1 x86 server with Ubuntu 16.04 and (at least):
    • 1 1G ethernet port
    • 1 10G ethernet port (this can be a second 1G interface as long as you have a media converter)

Setting up a client

The first thing you need to do is to install lxd on your server. To do that you can follow this guide: http://tutorials.ubuntu.com/tutorial/tutorial-setting-up-lxd-1604

Once lxd is successfully installed you need to initialize it with

lxd init

we recommend to use all the provided default values.

Once lxd is initialized you can create a container using these commands:

lxc launch ubuntu:16.04 <name>
lxc config device add <name> eth1 nic name=eth1 parent=<physical-intf> nictype=physical

where:

  • name is the desired container name, internally we use the ONU serial number as name
  • physical-intf is the name of the interface to which the ONU is physically connected

Once the container is created you canc check it's existance with lxc list. The output should look something like:

+---------------+---------+--------------------+------+------------+-----------+
|     NAME      |  STATE  |        IPV4        | IPV6 |    TYPE    | SNAPSHOTS |
+---------------+---------+--------------------+------+------------+-----------+
| voltha-client | RUNNING | 10.63.3.144 (eth0) |      | PERSISTENT | 0         |
+---------------+---------+--------------------+------+------------+-----------+

Please make sure the container has an assigned IP or we it won't be able to download and install the wpasupplicant.

Once the container is running you need to enter it and do some configuration.

To access the container you can use the lxc exec <name> /bin/bash command.

Once inside:

# activate the interface
ip link set eth1 up
# install the wpasupplicant tool
apt update
apt install -y wpasupplicant jq mz build-essential tcpdump

In the lxc container you also need to install iperf3. Iperf3 needs to be installed from source to have some options used in VOLTHA tests.

git clone https://github.com/esnet/iperf.git -b 3.9
cd iperf
./configure && make && sudo make install

NOTE wpasupplicant is a linux tool to perform 802.1X authentication. More informations here jq is a linux tool to perform json parsing. More information here mz alias for maushezan is a linux tool to perform traffic generations. More informations here iperf3 is a linux tool to perform speed tests. More information here

Create a configuration file for wpasupplicant in /etc/wpa_supplicant/wpa_supplicant.conf with the content:

ctrl_interface=/var/run/wpa_supplicant
eapol_version=1
ap_scan=0
fast_reauth=1
network={
        key_mgmt=WPA-EAP
        eap=MD5
        identity="user"
        password="password"
        ca_cert="/etc/cert/cacert.pem"
        client_cert="/etc/cert/client.pem"
        private_key="/etc/cert/client.key"
        private_key_passwd="whatever"
        eapol_flags=3
}

NOTE that the configuration in this file is not really important if you are using the freeradius server provided as part of the SEBA helm charts. Do not worry if the certificates do not exist, they won't affect authentication as that is password based.

At this point you'll be able kickoff the authentication process (by sending EAPOL packets into the system) with the command:

wpa_supplicant -i eth1 -Dwired -c /etc/wpa_supplicant/wpa_supplicant.conf

If everything has been set up correctly, you should see something similar in the VOLTHA logs:

cord@node1:~$ kubectl logs -f -n voltha vcore-0 | grep -E "packet_indication|packet-in" | grep 888e
20180912T003237.453 DEBUG    MainThread adapter_agent.send_packet_in {adapter_name: openolt, logical_port_no: 16, logical_device_id: 000100000a5a0097, packet: 0180c200000390e2ba82fa8281000ffb888e01000009020100090175736572000000000000000000000000000000000000000000000000000000000000000000, event: send-packet-in, instance_id: compose_voltha_1_1536712228, vcore_id: 0001}

Setting up the emulated BNG

The emulated BNG needs to perform only two operations: DHCP and NAT.

To setup a NAT router on an Ubuntu 16.04 server you can look at this tutorial: https://web.archive.org/web/20191024152605/https://www.nairabytes.net/81-linux/418-how-to-set-up-a-nat-router-on-ubuntu-server-16-04

To install a DHCP server you can follow this tutorial: https://web.archive.org/web/20191119233306/http://nairabytes.net/81-linux/415-how-to-install-a-dhcp-server-in-ubuntu-server-16-04

Once the DHCP server is installed, you need to configure it.

Create Q-in-Q interfaces

On the interface that connects to the Agg Switch (upstream) you are going to receive double tagged traffic, so you'll need to create interfaces to received it.

Supposing that your subscriber is using s_tag=111, c_tag=222 and the upstream interface name is eth2 you can use this commands to create it:

ip link set eth2 up
ip link add link eth2 name eth2.111 type vlan id 111
ip link set eth2.111 up
ip link add link eth2.111 name eth2.111.222 type vlan id 222
ip link set eth2.111.222 up
ip addr add 10.11.2.254/24 dev eth2.111.222

Then you'll need to tell the dhcp server to listen on that interface, you can do that by editing the file /etc/default/isc-dhcp-server so that it looks like:

INTERFACES="eth2.111.222"

NOTE that you can list multiple interfaces, separated by spaces, in case you have multiple subscribers in your setup

In the /etc/dhcp/dhcpd.conf config file, configure the ip address range to assing to the double tagged interface we created:

subnet 10.11.2.0 netmask 255.255.255.0 {
  range 10.11.2.1 10.11.2.100;
  option routers 10.11.2.254;
  option domain-name-servers 8.8.8.8;
}

Installing the required tools on the emulated BNG server

Some tools are required to perform data plane tests present in voltha-system-tests. The following commands install them:

sudo apt update
sudo apt-get install -y jq mz build-essential tcpdump

In the BNG you also need to install iperf3. Iperf3 needs to be installed from source to have some options used in the tests.

#remove existing installation if any
sudo apt-get remove --purge iperf3
#Clone and install from source
git clone https://github.com/esnet/iperf.git -b 3.9
cd iperf
./configure && make && sudo make install

After installing iperf3 on the BNG node it needs to be configured. Create the iperf3.service file:

vi /etc/systemd/system/iperf3.service

Include this content in the newly created file:

[Unit]
Description=iperf3
[Service]
ExecStart=/usr/local/bin/iperf3 --server
[Install]
WantedBy=multi-user.target

Finally, start the iperf3 service

sudo service iperf3 start