1 <!DOCTYPE HTML PUBLIC
"-//W3C//DTD HTML 4.01 Transitional//EN"
2 "http://www.w3.org/TR/html4/loose.dtd">
6 <TITLE>L2TPNS Manual
</TITLE>
7 <STYLE TYPE=
"text/css">
13 border-top:
1px solid black;
14 border-bottom:
1px solid black;
15 background-color: lightblue;
20 text-decoration: underline;
25 <H1>L2TPNS Manual
</H1>
27 <LI><A HREF=
"#Overview">Overview
</A></LI>
28 <LI><A HREF=
"#Installation">Installation
</A>
30 <LI><A HREF=
"#Requirements">Requirements
</A></LI>
31 <LI><A HREF=
"#Compile">Compile
</A></LI>
32 <LI><A HREF=
"#Install">Install
</A></LI>
33 <LI><A HREF=
"#Running">Running
</A></LI>
36 <LI><A HREF=
"#Configuration">Configuration
</A>
38 <LI><A HREF=
"#startup-config">startup-config
</A></LI>
39 <LI><A HREF=
"#users">users
</A></LI>
40 <LI><A HREF=
"#ip-pool">ip_pool
</A></LI>
41 <LI><A HREF=
"#build-garden">build-garden
</A></LI>
44 <LI><A HREF=
"#ControllingtheProcess">Controlling the Process
</A>
46 <LI><A HREF=
"#Command-LineInterface">Command-Line Interface
</A></LI>
47 <LI><A HREF=
"#nsctl">nsctl
</A></LI>
48 <LI><A HREF=
"#Signals">Signals
</A></LI>
51 <LI><A HREF=
"#Throttling">Throttling
</A></LI>
52 <LI><A HREF=
"#Interception">Interception
</A></LI>
53 <LI><A HREF=
"#Authentication">Authentication
</A></LI>
54 <LI><A HREF=
"#Plugins">Plugins
</A></LI>
55 <LI><A HREF=
"#Walled Garden">Walled Garden
</A></LI>
56 <LI><A HREF=
"#Clustering">Clustering
</A></LI>
57 <LI><A HREF=
"#Routing">Routing
</A></LI>
58 <LI><A HREF=
"#Performance">Performance
</A></LI>
61 <H2 ID=
"Overview">Overview
</H2>
62 l2tpns is half of a complete L2TP implementation. It supports only the
63 LNS side of the connection.
<P>
65 L2TP (Layer
2 Tunneling Protocol) is designed to allow any layer
2
66 protocol (e.g. Ethernet, PPP) to be tunneled over an IP connection. l2tpns
67 implements PPP over L2TP only.
<P>
69 There are a couple of other L2TP imlementations, of which
<A
70 HREF=
"http://sourceforge.net/projects/l2tpd">l2tpd
</A> is probably the
71 most popular. l2tpd also will handle being either end of a tunnel, and
72 is a lot more configurable than l2tpns. However, due to the way it works,
73 it is nowhere near as scalable.
<P>
75 l2tpns uses the TUN/TAP interface provided by the Linux kernel to receive
76 and send packets. Using some packet manipulation it doesn't require a
77 single interface per connection, as l2tpd does.
<P>
79 This allows it to scale extremely well to very high loads and very high
80 numbers of connections.
<P>
82 It also has a plugin architecture which allows custom code to be run
83 during processing. An example of this is in the walled garden module
87 <EM>Documentation is not my best skill. If you find any problems
88 with this document, or if you wish to contribute, please email
<A
89 HREF=
"mailto:david@dparrish.com?subject=L2TPNS+Documentation">david@dparrish.com
</A>.
</EM><P>
91 <H2 ID=
"Installation">Installation
</H2>
92 <H3 ID=
"Requirements">Requirements
</H3>
95 <LI>Linux kernel version
2.4 or above, with the Tun/Tap interface either
96 compiled in, or as a module.
</LI>
98 <LI>libcli
1.8.0 or greater.
<BR>You can get this from
<A
99 HREF=
"http://sourceforge.net/projects/libcli">http://sourceforge.net/projects/libcli
</A></LI>
102 <H3 ID=
"Compile">Compile
</H3>
104 You can generally get away with just running
<B>make
</B> from the source
105 directory. This will compile the daemon, associated tools and any modules
106 shipped with the distribution.
<P>
108 <H3 ID=
"Install">Install
</H3>
110 After you have successfully compiled everything, run
<B>make
111 install
</B> to install it. By default, the binaries are installed into
112 <EM>/usr/sbin
</EM>, the configuration into
<EM>/etc/l2tpns
</EM>, and the
113 modules into
<EM>/usr/lib/l2tpns
</EM>.
