ViewVC Help
View File | Revision Log | Show Annotations | Download File
/cvs/gvpe/doc/gvpe.conf.5.pod
Revision: 1.19
Committed: Thu Aug 7 19:07:02 2008 UTC (15 years, 9 months ago) by pcg
Branch: MAIN
Changes since 1.18: +7 -4 lines
Log Message:
*** empty log message ***

File Contents

# Content
1 =head1 NAME
2
3 gvpe.conf - configuration file for the GNU VPE daemon
4
5 =head1 SYNOPSIS
6
7 udp-port = 407
8 mtu = 1492
9 ifname = vpn0
10
11 node = branch1
12 hostname = 1.2.3.4
13
14 node = branch2
15 hostname = www.example.net
16 udp-port = 500 # this host uses a different udp-port
17
18 node = branch3
19 connect = ondemand
20
21 =head1 DESCRIPTION
22
23 The gvpe config file consists of a series of lines that contain C<variable
24 = value> pairs. Empty lines are ignored. Comments start with a C<#> and
25 extend to the end of the line. They can be used on their own lines, or
26 after any directives. Whitespace is allowed around the C<=> sign or after
27 values, but not within the variable names or values themselves.
28
29 The only exception to the above is the "on" directive that can prefix any
30 C<name = value> setting and will only "execute" it on the named node, or
31 (if the nodename starts with "!") on all nodes except the named one.
32
33 name = value
34 on branch1 loglevel = noise
35 on !branch2 connect = ondemand
36
37 All settings are executed "in order", that is, later settings of the same
38 variable overwrite earlier ones.
39
40 =head1 ANATOMY OF A CONFIG FILE
41
42 Usually, a config file starts with global settings (like the udp port to
43 listen on), followed by node-specific sections that begin with a C<node =
44 nickname> line.
45
46 Every node that is part of the network must have a section that starts
47 with C<node = nickname>. The number and order of the nodes is important
48 and must be the same on all hosts. It is not uncommon for node sections to
49 be completely empty - if the default values are right.
50
51 Node-specific settings can be used at any time. If used before the first
52 node section they will set the default values for all following nodes.
53
54 =head1 CONFIG VARIABLES
55
56 =head2 GLOBAL SETTINGS
57
58 Global settings will affect the behaviour of the running gvpe daemon, that
59 is, they are in some sense node-specific (config files can set different
60 values on different nodes using C<on>), but will affect the behaviour of
61 the gvpe daemon and all connections it creates.
62
63 =over 4
64
65 =item dns-forw-host = hostname/ip
66
67 The dns server to forward dns requests to for the DNS tunnel protocol
68 (default: C<127.0.0.1>, changing it is highly recommended).
69
70 =item dns-forw-port = port-number
71
72 The port where the C<dns-forw-host> is to be contacted (default: C<53>,
73 which is fine in most cases).
74
75 =item dns-max-outstanding = integer-number-of-requests
76
77 The maximum number of outstanding DNS transport requests
78 (default: C<100>). GVPE will never issue more requests then the given
79 limit without receiving replies. In heavily overloaded situations it might
80 help to set this to a low number (e.g. C<3> or even C<1>) to limit the
81 number of parallel requests.
82
83 The default should be working ok for most links.
84
85 =item dns-overlap-factor = float
86
87 The DNS transport uses the minimum request latency (B<min_latency>) seen
88 during a connection as it's timing base. This factor (default: C<0.5>,
89 must be > 0) is multiplied by B<min_latency> to get the maximum sending
90 rate (= minimum send interval), i.e. a factor of C<1> means that a new
91 request might be generated every B<min_latency> seconds, which means on
92 average there should only ever be one outstanding request. A factor of
93 C<0.5> means that GVPE will send requests twice as often as the minimum
94 latency measured.
95
96 For congested or picky dns forwarders you could use a value nearer to or
97 exceeding C<1>.
98
99 The default should be working ok for most links.
100
101 =item dns-send-interval = send-interval-in-seconds
102
103 The minimum send interval (= maximum rate) that the DNS transport will
104 use to send new DNS requests. GVPE will not exceed this rate even when
105 the latency is very low. The default is C<0.01>, which means GVPE will
106 not send more than 100 DNS requests per connection per second. For
107 high-bandwidth links you could go lower, e.g. to C<0.001> or so. For
108 congested or rate-limited links, you might want to go higher, say C<0.1>,
109 C<0.2> or even higher.
110
111 The default should be working ok for most links.
112
113 =item dns-timeout-factor = float
114
115 Factor to multiply the C<min_latency> (see C<dns-overlap-factor>) by to
116 get request timeouts. The default of C<8> means that the DNS transport
117 will resend the request when no reply has been received for longer than
118 eight times the minimum (= expected) latency, assuming the request or
119 reply has been lost.
