1 <!DOCTYPE refentry PUBLIC "-//OASIS//DTD DocBook V4.1//EN"
2 "http://www.oasis-open.org/docbook/xml/4.1.2/docbookx.dtd">
5 <date>2012-10-25</date>
9 <application>radsecproxy.conf</application>
11 <manvolnum>5</manvolnum>
12 <refmiscinfo>radsecproxy 1.6.3</refmiscinfo>
16 <application>radsecproxy.conf</application>
18 <refpurpose>Radsec proxy configuration file</refpurpose>
21 <title>Description</title>
23 When the proxy server starts, it will first check the command
24 line arguments, and then read the configuration file. Normally
25 radsecproxy will read the configuration file
26 <filename>/usr/local/etc/radsecproxy.conf</filename>. The command line
27 <option>-c</option> option can be used to instead read an
30 <refentrytitle>radsecproxy</refentrytitle><manvolnum>1</manvolnum>
35 If the configuration file can not be found, the proxy will exit
36 with an error message. Note that there is also an include facility
37 so that any configuration file may include other configuration
38 files. The proxy will also exit on configuration errors.
42 <title>Configuration Syntax</title>
44 When the configuration file is processed, whitespace (spaces and
45 tabs) are generally ignored. For each line, leading and trailing
46 whitespace are ignored. A line is ignored if it is empty, only
47 consists of whitespace, or if the first non-whitespace character
48 is a <literal>#</literal>. The configuration is generally case
49 insensitive, but in some cases the option values (see below) are
53 There are two types of configuration structures than can be
54 used. The first and simplest are lines on the format
55 <emphasis>option value</emphasis>. That is, an option name, see
56 below for a list of valid options, followed by whitespace (at
57 least one space or tab character), followed by a value. Note
58 that if the value contains whitespace, then it must be quoted
59 using <literal>""</literal> or <literal>''</literal>. Any
60 whitespace in front of the option or after the value will be
64 The other type of structure is a block. A block spans at least
65 two lines, and has the format:
66 <blockquote><literallayout>
72 </literallayout></blockquote>
73 That is, some blocktype, see below for a list of the different
74 block types, and then enclosed in braces you have zero or more
75 lines that each have the previously described <emphasis>option
76 value</emphasis> format. Different block types have different
77 rules for which options can be specified, they are listed
78 below. The rules regarding white space, comments and quotes are
79 as above. Hence you may do things like:
80 <blockquote><literallayout>
83 option "value with space"
86 </literallayout></blockquote>
89 Option value characters can also be written in hex. This is done
90 by writing the character <literal>%</literal> followed by two
91 hexadecimal digits. If a <literal>%</literal> is used without
92 two following hexadecimal digits, the <literal>%</literal> and
93 the following characters are used as written. If you want to
94 write a <literal>%</literal> and not use this decoding, you may
95 of course write <literal>%</literal> in hex; i.e.,
96 <literal>%25</literal>.
99 There is one special option that can be used both as a basic
100 option and inside all blocks. That is the option
101 <literal>Include</literal> where the value specifies files to be
102 included. The value can be a single file, or it can use normal
103 shell globbing to specify multiple files, e.g.:
106 include /usr/local/etc/radsecproxy.conf.d/*.conf
109 The files are sorted alphabetically. Included files are read in
110 the order they are specified, when reaching the end of a file,
111 the next file is read. When reaching the end of the last
112 included file, the proxy returns to read the next line following
113 the <literal>Include</literal> option. Included files may again
118 <title>Basic Options</title>
120 The following basic options may be specified in the
121 configuration file. Note that blocktypes and options inside
122 blocks are discussed later. Note that none of these options are
123 required, and indeed in many cases they are not needed. Note
124 that you should specify each at most once. The behaviour with
125 multiple occurences is undefined.
129 <term><literal>PidFile</literal></term>
132 The PidFile option specifies the name of a file to which
133 the process id (PID) will be written. This is overridden
134 by the <option>-i</option> command line option. There is
135 no default value for the PidFile option.
140 <term><literal>LogLevel</literal></term>
143 This option specifies the debug level. It must be set to
144 1, 2, 3, 4 or 5, where 1 logs only serious errors, and 5
145 logs everything. The default is 2 which logs errors,
146 warnings and a few informational messages. Note that the
147 command line option <option>-d</option> overrides this.
