n2n/edge.8

.TH edge 8  "17 Mar 2010" "n2n-2.1" "SUPERUSER COMMANDS"
.SH NAME
edge \- n2n edge node daemon
.SH SYNOPSIS
.B edge
[\-d <tun device>] \-a <tun IP address> \-c <community> {\-k <encrypt key>|\-K <keyfile>} 
[\-s <netmask>] \-l <supernode host:port> [\-L <reg_ttl>]
[\-p <local port>] [\-u <UID>] [\-g <GID>] [-f] [\-m <MAC address>] [\-r] [\-v]
.SH DESCRIPTION
N2N is a peer-to-peer VPN system. Edge is the edge node daemon for n2n which
creates a TAP interface to expose the n2n virtual LAN. On startup n2n creates
the TAP interface and configures it then registers with the supernode so it can
begin to find other nodes in the community.
.PP
.SH OPTIONS
.TP
\-d <name>
sets the TAP device name as seen in ifconfig. Only available on Linux.
.TP
\-a {<addr>|static:<addr>|dhcp:0.0.0.0}
sets the n2n virtual LAN IP address being claimed. This is a private IP
address. All IP addresses in an n2n community typical belong to the same /24
network (ie. only the last octet of the IP addresses varies). If DHCP is used to
assign interface addresses then specify the address as
.B -a dhcp:0.0.0.0 
.TP
\-b
cause edge to perform hostname resolution for the supernode address each time
the supernode is periodically contacted. This can cause reliability problems
because all packet processing stops while the supernode address is resolved
which might take 15 seconds.
.TP
\-c <community>
sets the n2n community name. All edges within the same community appear on the
same LAN (layer 2 network segment). Community name is 16 bytes in length. A name
smaller than this is padded with 0x00 bytes and a name longer than this is
truncated to take the first 16 bytes.
.TP
\-h
write usage then exit.
.TP
\-i <register_interval>
Supernode registration interval. It specifies the interval in seconds
between consecutive REGISTER_SUPER packets and it's used to keep NAT hole
open via the UDP NAT hole punching technique. This only works for asymmetric
NATs and allows for P2P communication.
.TP
\-k <keystring>
sets the twofish encryption key from ASCII text (see also N2N_KEY in
ENVIRONMENT). All edges communicating must use the same key and community
name. If neither -k nor -K is used to specify a key source then edge uses
cleartext mode (no encryption). The -k and -K options are mutually exclusive.
.TP
\-K <keyfile>
Reads a key-schedule file <keyfile> and populates the internal transform
operations with the data found there. This mechanism allows keys to roll at
pre-determined times for a group of hosts. Accurate time synchronisation is not
required as older keys can be decoded for some time after expiry.  If neither -k
nor -K is used to specify a key source then edge uses cleartext mode (no
encryption). The -k and -K options are mutually exclusive.
.TP
\-l <addr>:<port>
sets the n2n supernode IP address and port to register to. Up to 2 supernodes
can be specified by two invocations of -l <addr>:<port>. eg.
.B edge -l 12.34.56.78:7654 -l 98.76.54.32:7654
.
.TP
\-p <num>
binds edge to the given UDP port. Useful for keeping the same external socket
across restarts of edge. This allows peer edges which know the edge socket to
continue p2p operation without going back to the supernode.
.TP
\-t <num>
binds the edge management system to the given UDP port. Default 5644. Use this
if you need to run multiple instance of edge; or something is bound to that
port.
.TP
\-u <uid>
causes the edge process to drop to the given user ID when privileges are no
longer required (UNIX).
.TP
\-g <gid>
causes the edge process to drop to the given group ID when privileges are no
longer required (UNIX).
.TP
\-f
disables daemon mode (UNIX) and causes edge to run in the foreground.
.TP
\-m <MAC>
start the TAP interface with the given MAC address. This is highly recommended
as it means the same address will be used if edge stops and restarts. If this is
not done, the ARP caches of all peers will be wrong and packets will not flow to
this edge until the next ARP refresh.
.TP
\-M <MTU>
set the MTU of the edge interface in bytes. MTU is the largest packet fragment
size allowed to be moved throught the interface. The default is 1400.
.TP
\-s <netmask> 
set the netmask of edge interface in IPv4 dotted decimal notation. The default
is 255.255.255.0 (ie. /24).
.TP
\-r
enable IP packet forwarding/routing through the n2n virtual LAN. Without this
option, IP packets arriving over n2n are dropped if not for the -a <addr> (or
DHCP assigned) IP address of the edge interface.
.TP
\-E 
accept packets destined for multicast ethernet MAC addresses. These addresses
are used in multicast ethernet and IPv6 neighbour discovery. If this option is
not present these multicast packets are discarded as most users do not need or
understand them.
.TP
\-L
set the TTL for the hole punching packet. This is an advanced flag to make
sure that the registration packet is dropped immediately when it goes out of
local nat so that it will not trigger some firewall behavior on target peer.
Actually, the registration packet is only expected to make local nat UDP hole
and is not expected to reach the target peer, see
https://tools.ietf.org/html/rfc5389. To achieve this, the flag should be set as
nat level + 1. For example, if we have 2 layer nat in local, we should set -L 3.
Usually we know exactly how much nat layers in local.
If we are not sure how much nat layers in local, we can use traceroute on
Linux to check. The following example shows a local single layer nat because on
second jump it shows a public ip address. In this case it should set -L 2.