<P>
115 You will definately need to edit the configuration files before you
116 start. See the
<A HREF=
"#Configuration">Configuration
</A> section for
119 <H3 ID=
"Running">Running
</H3>
121 You only need to run
<B>/usr/sbin/l2tpns
</B> as root to start it. It does
122 not detach to a daemon process, so you should perhaps run it from init.
<P>
124 By default there is no log destination set, so all log messages will go to
127 <H2 ID=
"Configuration">Configuration
</H2>
129 All configuration of the software is done from the files installed into
132 <H3 ID=
"startup-config">startup-config
</H3>
134 This is the main configuration file for l2tpns. The format of the file is a
135 list of commands that can be run through the command-line interface. This
136 file can also be written directly by the l2tpns process if a user runs the
137 <EM>write memory
</EM> command, so any comments will be lost. However if your
138 policy is not to write the config by the program, then feel free to comment
139 the file with a # or ! at the beginning of the line.
<P>
141 A list of the possible configuration directives follows. Each of these
142 should be set by a line like:
<P>
144 set configstring
"value"
145 set ipaddress
192.168.1.1
150 <LI><B>debug
</B> (int)
<BR>
151 Sets the level of messages that will be written to the log file. The value
152 should be between
0 and
5, with
0 being no debugging, and
5 being the
153 highest. A rough description of the levels is:
155 <LI VALUE=
0>Critical Errors - Things are probably broken
</LI>
156 <LI>Errors - Things might have gone wrong, but probably will recover
</LI>
157 <LI>Warnings - Just in case you care what is not quite perfect
</LI>
158 <LI>Information - Parameters of control packets
</LI>
159 <LI>Calls - For tracing the execution of the code
</LI>
160 <LI>Packets - Everything, including a hex dump of all packets processed... probably twice
</LI>
162 Note that the higher you set the debugging level, the slower the program
163 will run. Also, at level
5 a LOT of information will be logged. This should
164 only ever be used for working out why it doesn't work at all.
168 <LI><B>log_file
</B> (string)
<BR>
169 This will be where all logging and debugging information is written
170 to. This can be either a filename, such as
<EM>/var/log/l2tpns
</EM>, or
171 the special magic string
<EM>syslog:facility
</EM>, where
<EM>facility
</EM>
172 is any one of the syslog logging facilities, such as local5.
176 <LI><B>l2tp_secret
</B> (string)
<BR>
177 This sets the string that l2tpns will use for authenticating tunnel request.
178 This must be the same as the LAC, or authentication will fail. This will
179 only actually be used if the LAC requests authentication.
183 <LI><B>primary_dns
</B> (ip address)
184 <LI><B>secondary_dns
</B> (ip address)
<BR>
185 Whenever a PPP connection is established, DNS servers will be sent to the
186 user, both a primary and a secondary. If either is set to
0.0.0.0, then that
187 one will not be sent.
191 <LI><B>save_state
</B> (boolean)
<BR>
192 When l2tpns receives a STGTERM it will write out its current
193 ip_address_pool, session and tunnel tables to disk prior to exiting to
194 be re-loaded at startup. The validity of this data is obviously quite
195 short and the intent is to allow an sessions to be retained over a
200 <LI><B>primary_radius
</B> (ip address)
201 <LI><B>secondary_radius
</B> (ip address)
<BR>
202 This sets the radius servers used for both authentication and
203 accounting. If the primary server does not respond, then the
204 secondary radius server will be tried.
208 <LI><B>radius_accounting
</B> (boolean)
<BR>
209 If set to true, then radius accounting packets will be sent. This
210 means that a Start record will be sent when the session is
211 successfully authenticated, and a Stop record will be sent when the
216 <LI><B>radius_secret
</B> (string)
<BR>
217 This secret will be used in all radius queries. If this is not set then
218 radius queries will fail.
222 <LI><B>bind_address
</B> (ip address)
<BR>
223 When the tun interface is created, it is assigned the address
224 specified here. If no address is given,
1.1.1.1 is used. Packets
225 containing user traffic should be routed via this address if given,
226 otherwise the primary address of the machine.
230 <LI><B>send_garp
</B> (boolean)
<BR>
231 Determines whether or not to send a gratuitous ARP for the
232 bind_address when the server is ready to handle traffic (default:
234 This value is ignored if BGP is configured.
238 <LI><B>throttle_speed
</B> (int)
<BR>
239 Sets the speed (in kbits/s) which sessions will be limited to. If this is
240 set to
0, then throttling will not be used at all. Note: You can set this by
241 the CLI, but changes will not affect currently connected users.