120
121 For congested links a higher value might be necessary (e.g. C<30>). If
122 the link is very stable lower values (e.g. C<2>) might work
123 nicely. Values near or below C<1> makes no sense whatsoever.
124
125 The default should be working ok for most links but will result in low
126 throughput if packet loss is high.
127
128 =item if-up = relative-or-absolute-path
129
130 Sets the path of a script that should be called immediately after the
131 network interface is initialized (but not neccessarily up). The following
132 environment variables are passed to it (the values are just examples).
133
134 Variables that have the same value on all nodes:
135
136 =over 4
137
138 =item CONFBASE=/etc/gvpe
139
140 The configuration base directory.
141
142 =item IFNAME=vpn0
143
144 The network interface to initialize.
145
146 =item IFTYPE=native # or tincd
147
148 =item IFSUBTYPE=linux # or freebsd, darwin etc..
149
150 The interface type (C<native> or C<tincd>) and the subtype (usually the
151 OS name in lowercase) that this GVPE was configured for. Can be used to
152 select the correct syntax to use for network-related commands.
153
154 =item MTU=1436
155
156 The MTU to set the interface to. You can use lower values (if done
157 consistently on all hosts), but this is usually ineffective.
158
159 =item NODES=5
160
161 The number of nodes in this GVPE network.
162
163 =back
164
165 Variables that are node-specific and with values pertaining to the node
166 running this GVPE:
167
168 =over 4
169
170 =item IFUPDATA=string
171
172 The value of the configuration directive C<if-up-data>.
173
174 =item MAC=fe:fd:80:00:00:01
175
176 The MAC address the network interface has to use.
177
178 Might be used to initialize interfaces on platforms where GVPE does not
179 do this automatically. Please see the C<gvpe.osdep(5)> manpage for
180 platform-specific information.
181
182 =item NODENAME=branch1
183
184 The nickname of the node.
185
186 =item NODEID=1
187
188 The numerical node ID of the node running this instance of GVPE. The first
189 node mentioned in the config file gets ID 1, the second ID 2 and so on.
190
191 =back
192
193 In addition, all node-specific variables (except C<NODEID>) will be
194 available with a postfix of C<_nodeid>, which contains the value for that
195 node, e.g. the C<MAC_1> variable contains the MAC address of node #1, while
196 the C<NODENAME_22> variable contains the name of node #22.
197
198 Here is a simple if-up script:
199
200 #!/bin/sh
201 ip link set $IFNAME up
202 [ $NODENAME = branch1 ] && ip addr add 10.0.0.1 dev $IFNAME
203 [ $NODENAME = branch2 ] && ip addr add 10.1.0.1 dev $IFNAME
204 ip route add 10.0.0.0/8 dev $IFNAME
205
206 More complicated examples (using routing to reduce arp traffic) can be
207 found in the etc/ subdirectory of the distribution.
208
209 =item ifname = devname
210
211 Sets the tun interface name to the given name. The default is OS-specific
212 and most probably something like C<tun0>.
213
214 =item ifpersist = yes|true|on | no|false|off
215
216 Should the tun/tap device be made persistent, that is, should the device
217 stay up even when gvpe exits? Some versions of the tunnel device have
218 problems sending packets when gvpe is restarted in persistent mode, so
219 if the connections can be established but you cannot send packets from
220 the local node, try to set this to C<off> and do an ifconfig down on the
221 device.
222
223 =item ip-proto = numerical-ip-protocol
224
225 Sets the protocol number to be used for the rawip protocol. This is a
226 global option because all hosts must use the same protocol, and since
227 there are no port numbers, you cannot easily run more than one gvpe
228 instance using the same protocol, nor can you share the protocol with
229 other programs.
230
231 The default is 47 (GRE), which has a good chance of tunneling through
232 firewalls (but note that the rawip protocol is not GRE compatible). Other
233 common choices are 50 (IPSEC, ESP), 51 (IPSEC, AH), 4 (IPIP tunnels) or 98
234 (ENCAP, rfc1241)
235
236 =item http-proxy-host = hostname/ip
237
238 The C<http-proxy-*> family of options are only available if gvpe was
239 compiled with the C<--enable-http-proxy> option and enable tunneling of
240 tcp connections through a http proxy server.
