The 1st article presented you the hardware platform and the rationale behind the choices. Let's dive into the subject now!

Requirement

  • Basic Linux/Unix knowledge
  • Service provider networking knowledge

Overview

Several choices were possible, we finally ended up in following the KISS method. The Operating system requirements are:

The hardest path would be:

The objective is to have tight control of the software installed on the appliance. This guarantees the smallest footprint we hope to obtain. For those familiar with OpenWRT, we can reach a tiny image size. My OpenWRT image is 5Mb.

This provides an incredible feature: commit/rollback functionality at the package management level!

The features above are still under study into RARE group. We will introduce these technologies once we feel more confident on how to integrate these technologies into a streamlined deployment process.

Article objective

In this article we will go through the major steps in deploying Debian 10 stable aka Buster in order to prepare freeRouter installation.

Diagrams

[ #002 ] - Cookbook


wget http://ftp.nl.debian.org/debian/dists/buster/main/installer-amd64/current/images/


balenaEtcher can be downloaded here

Via the appliance BIOS settings:

  • activate console port redirection:

Option d'activation du port série

  • configure serial port settings

Now that you have activated console port:

  • plug the USB key on which you previously burnt Debian 10
  • make sure you  set boot option from USB in BIOS settings
  • reboot

You can now proceed to the next step: Debian 10 installation


We will assume that you have installed Debian 10 on the 256 Gb SSD.

Just as a side note during the installation process you'll be prompted the: "Software selection" window, in this steps we will:

  • unselect everything
  • select "SSH server"

Software selection

This will guarantee the tiniest Debian 10 operating system software footprint. We will on demand install the needed packages manually.


On minimal installation, sudo is not installed, so all the software will be done as root.

apt-get update
apt-get install default-jre-headless

The latest DPDK software is needed. We use the Debian 10 backport repository in orcer to get DPDK 19.11.2-1~bpo10+1

echo "deb http://deb.debian.org/debian buster-backports main" | tee /etc/apt/sources.list.d/buster-backports.list
apt-get update
apt-get install dpdk dpdk-dev


dpkg -l | grep dpdk
ii  dpdk                                    19.11.2-1~bpo10+1            amd64        Data Plane Development Kit (runtime)
ii  dpdk-dev                                19.11.2-1~bpo10+1            amd64        Data Plane Development Kit (dev tools)
ii  libdpdk-dev:amd64                       19.11.2-1~bpo10+1            amd64        Data Plane Development Kit (basic development files)


apt-get update
apt-get install unzip net-tools libpcap-dev ethtool default-jre-headless psmisc tcpdump



In this setup we will create a freeRouter folder at the filesystem root directory

mkdir /rtr


cd /rtr 
wget http://freerouter.nop.hu/rtr.jar


cd /rtr 
tar xvf rtr.tar -C /rtr
rm rtr.tar



As freeRouter is handling the networking task, we have to disable the appliance networking. Forgetting to do so will result in conflicts and unpredictable behaviour. 

systemctl set-default multi-user.target
rm /usr/lib/systemd/network/*
SVC="network-manager NetworkManager ModemManager systemd-network-generator systemd-networkd systemd-networkd-wait-online systemd-resolved hostapd wpa_supplicant"
systemctl disable $SVC
systemctl mask $SVC




cat /lib/systemd/system/rtr.service

[Unit]
Description=router processes
Wants=network.target
After=network-pre.target
Before=network.target

[Service]
Type=forking
ExecStart=/rtr/hwdet-all.sh

[Install]
WantedBy=multi-user.target


cat /rtr/hwdet-all.sh

#!/bin/sh

cd /rtr
echo 1 > /proc/sys/net/ipv6/conf/all/disable_ipv6
echo 1 > /proc/sys/net/ipv6/conf/default/disable_ipv6
echo 0 > /proc/sys/net/ipv6/conf/lo/disable_ipv6
ip link set lo up mtu 65535
ip addr add 127.0.0.1/8 dev lo
ip addr add ::1/128 dev lo

# DPDK
echo 96 > /proc/sys/vm/nr_hugepages
modprobe uio_pci_generic

dpdk-devbind.py -b uio_pci_generic 01:00.0 
dpdk-devbind.py -b uio_pci_generic 02:00.0 
dpdk-devbind.py -b uio_pci_generic 05:00.0 
dpdk-devbind.py -b uio_pci_generic 06:00.0 
dpdk-devbind.py -b uio_pci_generic 07:00.0 
dpdk-devbind.py -b uio_pci_generic 08:00.0 