245 <LI><B>accounting_dir
</B> (string)
<BR>
246 If set to a directory, then every
5 minutes the current usage for
247 every connected use will be dumped to a file in this directory. Each
248 file dumped begins with a header, where each line is prefixed by #.
249 Following the header is a single line for every connected user, fields
250 separated by a space.
<BR> The fields are username, ip, qos,
251 uptxoctets, downrxoctets. The qos field is
1 if a standard user, and
252 2 if the user is throttled.
256 <LI><B>setuid
</B> (int)
<BR>
257 After starting up and binding the interface, change UID to this. This
258 doesn't work properly.
262 <LI><B>dump_speed
</B> (boolean)
<BR>
263 If set to true, then the current bandwidth utilization will be logged every
264 second. Even if this is disabled, you can see this information by running
265 the
<EM>uptime
</EM> command on the CLI.
269 <LI><B>cleanup_interval
</B> (int)
<BR>
270 Interval between regular cleanups (in seconds).
274 <LI><B>multi_read_count
</B> (int)
<BR>
275 Number of packets to read off each of the UDP and TUN fds when
276 returned as readable by select (default:
10). Avoids incurring the
277 unnecessary system call overhead of select on busy servers.
281 <LI><B>scheduler_fifo
</B> (boolean)
<BR>
282 Sets the scheduling policy for the l2tpns process to SCHED_FIFO. This
283 causes the kernel to immediately preempt any currently SCHED_OTHER
284 (normal) process in favour of l2tpns when it becomes runnable.
285 Ignored on uniprocessor systems.
289 <LI><B>lock_pages
</B> (boolean)
<BR>
290 Keep all pages mapped by the l2tpns process in memory.
294 <LI><B>icmp_rate
</B> (int)
<BR>
295 Maximum number of host unreachable icmp packets to send per second.
299 <LI><B>cluster_address
</B> (ip address)
<BR>
300 Multicast cluster address (default:
239.192.13.13). See the section
301 on
<A HREF=
"#Clustering">Clustering
</A> for more information.
305 <LI><B>cluster_interface
</B> (string)
<BR>
306 Interface for cluster packets (default: eth0).
310 <LI><B>cluster_hb_interval
</B> (int)
<BR>
311 Interval in tenths of a second between cluster heartbeat/pings.
315 <LI><B>cluster_hb_timeout
</B> (int)
<BR>
316 Cluster heartbeat timeout in tenths of a second. A new master will be
317 elected when this interval has been passed without seeing a heartbeat
322 <LI><B>as_number
</B> (int)
<BR>
323 Defines the local AS number for BGP (see
<A HREF=
"#Routing">Routing
</A>).
327 <LI><B>bgp_peer1
</B> (string)
328 <LI><B>bgp_peer1_as
</B> (int)
329 <LI><B>bgp_peer2
</B> (string)
330 <LI><B>bgp_peer2_as
</B> (int)
<BR>
332 DNS name (or IP) and AS number of BGP peers.
336 <H3 ID=
"users">users
</H3>
338 Usernames and passwords for the command-line interface are stored in
339 this file. The format is
<I>username
</I><B>:
</B><I>password
</I> where
340 <I>password
</I> may either by plain text, an MD5 digest (prefixed by
341 <B>$
1</B><I>salt
</I><B>$
</B>) or a DES password, distinguished from
342 plain text by the prefix
<B>{crypt}
</B>.
<P>
344 The username
<B>enable
</B> has a special meaning and is used to set
345 the enable password.
<P>
347 <B>Note:
</B> If this file doesn't exist, then anyone who can get to
348 port
23 will be allowed access without a username / password.
<P>
350 <H3 ID=
"ip-pool">ip_pool
</H3>
352 This file is used to configure the IP address pool which user
353 addresses are assigned from. This file should contain either an IP
354 address or a CIDR network per line. e.g.:
<P>
365 Keep in mind that l2tpns can only handle
65535 connections per
366 process, so don't put more than
65535 IP addresses in the
367 configuration file. They will be wasted.