241
242 C<http-proxy-host> and C<http-proxy-port> should specify the hostname and
243 port number of the proxy server. See C<http-proxy-loginpw> if your proxy
244 requires authentication.
245
246 Please note that gvpe will still try to resolve all hostnames in the
247 configuration file, so if you are behind a proxy without access to a dns
248 server better use numerical IP addresses.
249
250 To make best use of this option disable all protocols except tcp in your
251 config file and make sure your routers (or all other hosts) are listening
252 on a port that the proxy allows (443, https, is a common choice).
253
254 If you have a router, connecting to it will suffice. Otherwise tcp must be
255 enabled on all hosts.
256
257 Example:
258
259 http-proxy-host = proxy.example.com
260 http-proxy-port = 3128 # 8080 is another common choice
261 http-proxy-auth = schmorp:grumbeere
262
263 =item http-proxy-port = proxy-tcp-port
264
265 The port where your proxy server listens.
266
267 =item http-proxy-auth = login:password
268
269 The optional login and password used to authenticate to the proxy server,
270 seperated by a literal colon (C<:>). Only basic authentication is
271 currently supported.
272
273 =item keepalive = seconds
274
275 Sets the keepalive probe interval in seconds (default: C<60>). After this
276 many seconds of inactivity the daemon will start to send keepalive probe
277 every 5 seconds until it receives a reply from the other end. If no reply
278 is received within 30 seconds, the peer is considered unreachable and the
279 connection is closed.
280
281 =item loglevel = noise|trace|debug|info|notice|warn|error|critical
282
283 Set the logging level. Connection established messages are logged at level
284 C<info>, notable errors are logged with C<error>. Default is C<info>.
285
286 =item mtu = bytes
287
288 Sets the maximum MTU that should be used on outgoing packets (basically
289 the MTU of the outgoing interface) The daemon will automatically calculate
290 maximum overhead (e.g. udp header size, encryption blocksize...) and pass
291 this information to the C<if-up> script.
292
293 Recommended values are 1500 (ethernet), 1492 (pppoe), 1472 (pptp).
294
295 This value must be the minimum of the mtu values of all hosts.
296
297 =item node = nickname
298
299 Not really a config setting but introduces a node section. The nickname is
300 used to select the right configuration section and must be passed as an
301 argument to the gvpe daemon.
302
303 =item node-up = relative-or-absolute-path
304
305 Sets a command (default: none) that should be called whenever a connection
306 is established (even on rekeying operations). Note that node-up/down
307 scripts will be run asynchronously, but execution is serialised, so there
308 will only ever be one such script running.
309
310 In addition to all the variables passed to C<if-up> scripts, the following
311 environment variables will be set:
312
313 =over 4
314
315 =item DESTNODE=branch2
316
317 The name of the remote node.
318
319 =item DESTID=2
320
321 The node id of the remote node.
322
323 =item DESTIP=188.13.66.8
324
325 The numerical IP address of the remote host (gvpe accepts connections from
326 everywhere, as long as the other host can authenticate itself).
327
328 =item DESTPORT=655 # deprecated
329
330 The UDP port used by the other side.
331
332 =item STATE=UP
333
334 Node-up scripts get called with STATE=UP, node-down scripts get called
335 with STATE=DOWN.
336
337 =back
338
339 Here is a nontrivial example that uses nsupdate to update the name => ip
340 mapping in some dns zone:
341
342 #!/bin/sh
343 {
344 echo update delete $DESTNODE.lowttl.example.net. a
345 echo update add $DESTNODE.lowttl.example.net. 1 in a $DESTIP
346 echo
347 } | nsupdate -d -k $CONFBASE:key.example.net.
348
349 =item node-down = relative-or-absolute-path
350
351 Same as C<node-up>, but gets called whenever a connection is lost.
352
353 =item pid-file = path
354
355 The path to the pid file to check and create
356 (default: C<LOCALSTATEDIR/run/gvpe.pid>).
357
358 =item private-key = relative-path-to-key
359
360 Sets the path (relative to the config directory) to the private key
361 (default: C<hostkey>). This is a printf format string so every C<%> must
362 be doubled. A single C<%s> is replaced by the hostname, so you could
363 use paths like C<hostkeys/%s> to fetch the files at the location where
364 C<gvpectrl> puts them.
365
366 Since only the private key file of the current node is used and the
367 private key file should be kept secret per-host to avoid spoofings, it is
368 not recommended to use this feature.
