22. Some sample configurations

22.1. Small Ethernet LAN with Linux as a router to Radio LAN

Many of you may have small local area networks at home and want to connect the machines on that network to your local radio LAN. This is the type of configuration I use at home. I arranged to have a suitable block of addresses allocated to me that I could capture in a single route for convenience and I use these on my Ethernet LAN. Your local IP coordinator will assist you in doing this if you want to try it as well. The addresses for the Ethernet LAN form a subset of the radio LAN addresses. The following configuration is the actual one for my linux router on my network at home:

                                         . . . . . .
  ___               _________         .
   | Network       /         \     .    Network
   | 44.136.8.96/29|         |    .     44.136.8/24        \ | /
   |               | Linux   |   .                          \|/
   |               |         |  .   _____      __________    |
   |          eth0 | Router  |  .  /     \    /          \   |
   |_______________|         |_____| TNC |____|  Radio   |__/
   |   44.136.8.97 |  and    |  .  \_____/    \__________/
   |               |         | sl0
   |               | Server  | 44.136.8.5
   |               |         |   .
   |               |         |     .
   |               \_________/        .
  _|_                                    . . . . . .

#!/bin/sh
# /etc/rc.net
# This configuration provides one KISS based AX.25 port and one
# Ethernet device.

echo "/etc/rc.net"
echo "  Configuring:"

echo -n "    loopback:"
/sbin/ifconfig lo 127.0.0.1
/sbin/route add 127.0.0.1
echo " done."

echo -n "    ethernet:"
/sbin/ifconfig eth0 44.136.8.97 netmask 255.255.255.248 \
                broadcast 44.136.8.103 up
/sbin/route add 44.136.8.97 eth0
/sbin/route add -net 44.136.8.96 netmask 255.255.255.248 eth0
echo " done."

echo -n "    AX.25: "
kissattach -i 44.136.8.5 -m 512 /dev/ttyS1 4800
ifconfig sl0 netmask 255.255.255.0 broadcast 44.136.8.255
route add -host 44.136.8.5 sl0
route add -net 44.136.8.0 window 1024 sl0

echo -n "    NET/ROM: "
nrattach -i 44.136.8.5 netrom

echo "  Routing:"
/sbin/route add default gw 44.136.8.68 window 1024 sl0
echo "    default route."
echo done.

# end

/etc/ax25/axports

# name  callsign        speed   paclen  window  description
4800    VK2KTJ-0        4800    256     2       144.800 MHz

/etc/ax25/nrports

# name  callsign        alias   paclen  description
netrom  VK2KTJ-9        LINUX   235     Linux Switch Port

/etc/ax25/nrbroadcast

# ax25_name     min_obs def_qual        worst_qual      verbose
4800            1       120             10              1

• You must have IP_FORWARDING enabled in your kernel.

• The AX.25 configuration files are pretty much those used as examples in the earlier sections, refer to those where necessary.

• I've chosen to use an IP address for my radio port that is not within my home network block. I needn't have done so, I could have easily used 44.136.8.97 for that port too.

• 44.136.8.68 is my local IPIP encapsulated gateway and hence is where I point my default route.

• Each of the machines on my Ethernet network have a route:

  route add -net 44.0.0.0 netmask 255.0.0.0 \ gw 44.136.8.97 window 512 mss 512 eth0

  The use of the mss and window parameters means that I can get optimum performance from both local Ethernet and radio based connections.

• I also run my smail, http, ftp and other daemons on the router machine so that it needs to be the only machine to provide others with facilities.

• The router machine is a lowly 386DX20 with a 20Mb hard drive and a very minimal linux configuration.

22.2. IPIP encapsulated gateway configuration

Some information here on tunnelling is out of date. The setup has changed since the 2.0.x kernel, now the "ip" command from the iproute2 package should be used, as described in the Advanced Routing HOWTO.

Linux is now very commonly used for TCP/IP encapsulated gateways around the world. The new tunnel driver supports multiple encapsulated routes and makes the older ipip daemon obsolete.

A typical configuration would look similar to the following.

                                         . . . . . .
  ___               _________        .
   | Network       /         \     .    Network
   | 154.27.3/24   |         |    .     44.136.16/24       \ | /
   |               | Linux   |   .                          \|/
   |               |         |  .   _____      __________    |
   |          eth0 | IPIP    |  .  /     \    /          \   |
___|_______________|         |_____| TNC |____|  Radio   |___/
   |   154.27.3.20 | Gateway |  .  \_____/    \__________/
   |               |         | sl0
   |               |         | 44.136.16.1
   |               |         |    .
   |               |         |     .
   |               \_________/       .
  _|_                                    . . . . . .