369 <H3 ID=
"build-garden">build-garden
</H3>
371 The garden plugin on startup creates a NAT table called
"garden" then
372 sources the
<B>build-garden
</B> script to populate that table. All
373 packets from gardened users will be sent through this table. Example:
376 iptables -t nat -A garden -p tcp -m tcp --dport
25 -j DNAT --to
192.168.1.1
377 iptables -t nat -A garden -p udp -m udp --dport
53 -j DNAT --to
192.168.1.1
378 iptables -t nat -A garden -p tcp -m tcp --dport
53 -j DNAT --to
192.168.1.1
379 iptables -t nat -A garden -p tcp -m tcp --dport
80 -j DNAT --to
192.168.1.1
380 iptables -t nat -A garden -p tcp -m tcp --dport
110 -j DNAT --to
192.168.1.1
381 iptables -t nat -A garden -p tcp -m tcp --dport
443 -j DNAT --to
192.168.1.1
382 iptables -t nat -A garden -p icmp -m icmp --icmp-type echo-request -j DNAT --to
192.168.1.1
383 iptables -t nat -A garden -p icmp -j ACCEPT
384 iptables -t nat -A garden -j DROP
387 <H2 ID=
"ControllingtheProcess">Controlling the Process
</H2>
389 A running l2tpns process can be controlled in a number of ways. The primary
390 method of control is by the Command-Line Interface (CLI).
<P>
392 You can also remotely send commands to modules via the nsctl client
393 provided. This currently only works with the walled garden module, but
394 modification is trivial to support other modules.
<P>
396 Also, there are a number of signals that l2tpns understands and takes action
397 when it receives them.
399 <H3 ID=
"Command-LineInterface">Command-Line Interface
</H3>
401 You can access the command line interface by telnet'ing to port
23.
402 There is no IP address restriction, so it's a good idea to firewall
403 this port off from anyone who doesn't need access to it. See
404 <A HREF=
"#users">users
</A> for information on restricting access based
405 on a username and password.
<P>
407 The CLI gives you real-time control over almost everything in
408 the process. The interface is designed to look like a Cisco
409 device, and supports things like command history, line editing and
410 context sensitive help. This is provided by linking with the
411 <A HREF=
"http://sourceforge.net/projects/libcli">libcli
</A>
412 library. Some general documentation of the interface is
413 <A HREF=
"http://sourceforge.net/docman/display_doc.php?docid=20501&group_id=79019">
416 After you have connected to the telnet port (and perhaps logged in), you
417 will be presented with a
<I>hostname
</I><B>></B> prompt.
<P>
419 Enter
<EM>help
</EM> to get a list of possible commands. A brief
420 overview of the more important commands follows:
423 <LI><B>show session
</B><BR>
424 Without specifying a session ID, this will list all tunnels currently
425 connected. If you specify a session ID, you will be given all
426 information on a single tunnel. Note that the full session list can
427 be around
185 columns wide, so you should probably use a wide terminal
428 to see the list properly.
<P>
429 The columns listed in the overview are:
431 <TR><TD><B>SID
</B></TD><TD>Session ID
</TD></TR>
432 <TR><TD><B>TID
</B></TD><TD>Tunnel ID - Use with
<EM>show tunnel tid
</EM></TD></TR>
433 <TR><TD><B>Username
</B></TD><TD>The username given in the PPP
434 authentication. If this is *, then LCP authentication has not
436 <TR><TD><B>IP
</B></TD><TD>The IP address given to the session. If
437 this is
0.0.0.0, LCP negotiation has not completed.
</TD></TR>
438 <TR><TD><B>I
</B></TD><TD>Intercept - Y or N depending on whether the
439 session is being snooped. See
<EM>snoop
</EM>.
</TD></TR>
440 <TR><TD><B>T
</B></TD><TD>Throttled - Y or N if the session is
441 currently throttled. See
<EM>throttle
</EM>.
</TD></TR>
442 <TR><TD><B>G
</B></TD><TD>Walled Garden - Y or N if the user is
443 trapped in the walled garden. This field is present even if the
444 garden module is not loaded.
</TD></TR>
445 <TR><TD><B>opened
</B></TD><TD>The number of seconds since the
446 session started
</TD></TR>
447 <TR><TD><B>downloaded
</B></TD><TD>Number of bytes downloaded by the user
</TD></TR>
448 <TR><TD><B>uploaded
</B></TD><TD>Number of bytes uploaded by the user
</TD></TR>
449 <TR><TD><B>idle
</B></TD><TD>The number of seconds since traffic was
450 detected on the session
</TD></TR>
451 <TR><TD><B>LAC
</B></TD><TD>The IP address of the LAC the session is
452 connected to.
</TD></TR>
453 <TR><TD><B>CLI
</B></TD><TD>The Calling-Line-Identification field
454 provided during the session setup. This field is generated by the
460 <LI><B>show users
</B><BR>
461 With no arguments, display a list of currently connected users. If an
462 argument is given, the session details for the given username are
466 <LI><B>show tunnel
</B><BR>
467 This will show all the open tunnels in a summary, or detail on a single
468 tunnel if you give a tunnel id.