#VETH for CPU_PORT and OOBM_PORT
ip link add veth0a type veth peer name veth0b

ip link set veth0a multicast on
ip link set veth0a allmulti on
ip link set veth0a promisc on
ip link set veth0a mtu 8192
ip link set veth0a up

ip link set veth0b multicast on
ip link set veth0b allmulti on
ip link set veth0b promisc on
ip link set veth0b mtu 8192
ip link set veth0b up

ethtool -K veth0a rx off
ethtool -K veth0a tx off
ethtool -K veth0a sg off
ethtool -K veth0a tso off
ethtool -K veth0a ufo off
ethtool -K veth0a gso off
ethtool -K veth0a gro off
ethtool -K veth0a lro off
ethtool -K veth0a rxvlan off
ethtool -K veth0a txvlan off
ethtool -K veth0a ntuple off
ethtool -K veth0a rxhash off
ethtool --set-eee veth0a eee off

ethtool -K veth0b rx off
ethtool -K veth0b tx off
ethtool -K veth0b sg off
ethtool -K veth0b tso off
ethtool -K veth0b ufo off
ethtool -K veth0b gso off
ethtool -K veth0b gro off
ethtool -K veth0b lro off
ethtool -K veth0b rxvlan off
ethtool -K veth0b txvlan off
ethtool -K veth0b ntuple off
ethtool -K veth0b rxhash off
ethtool --set-eee veth0b eee off

ip link add veth1a type veth peer name veth1b

ip link set veth1a multicast on
ip link set veth1a allmulti on
ip link set veth1a promisc on
ip link set veth1a mtu 1500
ip link set veth1a up

ip link set veth1b multicast on
ip link set veth1b allmulti on
ip link set veth1b promisc on
ip link set veth1b mtu 8192
ip link set veth1b up

ip link set wlan0 up

ethtool -K veth1a rx off
ethtool -K veth1a tx off
ethtool -K veth1a sg off
ethtool -K veth1a tso off
ethtool -K veth1a ufo off
ethtool -K veth1a gso off
ethtool -K veth1a gro off
ethtool -K veth1a lro off
ethtool -K veth1a rxvlan off
ethtool -K veth1a txvlan off
ethtool -K veth1a ntuple off
ethtool -K veth1a rxhash off
ethtool --set-eee veth1a eee off

ethtool -K veth1b rx off
ethtool -K veth1b tx off
ethtool -K veth1b sg off
ethtool -K veth1b tso off
ethtool -K veth1b ufo off
ethtool -K veth1b gso off
ethtool -K veth1b gro off
ethtool -K veth1b lro off
ethtool -K veth1b rxvlan off
ethtool -K veth1b txvlan off
ethtool -K veth1b ntuple off
ethtool -K veth1b rxhash off
ethtool --set-eee veth1b eee off

ip addr flush dev veth1a 
ip addr add 192.168.128.254/24 dev veth1a

#ADD DEFAULT ROUTE to OOBM SDN999
route add default gw 192.168.128.1

# START RTR !
start-stop-daemon -S -b -x /rtr/hwdet-main.sh


chmod u+x /rtr/hwdet-main.sh



echo 1 > /proc/sys/net/ipv6/conf/all/disable_ipv6
echo 1 > /proc/sys/net/ipv6/conf/default/disable_ipv6
echo 0 > /proc/sys/net/ipv6/conf/lo/disable_ipv6
ip link set lo up mtu 65535

Note that IPv6 operation would occur on the host itself, IPv6 will be handled at freeRouter level

echo 96 > /proc/sys/vm/nr_hugepages
modprobe uio_pci_generic

dpdk-devbind.py -b uio_pci_generic 01:00.0 
dpdk-devbind.py -b uio_pci_generic 02:00.0 
dpdk-devbind.py -b uio_pci_generic 05:00.0 
dpdk-devbind.py -b uio_pci_generic 06:00.0 
dpdk-devbind.py -b uio_pci_generic 07:00.0 
dpdk-devbind.py -b uio_pci_generic 08:00.0 

In the stanza above, we configure DPDK (required)

  • Configure HugePages

In this case we use 96 hugepages, this value can be different if you are using a box with different characteristics (# of ports, memory etc.) The objective is to configure a value that is not too high (waste of resources) and not too small. otherwise p4dpdk won't run. In this case this leaves 10 Free HugePages.

grep HugePages_ /proc/meminfo
HugePages_Total:      96
HugePages_Free:       10
HugePages_Rsvd:        0
HugePages_Surp:        0 
  • Activate UIO_PCI_GENERIC driver
  • Bind the interfaces to DPDK, DPDK will control them now. Keep in mind that now they will be invisible from the linux kernel.