152 <term><literal>LogDestination</literal></term>
155 This specifies where the log messages should go. By
156 default the messages go to syslog with facility
157 <literal>LOG_DAEMON</literal>. Using this option you can
158 specify another syslog facility, or you may specify that
159 logging should be to a particular file, not using
160 syslog. The value must be either a file or syslog URL. The
161 file URL is the standard one, specifying a local file that
162 should be used. For syslog, you must use the syntax:
163 <literal>x-syslog:///FACILITY</literal> where
164 <literal>FACILITY</literal> must be one of
165 <literal>LOG_DAEMON</literal>,
166 <literal>LOG_MAIL</literal>, <literal>LOG_USER</literal>,
167 <literal>LOG_LOCAL0</literal>,
168 <literal>LOG_LOCAL1</literal>,
169 <literal>LOG_LOCAL2</literal>,
170 <literal>LOG_LOCAL3</literal>,
171 <literal>LOG_LOCAL4</literal>,
172 <literal>LOG_LOCAL5</literal>,
173 <literal>LOG_LOCAL6</literal> or
174 <literal>LOG_LOCAL7</literal>. You may omit the facility
175 from the URL to specify logging to the default facility,
176 but this is not very useful since this is the default log
177 destination. Note that this option is ignored if
178 <option>-f</option> is specified on the command line.
184 <term><literal>FTicksReporting</literal></term>
187 The FTicksReporting option is used to enable F-Ticks
188 logging and can be set to <literal>None</literal>,
189 <literal>Basic</literal> or <literal>Full</literal>. Its
190 default value is <literal>None</literal>. If
191 FTicksReporting is set to anything other than
192 <literal>None</literal>, note that the default value for
193 FTicksMAC is <literal>VendorKeyHashed</literal> which
194 needs FTicksKey to be set.
197 See <literal>radsecproxy.conf-example</literal> for
198 details. Note that radsecproxy has to be configured with
199 F-Ticks support (<literal>--enable-fticks</literal>) for
200 this option to have any effect.
206 <term><literal>FTicksMAC</literal></term>
209 The FTicksMAC option can be used to control if and how
210 Calling-Station-Id (the users Ethernet MAC address) is
211 being logged. It can be set to one of
212 <literal>Static</literal>, <literal>Original</literal>,
213 <literal>VendorHashed</literal>,
214 <literal>VendorKeyHashed</literal>,
215 <literal>FullyHashed</literal> or
216 <literal>FullyKeyHashed</literal>.
219 The default value for FTicksMAC is
220 <literal>VendorKeyHashed</literal>. This means that
221 FTicksKey has to be set.
223 Before chosing any of <literal>Original</literal>,
224 <literal>FullyHashed</literal> or
225 <literal>VendorHashed</literal>, consider the implications
226 for user privacy when MAC addresses are collected. How
227 will the logs be stored, transferred and accessed?
231 See <literal>radsecproxy.conf-example</literal> for
232 details. Note that radsecproxy has to be configured with
233 F-Ticks support (<literal>--enable-fticks</literal>) for
234 this option to have any effect.
240 <term><literal>FTicksKey</literal></term>
243 The FTicksKey option is used to specify the key to use
244 when producing HMAC's as an effect of specifying
245 VendorKeyHashed or FullyKeyHashed for the FTicksMAC
249 Note that radsecproxy has to be configured with F-Ticks
250 support (<literal>--enable-fticks</literal>) for this
251 option to have any effect.
257 <term><literal>FTicksSyslogFacility</literal></term>
260 The FTicksSyslogFacility option is used to specify a
261 dedicated syslog facility for F-Ticks messages. This
262 allows for easier filtering of F-Ticks messages. If no
263 FTicksSyslogFacility option is given, F-Ticks messages are
264 written to what the LogDestination option specifies.
267 F-Ticks messages are always logged using the log level
268 LOG_DEBUG. Note that specifying a file in
269 FTicksSyslogFacility (using the file:/// prefix) is
276 <term><literal>ListenUDP</literal></term>
279 Normally the proxy will listen to the standard RADIUS UDP
280 port <literal>1812</literal> if configured to handle UDP
281 clients. On most systems it will do this for all of the
282 system's IP addresses (both IPv4 and IPv6). On some
283 systems however, it may respond to only IPv4 or only
284 IPv6. To specify an alternate port you may use a value on
285 the form <literal>*:port</literal> where port is any valid
286 port number. If you also want to specify a specific
288 e.g. <literal>192.168.1.1:1812</literal> or
289 <literal>[2001:db8::1]:1812</literal>. The port may be
290 omitted if you want the default one (like in these
291 examples). These examples are equivalent to
292 <literal>192.168.1.1</literal> and
293 <literal>2001:db8::1</literal>. Note that you must use
294 brackets around the IPv6 address. This option may be
295 specified multiple times to listen to multiple addresses
301 <term><literal>ListenTCP</literal></term>
304 This option is similar to the <literal>ListenUDP</literal>
305 option, except that it is used for receiving connections
306 from TCP clients. The default port number is
307 <literal>1812</literal>.