<P>
469 The columns listed in the overview are:
471 <TR><TD><B>TID
</B></TD><TD>Tunnel ID
</TD></TR>
472 <TR><TD><B>Hostname
</B></TD><TD>The hostname for the tunnel as
473 provided by the LAC. This has no relation to DNS, it is just
474 a text field.
</TD></TR>
475 <TR><TD><B>IP
</B></TD><TD>The IP address of the LAC
</TD></TR>
476 <TR><TD><B>State
</B></TD><TD>Tunnel state - Free, Open, Dieing,
478 <TR><TD><B>Sessions
</B></TD><TD>The number of open sessions on the
484 <LI><B>show pool
</B><BR>
485 Displays the current IP address pool allocation. This will only display
486 addresses that are in use, or are reserved for re-allocation to a
487 disconnected user.
<P>
488 If an address is not currently in use, but has been used, then in the User
489 column the username will be shown in square brackets, followed by the time
490 since the address was used:
492 IP Address Used Session User
493 192.168.100.6 N [joe.user]
1548s
498 <LI><B>show radius
</B><BR>
499 Show a summary of the in-use radius sessions. This list should not be very
500 long, as radius sessions should be cleaned up as soon as they are used. The
503 <TR><TD><B>Radius
</B></TD><TD>The ID of the radius request. This is
504 sent in the packet to the radius server for identification.
</TD></TR>
505 <TR><TD><B>State
</B></TD><TD>The state of the request - WAIT, CHAP,
506 AUTH, IPCP, START, STOP, NULL.
</TD></TR>
507 <TR><TD><B>Session
</B></TD><TD>The session ID that this radius
508 request is associated with
</TD></TR>
509 <TR><TD><B>Retry
</B></TD><TD>If a response does not appear to the
510 request, it will retry at this time. This is a unix timestamp.
</TD></TR>
511 <TR><TD><B>Try
</B></TD><TD>Retry count. The radius request is
512 discarded after
3 retries.
</TD></TR>
517 <LI><B>show running-config
</B><BR>
518 This will list the current running configuration. This is in a format that
519 can either be pasted into the configuration file, or run directly at the
524 <LI><B>show counters
</B><BR>
525 Internally, counters are kept of key values, such as bytes and packets
526 transferred, as well as function call counters. This function displays all
527 these counters, and is probably only useful for debugging.
<P>
528 You can reset these counters by running
<EM>clear counters
</EM>.
532 <LI><B>show cluster
</B><BR>
533 Show cluster status. Shows the cluster state for this server
534 (Master/Slave), information about known peers and (for slaves) the
535 master IP address, last packet seen and up-to-date status.
<P>
536 See
<A HREF=
"#Clustering">Clustering
</A> for more information.
540 <LI><B>write memory
</B><BR>
541 This will write the current running configuration to the config file
542 <B>startup-config
</B>, which will be run on a restart.
547 You must specify a username, IP address and port. All packets for the
548 current session for that username will be forwarded to the given
549 host/port. Specify
<EM>no snoop username
</EM> to disable interception
552 If you want interception to be permanent, you will have to modify the radius
553 response for the user. See
<A HREF=
"#Interception">Interception
</A>.
557 <LI><B>throttle
</B><BR>
558 You must specify a username, which will be throttled for the current
559 session. Specify
<EM>no throttle username
</EM> to disable throttling
560 for the current session.
<P>
562 If you want throttling to be permanent, you will have to modify the
563 radius response for the user. See
<A HREF=
"#THrottling">Throttling
</A>.
567 <LI><B>drop session
</B><BR>
568 This will cleanly disconnect a session. You must specify a session id, which
569 you can get from
<EM>show session
</EM>. This will send a disconnect message
574 <LI><B>drop tunnel
</B><BR>
575 This will cleanly disconnect a tunnel, as well as all sessions on that
576 tunnel. It will send a disconnect message for each session individually, and
577 after
10 seconds it will send a tunnel disconnect message.
581 <LI><B>uptime
</B><BR>
582 This will show how long the l2tpns process has been running, and the current
583 bandwidth utilization:
585 17:
10:
35 up
8 days,
2212 users, load average:
0.21,
0.17,
0.16
586 Bandwidth: UDP-ETH:
6/
6 ETH-UDP:
13/
13 TOTAL:
37.6 IN:
3033 OUT:
2569
588 The bandwidth line contains
4 sets of values.
<BR>
589 UDP-ETH is the current bandwidth going from the LAC to the ethernet
590 (user uploads), in mbits/sec.
<BR>
591 ETH-UDP is the current bandwidth going from ethernet to the LAC (user
593 TOTAL is the total aggregate bandwidth in mbits/s.
<BR>
594 IN and OUT are packets/per-second going between UDP-ETH and ETH-UDP.
596 These counters are updated every second.
600 <LI><B>configure terminal
</B><BR>
601 Enter configuration mode. Use
<EM>exit
</EM> or ^Z to exit this mode.