This command use device PCI ID. In order to check device PCI ID just issue the below command:

 dpdk-devbind.py --status

Network devices using DPDK-compatible driver
============================================
0000:01:00.0 'I211 Gigabit Network Connection 1539' drv=uio_pci_generic unused=igb
0000:02:00.0 'I211 Gigabit Network Connection 1539' drv=uio_pci_generic unused=igb
0000:05:00.0 'I211 Gigabit Network Connection 1539' drv=uio_pci_generic unused=igb
0000:06:00.0 'I211 Gigabit Network Connection 1539' drv=uio_pci_generic unused=igb
0000:07:00.0 'I211 Gigabit Network Connection 1539' drv=uio_pci_generic unused=igb
0000:08:00.0 'I211 Gigabit Network Connection 1539' drv=uio_pci_generic unused=igb

Network devices using kernel driver
===================================
0000:09:00.0 'AR928X Wireless Network Adapter (PCI-Express) 002a' if=wlan0 drv=ath9k unused=uio_pci_generic 

No 'Baseband' devices detected
==============================

Other Crypto devices
====================
0000:00:1a.0 'Atom Processor Z36xxx/Z37xxx Series Trusted Execution Engine 0f18' unused=uio_pci_generic

No 'Eventdev' devices detected
==============================

No 'Mempool' devices detected
=============================

No 'Compress' devices detected
==============================

No 'Misc (rawdev)' devices detected
=================================== 
  • Configure the appliance OOBM via veth pair (as all physical ports are handled by DPDK and will be invisible from the Linux kernel)
#VETH for CPU_PORT and OOBM_PORT
ip link add veth0a type veth peer name veth0b

ip link set veth0a multicast on
ip link set veth0a allmulti on
ip link set veth0a promisc on
ip link set veth0a mtu 8192
ip link set veth0a up

ip link set veth0b multicast on
ip link set veth0b allmulti on
ip link set veth0b promisc on
ip link set veth0b mtu 8192
ip link set veth0b up

ethtool -K veth0a rx off
ethtool -K veth0a tx off
ethtool -K veth0a sg off
ethtool -K veth0a tso off
ethtool -K veth0a ufo off
ethtool -K veth0a gso off
ethtool -K veth0a gro off
ethtool -K veth0a lro off
ethtool -K veth0a rxvlan off
ethtool -K veth0a txvlan off
ethtool -K veth0a ntuple off
ethtool -K veth0a rxhash off
ethtool --set-eee veth0a eee off

ethtool -K veth0b rx off
ethtool -K veth0b tx off
ethtool -K veth0b sg off
ethtool -K veth0b tso off
ethtool -K veth0b ufo off
ethtool -K veth0b gso off
ethtool -K veth0b gro off
ethtool -K veth0b lro off
ethtool -K veth0b rxvlan off
ethtool -K veth0b txvlan off
ethtool -K veth0b ntuple off
ethtool -K veth0b rxhash off
ethtool --set-eee veth0b eee off

So the above section is pretty straightforward:

  • It creates veth0a / veth0b pair. For those familiar with P4, this is similar to the channel between the control plane (freeRouter) and p4dpdk (dataplane) using CPU_PORT
  • It sets for veth0a/veth0b: multicast/allmulti/promisc flag + mtu=8192
  • It disables TCP offload for veth0a/veth0b

We do the same thing for the Out Of Band management (linux access)

ip link add veth1a type veth peer name veth1b

ip link set veth1a multicast on
ip link set veth1a allmulti on
ip link set veth1a promisc on
ip link set veth1a mtu 1500
ip link set veth1a up

ip link set veth1b multicast on
ip link set veth1b allmulti on
ip link set veth1b promisc on
ip link set veth1b mtu 8192
ip link set veth1b up

ip link set wlan0 up

ethtool -K veth1a rx off
ethtool -K veth1a tx off
ethtool -K veth1a sg off
ethtool -K veth1a tso off
ethtool -K veth1a ufo off
ethtool -K veth1a gso off
ethtool -K veth1a gro off
ethtool -K veth1a lro off
ethtool -K veth1a rxvlan off
ethtool -K veth1a txvlan off
ethtool -K veth1a ntuple off
ethtool -K veth1a rxhash off
ethtool --set-eee veth1a eee off