369
370 =item rekey = seconds
371
372 Sets the rekeying interval in seconds (default: C<3600>). Connections are
373 reestablished every C<rekey> seconds.
374
375 =back
376
377 =head2 NODE SPECIFIC SETTINGS
378
379 The following settings are node-specific, that is, every node can have
380 different settings, even within the same gvpe instance. Settings that are
381 set before the first node section set the defaults, settings that are
382 set within a node section only apply to the given node.
383
384 =over 4
385
386 =item allow-direct = nodename
387
388 Allow direct connections to this node. See C<deny-direct> for more info.
389
390 =item compress = yes|true|on | no|false|off
391
392 Wether to compress data packets sent to this host (default: C<yes>).
393 Compression is really cheap even on slow computers and has no size
394 overhead at all, so enabling this is a good idea.
395
396 =item connect = ondemand | never | always | disabled
397
398 Sets the connect mode (default: C<always>). It can be C<always> (always
399 try to establish and keep a connection to the given host), C<never>
400 (never initiate a connection to the given host, but accept connections),
401 C<ondemand> (try to establish a connection when there are outstanding
402 packets in the queue and take it down after the keepalive interval) or
403 C<disabled> (node is bad, don't talk to it).
404
405 =item deny-direct = nodename | *
406
407 Deny direct connections to the specified node (or all nodes when C<*>
408 is given). Only one node can be specified, but you can use multiple
409 C<allow-direct> and C<deny-direct> statements. This only makes sense in
410 networks with routers, as routers are required for indirect connections.
411
412 Sometimes, a node cannot reach some other nodes for reasons of network
413 connectivity. For example, a node behind a firewall that only allows
414 conenctions to/from a single other node in the network. In this case one
415 should specify C<deny-direct = *> and C<allow-direct = othernodename> (the other
416 node I<must> be a router for this to work).
417
418 The algorithm to check wether a connection may be direct is as follows:
419
420 1. Other node mentioned in a C<allow-direct>? If yes, allow the connection.
421
422 2. Other node mentioned in a C<deny-direct>? If yes, deny direct connections.
423
424 3. Allow the connection.
425
426 That is, C<allow-direct> takes precedence over C<deny-direct>.
427
428 The check is done in both directions, i.e. both nodes must allow a direct
429 connection before one is attempted, so you only need to specify connect
430 limitations on one node.
431
432 =item dns-domain = domain-suffix
433
434 The DNS domain suffix that points to the DNS tunnel server for this node.
435
436 The domain must point to a NS record that points to the I<dns-hostname>,
437 i.e.
438
439 dns-domainname = tunnel.example.net
440 dns-hostname = tunnel-server.example.net
441
442 Corresponds to the following DNS entries in the C<example.net> domain:
443
444 tunnel.example.net. NS tunnel-server.example.net.
445 tunnel-server.example.net. A 13.13.13.13
446
447 =item dns-hostname = hostname/ip
448
449 The address to bind the DNS tunnel socket to, similar to the C<hostname>,
450 but for the DNS tunnel protocol only. Default: C<0.0.0.0>, but that might
451 change.
452
453 =item dns-port = port-number
454
455 The port to bind the DNS tunnel socket to. Must be C<53> on DNS tunnel servers.
456
457 =item enable-dns = yes|true|on | no|false|off
458
459 See gvpe.protocol(7) for a description of the DNS transport
460 protocol. Avoid this protocol if you can.
461
462 Enable the DNS tunneling protocol on this node, either as server or as
463 client. Support for this transport protocol is only available when gvpe
464 was compiled using the C<--enable-dns> option.
465
466 =item enable-icmp = yes|true|on | no|false|off
467
468 See gvpe.protocol(7) for a description of the ICMP transport protocol.
469
470 Enable the ICMP transport using icmp packets of type C<icmp-type> on this
471 node.
472
473 =item enable-rawip = yes|true|on | no|false|off
474
475 See gvpe.protocol(7) for a description of the RAW IP transport protocol.
476
477 Enable the RAW IPv4 transport using the C<ip-proto> protocol
478 (default: C<no>).
479
480 =item enable-tcp = yes|true|on | no|false|off
481
482 See gvpe.protocol(7) for a description of the TCP transport protocol.
483
484 Enable the TCPv4 transport using the C<tcp-port> port
485 (default: C<no>). Support for this transport protocol is only available
486 when gvpe was compiled using the C<--enable-tcp> option.