602 The following commands are valid in this mode:
<P>
605 <LI><B>load plugin
</B><BR>
606 Load a plugin. You must specify the plugin name, and it will search in
607 /usr/lib/l2tpns for
<EM>plugin
</EM>.so. You can unload a loaded plugin with
608 <EM>remove plugin
</EM>.
613 Set a configuration variable. You must specify the variable name, and
614 the value. If the value contains any spaces, you should quote the
615 value with double (
") or single (') quotes.<P>
617 You can set any <A HREF="#startup-config
">startup-config</A> value in
618 this way, although some may require a restart to take effect.<P>
622 <H3 ID="nsctl
">nsctl</H3>
624 nsctl was implemented (badly) to allow messages to be passed to modules.<P>
626 You must pass at least 2 parameters: <EM>host</EM> and <EM>command</EM>. The
627 host is the address of the l2tpns server which you want to send the message
630 Command can currently be either <EM>garden</EM> or <EM>ungarden</EM>. With
631 both of these commands, you must give a session ID as the 3rd parameter.
632 This will activate or deactivate the walled garden for a session
635 <H3 ID="Signals
">Signals</H3>
637 While the process is running, you can send it a few different signals, using
643 The signals understood are:
645 <LI>SIGHUP - Reload the config from disk and re-open log file<P></LI>
646 <LI>SIGTERM / SIGINT - Shut down for a restart. This will dump the current
647 state to disk (if <EM>save_state</EM> is set to true). Upon restart, the
648 process will read this saved state to resume active sessions.<P>
649 <LI>SIGQUIT - Shut down cleanly. This will send a disconnect message for
650 every active session and tunnel before shutting down. This is a good idea
651 when upgrading the code, as no sessions will be left with the remote end
652 thinking they are open.</LI>
655 <H2 ID="Throttling
">Throttling</H2>
657 l2tpns contains support for slowing down user sessions to whatever speed you
658 desire. You must first enable the global setting <EM>throttle_speed</EM>
659 before this will be activated.<P>
661 If you wish a session to be throttled permanently, you should set the
662 Vendor-Specific radius value <B>Cisco-Avpair="throttle=yes
"</B>, which
663 will be handled by the <EM>autothrottle</EM> module.<P>
665 Otherwise, you can enable and disable throttling an active session using
666 the <EM>throttle</EM> CLI command.<P>
668 <H2 ID="Interception
">Interception</H2>
670 You may have to deal with legal requirements to be able to intercept a
671 user's traffic at any time. l2tpns allows you to begin and end interception
672 on the fly, as well as at authentication time.<P>
674 When a user is being intercepted, a copy of every packet they send and
675 receive will be sent wrapped in a UDP packet to the IP address and port set
676 in the <EM>snoop_host</EM> and <EM>snoop_port</EM> configuration
679 The UDP packet contains just the raw IP frame, with no extra headers.<P>
681 To enable interception on a connected user, use the <EM>snoop username</EM>
682 and <EM>no snoop username</EM> CLI commands. These will enable interception
685 If you wish the user to be intercepted whenever they reconnect, you will
686 need to modify the radius response to include the Vendor-Specific value
687 <B>Cisco-Avpair="intercept=yes
"</B>. For this feature to be enabled,
688 you need to have the <EM>autosnoop</EM> module loaded.<P>
690 <H2 ID="Authentication
">Authentication</H2>
692 Whenever a session connects, it is not fully set up until authentication is
693 completed. The remote end must send a PPP CHAP or PPP PAP authentication
694 request to l2tpns.<P>
696 This request is sent to the radius server, which will hopefully respond with
697 Auth-Accept or Auth-Reject.<P>
699 If Auth-Accept is received, the session is set up and an IP address is
700 assigned. The radius server can include a Framed-IP-Address field in the
701 reply, and that address will be assigned to the client. It can also include
702 specific DNS servers, and a Framed-Route if that is required.<P>
704 If Auth-Reject is received, then the client is sent a PPP AUTHNAK packet,
705 at which point they should disconnect. The exception to this is when the
706 walled garden module is loaded, in which case the user still receives the
707 PPP AUTHACK, but their session is flagged as being a garden'd user, and they
708 should not receive any service.<P>
710 The radius reply can also contain a Vendor-Specific attribute called
711 Cisco-Avpair. This field is a freeform text field that most Cisco
712 devices understand to contain configuration instructions for the session. In
713 the case of l2tpns it is expected to be of the form
715 key=value,key2=value2,key3=value3,key<EM>n</EM>=<EM>value</EM>