ethtool -K veth1b rx off
ethtool -K veth1b tx off
ethtool -K veth1b sg off
ethtool -K veth1b tso off
ethtool -K veth1b ufo off
ethtool -K veth1b gso off
ethtool -K veth1b gro off
ethtool -K veth1b lro off
ethtool -K veth1b rxvlan off
ethtool -K veth1b txvlan off
ethtool -K veth1b ntuple off
ethtool -K veth1b rxhash off
ethtool --set-eee veth1b eee off

ip addr flush dev veth1a 
ip addr add 192.168.128.254/24 dev veth1a

Add default route to SDN999 for OOBM return traffic (192.168.128.1 is freeRouter sdn999: we will see the full config later)

route add default gw 192.168.128.1


Effectively start freeRouter main loop

start-stop-daemon -S -b -x /rtr/hwdet-main.sh

This main loop is triggered by the script hwdet-main.sh below:

cat /rtr/hwdet-main.sh 

#!/bin/sh

while (true); do
  cd /rtr/
  stty raw < /dev/tty
  java -Xmx4g -jar /rtr/rtr.jar router /rtr/rtr-
  if [ $? -eq 4 ] ; then
    sync
    reboot -f
  fi
  stty cooked < /dev/tty
  sleep 1
done  


Requirement considerations:

  • The box should run 24x7
  • It must survive a power cut, i.e the service should be restored each time the power is cut for any reasons
  • If no power cut but freeRouter has crashed for any reason, it should be restarted

Let me re-assure you, freeRouter usually don't crash, most often freeRouter has manual or better: auto-upgrades (smile) 

while (true); do
  ...
done  
  • The appliance has 8Gb RAM which is enough for JVM running freeRouter. (Full routing IPv4/IPv6 at the control plane is possible at home!  ← ok this is useless but cool, no? :3 )
    • RAM allocation is for JVM and its tables
    • Additional RAM allocation is for p4dpdk and p4emu, as we have to store the table once for the native code too
    • Lastly the kernel also needs memory, so it's a good idea to leave some free RAM and not give everything to JVM.
java -Xmx4g -jar /rtr/rtr.jar router /rtr/rtr-
  • freeRouter "Cold reboot"  
if [ $? -eq 4 ] ; then
  sync
  reboot -f
fi



Discussion

All the choices have been made in order to make the appliance resilient as much as possible and provide an enjoyable user experience. We will see in a later article, a feature that I love: auto-upgrade. This will keep your appliance up to date over the network with the latest freeRouter train during low traffic period. Of course, for ISP P/PE core router we don't want this, but hey! why not? As soon as all customers are dual homed to 2 different PEs reachable via 2 direct core paths, this can be achieved during low traffic period after having set the metric to infinity on all the PE/P boxes to be upgraded. (use IS-IS overload bit or OSPF max-metric router-lsa)

Conclusion

In this article, we got our hands dirty and manually installed freeRouter with DPDK dataplane from a clean slate environment. This is done on purpose, as I'd like you to understand the whole installation process in detail. There is an automated installation alternative that will install freeRouter also. However this is will install freeRouter with software backend. If your hardware CPU+NIC is compatible you can just replace the software backend by DPDK backend. At that precise point we have a vanilla genuine installation of freeRouter with DPDK dataplane on an appliance that can survive physical wild environment and power cut. We have just now to create the 2 freeRouter configuration files:

ls -l rtr-*
-rw-r--r-- 1 root root  646 Jul 31 17:03 rtr-hw.txt
-rw-r--r-- 1 root root 9027 Aug 25 10:02 rtr-sw.txt


  • freeRouter installation is not complex. It just boils down to installing a basic supported Linux OS, install Java, some 3rd party software and the freeRouter jar and binaries itself
  • In the binary list you'll have a special one called p4dpdk that corresponds to freeRouter DPDK dataplane that emulate RARE P4 program on BMv2 (It does not emulate BMv2 !)
  • Though this installation is manual for pedagogic purpose, the installation can be fully automated, just fire up a VM with a bunch of interfaces and test it ! 
  • The installation proposed is highly resilient and will ease upgrade of the appliance (we will see in subsequent article what it means (wink) )

In the next article, we will configure the freeRouter appliance, start the router, and provide configuration in order to have effective basic ping reachability to the FTTH BROADBAND internal IP.