The configuration files of interest are:

# /etc/rc.net
# This file is a simple configuration that provides one KISS AX.25
# radio port, one Ethernet device, and utilizes the kernel tunnel driver
# to perform the IPIP encapsulation/decapsulation
#
echo "/etc/rc.net"
echo "  Configuring:"
#
echo -n "    loopback:"
/sbin/ifconfig lo 127.0.0.1
/sbin/route add 127.0.0.1
echo " done."
#
echo -n "    ethernet:"
/sbin/ifconfig eth0 154.27.3.20 netmask 255.255.255.0 \
                broadcast 154.27.3.255 up
/sbin/route add 154.27.3.20 eth0
/sbin/route add -net 154.27.3.0 netmask 255.255.255.0 eth0
echo " done."
#
echo -n "    AX.25: "
kissattach -i 44.136.16.1 -m 512 /dev/ttyS1 4800
/sbin/ifconfig sl0 netmask 255.255.255.0 broadcast 44.136.16.255
/sbin/route add -host 44.136.16.1 sl0
/sbin/route add -net 44.136.16.0 netmask 255.255.255.0 window 1024 sl0
#
echo -n "    tunnel:"
/sbin/ifconfig tunl0 44.136.16.1 mtu 512 up
#
echo done.
#
echo -n "Routing ... "
source /etc/ipip.routes
echo done.
#
# end.

and:

# /etc/ipip.routes
# This file is generated using the munge script
#
/sbin/route add -net 44.134.8.0 netmask 255.255.255.0 tunl0 gw 134.43.26.1
/sbin/route add -net 44.34.9.0 netmask 255.255.255.0 tunl0 gw 174.84.6.17
/sbin/route add -net 44.13.28.0 netmask 255.255.255.0 tunl0 gw 212.37.126.3
   ...
   ...
   ...

/etc/ax25/axports

# name  callsign        speed   paclen  window  description
4800    VK2KTJ-0        4800    256     2       144.800 MHz

Some points to note here are:

• The new tunnel driver uses the gw field in the routing table in place of the pointopoint parameter to specify the address of the remote IPIP gateway. This is why it now supports multiple routes per interface.

• You can configure two network devices with the same address. In this example both the sl0 and the tunl0 devices have been configured with the IP address of the radio port. This is done so that the remote gateway sees the correct address from your gateway in encapsulated datagrams sent to it.

• The route commands used to specify the encapsulated routes can be automatically generated by a modified version of the munge script. This is included below. The route commands would then be written to a separate file and read in using the bash source /etc/ipip.routes command (assuming you called the file with the routing commands /etc/ipip.routes) as illustrated. The source file must be in the NOS route command format.

• Note the use of the window argument on the route command. Setting this parameter to an appropriate value improves the performance of your radio link.

The new tunnel-munge script:

#!/bin/sh
#
# From: Ron Atkinson <[email protected]>
#
#  This script is basically the 'munge' script written by Bdale N3EUA
#  for the IPIP daemon and is modified by Ron Atkinson N8FOW. It's
#  purpose is to convert a KA9Q NOS format gateways route file
#  (usually called 'encap.txt') into a Linux routing table format
#  for the IP tunnel driver.
#
#        Usage: Gateway file on stdin, Linux route format file on stdout.
#               eg.  tunnel-munge < encap.txt > ampr-routes
#
# NOTE: Before you use this script be sure to check or change the
#       following items:
#
#     1) Change the 'Local routes' and 'Misc user routes' sections
#        to routes that apply to your own area (remove mine please!)
#     2) On the fgrep line be sure to change the IP address to YOUR
#        gateway Internet address. Failure to do so will cause serious
#        routing loops.
#     3) The default interface name is 'tunl0'. Make sure this is
#        correct for your system.