718 Each key-value pair is separated and passed to any modules loaded. The
719 <EM>autosnoop</EM> and <EM>autothrottle</EM> understand the keys
720 <EM>intercept</EM> and <EM>throttle</EM> respectively. For example, to have
721 a user who is to be throttled and intercepted, the Cisco-Avpair value should
724 intercept=yes,throttle=yes
727 <H2 ID="Plugins
">Plugins</H2>
729 So as to make l2tpns as flexible as possible (I know the core code is pretty
730 difficult to understand), it includes a plugin API, which you can use to
731 hook into certain events.<P>
733 There are a few example modules included - autosnoop, autothrottle and
736 When an event happens that has a hook, l2tpns looks for a predefined
737 function name in every loaded module, and runs them in the order the modules
740 The function should return <B>PLUGIN_RET_OK</B> if it is all OK. If it returns
741 <B>PLUGIN_RET_STOP</B>, then it is assumed to have worked, but that no further
742 modules should be run for this event.<P>
743 A return of <B>PLUGIN_RET_ERROR</B> means that this module failed, and
744 no further processing should be done for this event. <EM>Use this with care.</EM>
746 Every event function called takes a specific structure named
747 param_<EM>event</EM>, which varies in content with each event. The
748 function name for each event will be <B>plugin_<EM>event</EM></B>,
749 so for the event <EM>timer</EM>, the function declaration should look like:
751 int plugin_timer(struct param_timer *data);
754 A list of the available events follows, with a list of all the fields in the
756 <TABLE CELLSPACING=0 CELLPADDING=0><TR BGCOLOR=LIGHTGREEN><TD>
757 <TABLE CELLSPACING=1 CELLPADDING=3>
758 <TR BGCOLOR=LIGHTGREEN><TH><B>Event</B></TH><TH><B>Description</B></TH><TH><B>Parameters</B></TH></TR>
759 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>pre_auth</B></TD>
760 <TD>This is called after a radius response has been
761 received, but before it has been processed by the
762 code. This will allow you to modify the response in
767 <LI>t - Tunnel ID</LI>
768 <LI>s - Session ID</LI>
771 <LI>protocol (0xC023 for PAP, 0xC223 for CHAP)</LI>
772 <LI>continue_auth - Set to 0 to stop processing authentication modules</LI>
776 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>post_auth</B></TD>
777 <TD>This is called after a radius response has been
778 received, and the basic checks have been performed. This
779 is what the garden module uses to force authentication
784 <LI>t - Tunnel ID</LI>
785 <LI>s - Session ID</LI>
787 <LI>auth_allowed - This is already set to true or
788 false depending on whether authentication has been
789 allowed so far. You can set this to 1 or 0 to force
790 allow or disallow authentication</LI>
791 <LI>protocol (0xC023 for PAP, 0xC223 for CHAP)</LI>
795 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>packet_rx</B></TD>
796 <TD>This is called whenever a session receives a
797 packet. <FONT COLOR=RED>Use this sparingly, as this will
798 seriously slow down the system.</FONT>
802 <LI>t - Tunnel ID</LI>
803 <LI>s - Session ID</LI>
804 <LI>buf - The raw packet data</LI>
805 <LI>len - The length of buf</LI>
809 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>packet_tx</B></TD>
810 <TD>This is called whenever a session sends a
811 packet. <FONT COLOR=RED>Use this sparingly, as this will
812 seriously slow down the system.</FONT>
816 <LI>t - Tunnel ID</LI>
817 <LI>s - Session ID</LI>
818 <LI>buf - The raw packet data</LI>
819 <LI>len - The length of buf</LI>
823 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>timer</B></TD>
824 <TD>This is run every second, no matter what is happening.
825 This is called from a signal handler, so make sure anything
830 <LI>time_now - The current unix timestamp</LI>
834 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>new_session</B></TD>
835 <TD>This is called after a session is fully set up. The
836 session is now ready to handle traffic.
840 <LI>t - Tunnel ID</LI>
841 <LI>s - Session ID</LI>
845 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>kill_session</B></TD>
846 <TD>This is called when a session is about to be shut down.
847 This may be called multiple times for the same session.
851 <LI>t - Tunnel ID</LI>
852 <LI>s - Session ID</LI>
856 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>radius_response</B></TD>
857 <TD>This is called whenever a radius response includes a
858 Cisco-Avpair value. The value is split up into
859 <EM>key=value</EM> pairs, and each is processed through all
864 <LI>t - Tunnel ID</LI>
865 <LI>s - Session ID</LI>
871 <TR VALIGN=TOP BGCOLOR=WHITE><TD><B>control</B></TD>
872 <TD>This is called in whenever a nsctl packet is received.