$ /usr/sbin/traceroute -w1 8.8.8.8
traceroute to 8.8.8.8 (8.8.8.8), 30 hops max, 60 byte packets
 1  192.168.3.1 (192.168.3.1)  0.464 ms  0.587 ms  0.719 ms
 2  112.65.17.217 (112.65.17.217)  5.269 ms  7.031 ms  8.666 ms

But this method does not always work due to various local network device policy.
.TP
\-v
more verbose logging (may be specified several times for more verbosity).
.SH ENVIRONMENT
.TP
.B N2N_KEY
set the encryption key so it is not visible on the command line
.SH EXAMPLES
.TP
.B edge \-d n2n0 \-c mynetwork \-k encryptme \-u 99 \-g 99 \-m DE:AD:BE:EF:01:23 \-a 192.168.254.7 \-p 50001 \-l 123.121.120.119:7654

Start edge with TAP device n2n0 on community "mynetwork" with community
supernode at 123.121.120.119 UDP port 7654 and bind the locally used UDP port to
50001. Use "encryptme" as the single permanent shared encryption key. Assign MAC
address DE:AD:BE:EF:01:23 to the n2n interface and drop to user=99 and group=99
after the TAP device is successfull configured.
.PP
Add the -f option to stop edge running as a daemon.
.PP
Somewhere else setup another edge with similar parameters, eg.

.B edge \-d n2n0 \-c mynetwork \-k encryptme \-u 99 \-g 99 \-m DE:AD:BE:EF:01:21 \-a 192.168.254.5 \-p 50001 \-l 123.121.120.119:7654
.PP
Now you can ping from 192.168.254.5 to 192.168.254.7.
.PP
The MAC address (-m <MAC>) and virtual IP address (-a <addr>) must be different
on all edges in the same community.

.SH KEY SCHEDULE FILES
(See
.B n2n_v2(7)
for more details).

The -K <keyfile> option reads a key schedule file.

.B edge \-d n2n0 \-c mynetwork \-K /path/to/file \-u 99 \-g 99 \-m DE:AD:BE:EF:01:21 \-a 192.168.254.5 \-p 50001 \-l 123.121.120.119:7654
.PP

The key schedule file consists of line, one per key in the schedule. The purpose
of key schedules is to encourage regular changing of the encryption keys used by
a community. The file structure also allows for full binary keys to be specified
as compared to the ASCII keys allowed by the single key injection. Each key line
consists of the following:

.B <from> <until> <transform> <data>

<from> and <until> are ASCII decimal values of the UNIX times during which the
key is valid. <transform> is the index of the transform that <data> applies
to. <data> is some text which is parsed by the transform module to derive the
key for that line.

Supported <transform> values are:
.TP
2 = TwoFish
<data> has the form <SA>_<hex_key>. eg.

.B 1252327945 1252328305 2 602_3d7c7769b34b2a4812f8c0e9d87ce9

This specifies security association number 602 and a 16-octet key of numeric
value 0x3d7c7769b34b2a4812f8c0e9d87ce9. <SA> is a 32-bit unsigned integer which
is used to identify the encryption key to the receiver. The SA number is sent
unencrypted so the receiver may find the correct key from the key
schedule. <hex_key> is up to 16 octets although shorter keys are allowed.

.TP
3 = AES-CBC
<data> has the form <SA>_<hex_key>. Same rules as TwoFish.

.SH CLEARTEXT MODE
If neither 
.B -k
nor
.B -K
is specified then edge uses cleartext mode. In cleartext mode there is no
transform of the packet data it is simply encrypted. This is useful for
debugging n2n as packet contents can be seen clearly.

To prevent accidental exposure of data, edge only enters cleartext mode when no
keying parameters are specified. In the case where keying parameters are
specified but no valid keys can be determined, edge exits with an error at
startup. If all keys become invalid while running, edge continues to encode
using the last key that was valid.

.SH MANAGEMENT INTERFACE
Edge provides a very simple management system on UDP port 5644. Send a newline
to receive a status output. Send 'reload' to cause re-read of the
keyfile. Send 'stop' to cause edge to exit cleanly.

.SH EXIT STATUS
edge is a daemon and any exit is an error.
.SH AUTHORS
.TP
Richard Andrews
andrews (at) ntop.org - n2n-1 maintainer and main author of n2n-2
.TP
Luca Deri
deri (at) ntop.org - original author of n2n
.TP
Don Bindner
(--) - significant contributions to n2n-1
.SH SEE ALSO
ifconfig(8) supernode(1) tunctl(8) n2n_v2(7)