echo "#"
echo "# IP tunnel route table built by $LOGNAME on `date`"
echo "# by tunnel-munge script v960307."
echo "#"
echo "# Local routes"
echo "route add -net 44.xxx.xxx.xxx netmask 255.mmm.mmm.mmm dev sl0"
echo "#"
echo "# Misc user routes"
echo "#"
echo "# remote routes"

fgrep encap | grep "^route" | grep -v " XXX.XXX.XXX.XXX" | \
awk '{
        split($3, s, "/")
        split(s[1], n,".")
        if      (n[1] == "")    n[1]="0"
        if      (n[2] == "")    n[2]="0"
        if      (n[3] == "")    n[3]="0"
        if      (n[4] == "")    n[4]="0"
        if      (s[2] == "1")   mask="128.0.0.0"
        else if (s[2] == "2")   mask="192.0.0.0"
        else if (s[2] == "3")   mask="224.0.0.0"
        else if (s[2] == "4")   mask="240.0.0.0"
        else if (s[2] == "5")   mask="248.0.0.0"
        else if (s[2] == "6")   mask="252.0.0.0"
        else if (s[2] == "7")   mask="254.0.0.0"
        else if (s[2] == "8")   mask="255.0.0.0"
        else if (s[2] == "9")   mask="255.128.0.0"
        else if (s[2] == "10")  mask="255.192.0.0"
        else if (s[2] == "11")  mask="255.224.0.0"
        else if (s[2] == "12")  mask="255.240.0.0"
        else if (s[2] == "13")  mask="255.248.0.0"
        else if (s[2] == "14")  mask="255.252.0.0"
        else if (s[2] == "15")  mask="255.254.0.0"
        else if (s[2] == "16")  mask="255.255.0.0"
        else if (s[2] == "17")  mask="255.255.128.0"
        else if (s[2] == "18")  mask="255.255.192.0"
        else if (s[2] == "19")  mask="255.255.224.0"
        else if (s[2] == "20")  mask="255.255.240.0"
        else if (s[2] == "21")  mask="255.255.248.0"
        else if (s[2] == "22")  mask="255.255.252.0"
        else if (s[2] == "23")  mask="255.255.254.0"
        else if (s[2] == "24")  mask="255.255.255.0"
        else if (s[2] == "25")  mask="255.255.255.128"
        else if (s[2] == "26")  mask="255.255.255.192"
        else if (s[2] == "27")  mask="255.255.255.224"
        else if (s[2] == "28")  mask="255.255.255.240"
        else if (s[2] == "29")  mask="255.255.255.248"
        else if (s[2] == "30")  mask="255.255.255.252"
        else if (s[2] == "31")  mask="255.255.255.254"
        else                    mask="255.255.255.255"

if (mask == "255.255.255.255")  
        printf "route add -host %s.%s.%s.%s gw %s dev tunl0\n"\
                ,n[1],n[2],n[3],n[4],$5
else                            
        printf "route add -net %s.%s.%s.%s gw %s netmask %s dev tunl0\n"\
                ,n[1],n[2],n[3],n[4],$5,mask
 }'

echo "#"
echo "# default the rest of amprnet via mirrorshades.ucsd.edu"
echo "route add -net 44.0.0.0 gw 128.54.16.18 netmask 255.0.0.0 dev tunl0"
echo "#"
echo "# the end"

22.3. AXIP encapsulated gateway configuration

Many Amateur Radio Internet gateways encapsulate AX.25, NET/ROM and ROSE in addition to tcp/ip. Encapsulation of AX.25 frames within IP datagrams is described in RFC-1226 by Brian Kantor. Mike Westerhof wrote an implementation of an AX.25 encapsulation daemon for UNIX in 1991. The ax25-utils package includes a marginally enhanced version of it for Linux.

An AXIP encapsulation program accepts AX.25 frames at one end, looks at the destination AX.25 address to determine what IP address to send them to, encapsulates them in a tcp/ip datagram and then transmits them to the appropriate remote destination. It also accepts tcp/ip datagrams that contain AX.25 frames, unwraps them and processes them as if it had received them directly from an AX.25 port. To distinguish IP datagrams containing AX.25 frames from other IP datagrams which don't, AXIP datagrams are coded with a protocol id of 4 (or 94 which is now deprecated). This process is described in RFC-1226.

The ax25ipd program included in the ax25-utils package presents itself as a program supporting a KISS interface across which you pass AX.25 frames, and an interface into the tcp/ip protocols. It is configured with a single configuration file called /etc/ax25/ax25ipd.conf.