873 This should handle the packet and form a response if
878 <LI>buf - The raw packet data</LI>
879 <LI>l - The raw packet data length</LI>
880 <LI>source_ip - Where the request came from</LI>
881 <LI>source_port - Where the request came from</LI>
882 <LI>response - Allocate a buffer and put your response in here</LI>
883 <LI>response_length - Length of response</LI>
884 <LI>send_response - true or false whether a response
885 should be sent. If you set this to true, you must
886 allocate a response buffer.</LI>
887 <LI>type - Type of request (see nsctl.c)</LI>
888 <LI>id - ID of request</LI>
889 <LI>data - I'm really not sure</LI>
890 <LI>data_length - Length of data</LI>
897 <H2 ID="WalledGarden
">Walled Garden</H2>
899 Walled Garden is implemented so that you can provide perhaps limited service
900 to sessions that incorrectly authenticate.<P>
902 Whenever a session provides incorrect authentication, and the
903 radius server responds with Auth-Reject, the walled garden module
904 (if loaded) will force authentication to succeed, but set the flag
905 <EM>garden</EM> in the session structure, and adds an iptables rule to
906 the <B>garden_users</B> chain to force all packets for the session's IP
907 address to traverse the <B>garden</B> chain.<P>
909 This doesn't <EM>just work</EM>. To set this all up, you will to
910 setup the <B>garden</B> nat table with the
911 <A HREF="#build-garden
">build-garden</A> script with rules to limit
912 user's traffic. For example, to force all traffic except DNS to be
913 forwarded to 192.168.1.1, add these entries to your
914 <EM>build-garden</EM>:
916 iptables -t nat -A garden -p tcp --dport ! 53 -j DNAT --to 192.168.1.1
917 iptables -t nat -A garden -p udp --dport ! 53 -j DNAT --to 192.168.1.1
920 l2tpns will add entries to the garden_users chain as appropriate.<P>
922 You can check the amount of traffic being captured using the following
925 iptables -t nat -L garden -nvx
928 <H2 ID="Clustering
">Clustering</H2>
930 An l2tpns cluster consists of of one* or more servers configured with
931 the same configuration, notably the multicast <B>cluster_address</B>.<P>
933 *A stand-alone server is simply a degraded cluster.<P>
935 Initially servers come up as cluster slaves, and periodically (every
936 <B>cluster_hb_interval</B>/10 seconds) send out ping packets
937 containing the start time of the process to the multicast
938 <B>cluster_address</B>.<P>
940 A cluster master sends heartbeat rather than ping packets, which
941 contain those session and tunnel changes since the last heartbeat.<P>
943 When a slave has not seen a heartbeat within
944 <B>cluster_hb_timeout</B>/10 seconds it "elects
" a new master by
945 examining the list of peers it has seen pings from and determines
946 which of these and itself is the "best
" candidate to be master.
947 "Best
" in this context means the server with the highest uptime (the
948 highest IP address is used as a tie-breaker in the case of equal
951 After discovering a master, and determining that it is up-to-date (has
952 seen an update for all in-use sessions and tunnels from heartbeat
953 packets) will raise a route (see <A HREF="#Routing
">Routing</A>) for
954 the <B>bind_address</B> and for all addresses/networks in
955 <B>ip_pool</B>. Any packets recieved by the slave which would alter
956 the session state, as well as packets for throttled or gardened
957 sessions are forwarded to the master for handling. In addition, byte
958 counters for session traffic are periodically forwarded.<P>
960 A master, when determining that it has at least one up-to-date slave
961 will drop all routes (raising them again if all slaves disappear) and
962 subsequently handle only packets forwarded to it by the slaves.<P>
964 <H2 ID="Routing
">Routing</H2>
965 If you are running a single instance, you may simply statically route
966 the IP pools to the <B>bind_address</B> (l2tpns will send a gratuitous
969 For a cluster, configure the members as BGP neighbours on your router
970 and configure multi-path load-balancing. Cisco uses "maximum-paths
971 ibgp
" for IBGP. If this is not supported by your IOS revision, you
972 can use "maximum-paths
" (which works for EBGP) and set
973 <B>as_number</B> to a private value such as 64512.<P>
975 <H2 ID="Performance
">Performance</H2>
977 Performance is great.<P>
979 I'd like to include some pretty graphs here that show a linear performance
980 increase, with no impact by number of connected sessions.<P>
982 That's really what it looks like.<P>
986 <A HREF="mailto:david@dparrish.com?subject=L2TPNS+Documentation
">david@dparrish.com</A>