487
488 =item enable-udp = yes|true|on | no|false|off
489
490 See gvpe.protocol(7) for a description of the UDP transport protocol.
491
492 Enable the UDPv4 transport using the C<udp-port> port (default: C<no>,
493 unless no other protocol is enabled for a node, in which case this
494 protocol is enabled automatically).
495
496 NOTE: Please specify C<enable-udp = yes> if you want t use it even though
497 it might get switched on automatically, as some future version might
498 default to another default protocol.
499
500 =item hostname = hostname | ip [can not be defaulted]
501
502 Forces the address of this node to be set to the given dns hostname or ip
503 address. It will be resolved before each connect request, so dyndns should
504 work fine. If this setting is not specified and a router is available,
505 then the router will be queried for the address of this node. Otherwise,
506 the connection attempt will fail.
507
508 =item icmp-type = integer
509
510 Sets the type value to be used for outgoing (and incoming) packets sent
511 via the ICMP transport.
512
513 The default is C<0> (which is C<echo-reply>, also known as
514 "ping-replies"). Other useful values include C<8> (C<echo-request>, a.k.a.
515 "ping") and C<11> (C<time-exceeded>), but any 8-bit value can be used.
516
517 =item if-up-data = value
518
519 The value specified using this directive will be passed to the C<if-up>
520 script in the environment variable C<IFUPDATA>.
521
522 =item inherit-tos = yes|true|on | no|false|off
523
524 Wether to inherit the TOS settings of packets sent to the tunnel when
525 sending packets to this node (default: C<yes>). If set to C<yes> then
526 outgoing tunnel packets will have the same TOS setting as the packets sent
527 to the tunnel device, which is usually what you want.
528
529 =item max-retry = positive-number
530
531 The maximum interval in seconds (default: C<3600>, one hour) between
532 retries to establish a connection to this node. When a connection cannot
533 be established, gvpe uses exponential backoff capped at this value. It's
534 sometimes useful to set this to a much lower value (e.g. C<120>) on
535 connections to routers that usually are stable but sometimes are down, to
536 assure quick reconnections even after longer downtimes.
537
538 =item max-ttl = seconds
539
540 Expire packets that couldn't be sent after this many seconds
541 (default: C<60>). Gvpe will normally queue packets for a node without an
542 active connection, in the hope of establishing a connection soon. This
543 value specifies the maximum lifetime a packet will stay in the queue, if a
544 packet gets older, it will be thrown away.
545
546 =item max-queue = positive-number
547
548 The maximum number of packets that will be queued (default: C<512>)
549 for this node. If more packets are sent then earlier packets will be
550 expired. See C<max-ttl>, above.
551
552 =item router-priority = 0 | 1 | positive-number>=2
553
554 Sets the router priority of the given host (default: C<0>, disabled). If
555 some host tries to connect to another host without a hostname, it asks
556 the router host for it's IP address. The router host is the one with the
557 highest priority larger than C<1> that is currently reachable.
558
559 Make sure all hosts always connect (C<connect = always>) to the router
560 hosts, otherwise connecting to them might be impossible.
561
562 The special value C<1> allows other hosts to route through the router
563 host, but they will never route through it by default. The value C<0>
564 disables routing. The idea behind this is that some hosts can, if
565 required, bump the C<router-priority> setting to higher than C<1> in their
566 local config to route through specific hosts. If C<router-priority> is
567 C<0>, then routing will be refused, so C<1> serves as a "enable, but do
568 not use by default" switch.
569
570 =item tcp-port = port-number
571
572 Similar to C<udp-port> (default: C<655>), but sets the TCP port number.
573
574 =item udp-port = port-number
575
576 Sets the port number used by the UDP protocol (default: C<655>, not
577 officially assigned by IANA!).
578
579 =back
580
581 =head1 CONFIG DIRECTORY LAYOUT
582
583 The default (or recommended) directory layout for the config directory is:
584
585 =over 4
586
587 =item X<gvpe.conf>
588
589 The config file.
590
591 =item X<if-up>
592
593 The if-up script
594
595 =item X<node-up>, X<node-down>
596
597 If used the node up or node-down scripts.
598
599 =item X<hostkey>
600
601 The private key (taken from C<hostkeys/nodename>) of the current host.
602
603 =item X<pubkey/nodename>
604
605 The public keys of the other nodes, one file per node.
606
607 =back
608
609 =head1 SEE ALSO
610
611 gvpe(5), gvpe(8), gvpectrl(8).
612
613 =head1 AUTHOR
614
615 Marc Lehmann <gvpe@schmorp.de>
616