#
# ax25ipd configuration file for station floyd.vk5xxx.ampr.org
#
# Select axip transport. 'ip' is what you want for compatibility
# with most other gateways.
#
socket ip
#
# Set ax25ipd mode of operation. (digi or tnc)
#
mode tnc
#
# If you selected digi, you must define a callsign. If you selected
# tnc mode, the callsign is currently optional, but this may change
# in the future! (2 calls if using dual port kiss)
#
#mycall vk5xxx-4
#mycall2 vk5xxx-5
#
# In digi mode, you may use an alias. (2 for dual port)
#
#myalias svwdns
#myalias2 svwdn2
#
# Send an ident every 540 seconds ...
#
#beacon after 540
#btext ax25ip -- tncmode rob/vk5xxx -- Experimental AXIP gateway
#
# Serial port, or pipe connected to a kissattach in my case
#
device /dev/ttyq0
#
# Set the device speed
#
speed 9600
#
# loglevel 0 - no output
# loglevel 1 - config info only
# loglevel 2 - major events and errors
# loglevel 3 - major events, errors, and AX.25 frame trace
# loglevel 4 - all events
# log 0 for the moment, syslog not working yet ...
#
loglevel 2
#
# If we are in digi mode, we might have a real tnc here, so use param to
# set the tnc parameters ...
#
#param 1 20
#
# Broadcast Address definition. Any of the addresses listed will be forwarded
# to any of the routes flagged as broadcast capable routes.
#
broadcast QST-0 NODES-0
#
# ax.25 route definition, define as many as you need.
# format is route (call/wildcard) (ip host at destination)
# ssid of 0 routes all ssid's
#
# route <destcall> <destaddr> [flags]
#
# Valid flags are:
#         b  - allow broadcasts to be transmitted via this route
#         d  - this route is the default route
#
route vk2sut-0 44.136.8.68 b
route vk5xxx 44.136.188.221 b
route vk2abc 44.1.1.1
#
#

22.4. Linking NOS and Linux using a pipe device

Many people like to run some version of NOS under Linux because it has all of the features and facilities they are used to. Most of those people would also like to have the NOS running on their machine capable of talking to the Linux kernel so that they can offer some of the linux capabilities to radio users via NOS.

Brandon S. Allbery, KF8NH, contributed the following information to explain how to interconnect the NOS running on a Linux machine with the kernel code using the Linux pipe device.

Since both Linux and NOS support the slip protocol it is possible to link the two together by creating a slip link. You could do this by using two serial ports with a loopback cable between them, but this would be slow and costly. Linux provides a feature that many other Unix-like operating systems provide called `pipes'. These are special pseudo devices that look like a standard tty device to software but in fact loopback to another pipe device. To use these pipes the first program must open the master end of the pipe, and the open then the second program can open the slave end of the pipe. When both ends are open the programs can communicate with each other simply by writing characters to the pipes in the way they would if they were terminal devices.

To use this feature to connect the Linux Kernel and a copy of NOS, or some other program you first must choose a pipe device to use. You can find one by looking in your /dev directory. The master end of the pipes are named: ptyq[1-f] and the slave end of the pipes are known as: ttyq[1-f]. Remember they come in pairs, so if you select /dev/ptyqf as your master end then you must use /dev/ttyqf as the slave end.

Once you have chosen a pipe device pair to use you should allocate the master end to you linux kernel and the slave end to the NOS program, as the Linux kernel starts first and the master end of the pipe must be opened first. You must also remember that your Linux kernel must have a different IP address to your NOS, so you will need to allocate a unique address for it if you haven't already.

You configure the pipe just as if it were a serial device, so to create the slip link from your linux kernel you can use commands similar to the following:

# /sbin/slattach -s 38400 -p slip /dev/ptyqf &
# /sbin/ifconfig sl0 broadcast 44.255.255.255 pointopoint 44.70.248.67 /
        mtu 1536 44.70.4.88
# /sbin/route add 44.70.248.67 sl0
# /sbin/route add -net 44.0.0.0 netmask 255.0.0.0 gw 44.70.248.67

In this example the Linux kernel has been given IP address 44.70.4.88 and the NOS program is using IP address 44.70.248.67. The route command in the last line simply tells your linux kernel to route all datagrams for the amprnet via the slip link created by the slattach command. Normally you would put these commands into your /etc/rc.d/rc.inet2 file after all your other network configuration is complete so that the slip link is created automatically when you reboot. Note: there is no advantage in using cslip instead of slip as it actually reduces performance because the link is only a virtual one and occurs fast enough that having to compress the headers first takes longer than transmitting the uncompressed datagram.

To configure the NOS end of the link you could try the following:

# you can call the interface anything you want; I use "linux" for convenience.
attach asy ttyqf - slip linux 1024 1024 38400
route addprivate 44.70.4.88 linux

These commands will create a slip port named `linux' via the slave end of the pipe device pair to your linux kernel, and a route to it to make it work. When you have started NOS you should be able to ping and telnet to your NOS from your Linux machine and vice versa. If not, double check that you have made no mistakes especially that you have the addresses configured properly and have the pipe devices around the right way.