312 <term><literal>ListenTLS</literal></term>
315 This is similar to the <literal>ListenUDP</literal>
316 option, except that it is used for receiving connections
317 from TLS clients. The default port number is
318 <literal>2083</literal>. Note that this option was
319 previously called <literal>ListenTCP</literal>.
324 <term><literal>ListenDTLS</literal></term>
327 This is similar to the <literal>ListenUDP</literal>
328 option, except that it is used for receiving connections
329 from DTLS clients. The default port number is
330 <literal>2083</literal>.
335 <term><literal>SourceUDP</literal></term>
338 This can be used to specify source address and/or source
339 port that the proxy will use for sending UDP client
340 messages (e.g. Access Request).
345 <term><literal>SourceTCP</literal></term>
348 This can be used to specify source address and/or source
349 port that the proxy will use for TCP connections.
354 <term><literal>SourceTLS</literal></term>
357 This can be used to specify source address and/or source
358 port that the proxy will use for TLS connections.
363 <term><literal>SourceDTLS</literal></term>
366 This can be used to specify source address and/or source
367 port that the proxy will use for DTLS connections.
372 <term><literal>TTLAttribute</literal></term>
375 This can be used to change the default TTL attribute. Only
376 change this if you know what you are doing. The syntax is
377 either a numerical value denoting the TTL attribute, or
378 two numerical values separated by column specifying a
380 i.e. <literal>vendorid:attribute</literal>.
385 <term><literal>AddTTL</literal></term>
388 If a TTL attribute is present, the proxy will decrement
389 the value and discard the message if zero. Normally the
390 proxy does nothing if no TTL attribute is present. If you
391 use the AddTTL option with a value 1-255, the proxy will
392 when forwarding a message with no TTL attribute, add one
393 with the specified value. Note that this option can also
394 be specified for a client/server. It will then override
395 this setting when forwarding a message to that
401 <term><literal>LoopPrevention</literal></term>
404 This can be set to <literal>on</literal> or
405 <literal>off</literal> with <literal>off</literal> being
406 the default. When this is enabled, a request will never be
407 sent to a server named the same as the client it was
408 received from. I.e., the names of the client block and the
409 server block are compared. Note that this only gives
410 limited protection against loops. It can be used as a
411 basic option and inside server blocks where it overrides
417 <term><literal>IPv4Only and IPv6Only</literal></term>
420 These can be set to <literal>on</literal> or
421 <literal>off</literal> with <literal>off</literal> being
422 the default. At most one of <literal>IPv4Only</literal>
423 and <literal>IPv6Only</literal> can be enabled. Enabling
424 <literal>IPv4Only</literal> or <literal>IPv6Only</literal>
425 makes radsecproxy resolve DNS names to the corresponding
426 address family only, and not the other. This is done for
427 both clients and servers. Note that this can be
428 overridden in <literal>client</literal> and
429 <literal>server</literal> blocks, see below.
434 <term><literal>Include</literal></term>
437 This is not a normal configuration option; it can be
438 specified multiple times. It can both be used as a basic
439 option and inside blocks. For the full description, see
440 the configuration syntax section above.
447 <title>Blocks</title>
449 There are five types of blocks, they are
450 <literal>client</literal>, <literal>server</literal>,
451 <literal>realm</literal>, <literal>tls</literal> and
452 <literal>rewrite</literal>. At least one instance of each of
453 <literal>client</literal> and <literal>realm</literal> is
454 required. This is necessary for the proxy to do anything useful,
455 and it will exit if not. The <literal>tls</literal> block is
456 required if at least one TLS/DTLS client or server is
457 configured. Note that there can be multiple blocks for each
458 type. For each type, the block names should be unique. The
459 behaviour with multiple occurences of the same name for the same
460 block type is undefined. Also note that some block option values
461 may reference a block by name, in which case the block name must
462 be previously defined. Hence the order of the blocks may be
467 <title>Client Block</title>
469 The client block is used to configure a client. That is, tell
470 the proxy about a client, and what parameters should be used for
471 that client. The name of the client block must (with one
472 exception, see below) be either the IP address (IPv4 or IPv6) of
473 the client, an IP prefix (IPv4 or IPv6) on the form
474 IpAddress/PrefixLength, or a domain name (FQDN). The way an
475 FQDN is resolved into an IP address may be influenced by the use
476 of the <literal>IPv4Only</literal> and
477 <literal>IPv6Only</literal> options. Note that literal IPv6
478 addresses must be enclosed in brackets.
481 If a domain name is specified, then this will be resolved
482 immediately to all the addresses associated with the name, and
483 the proxy will not care about any possible DNS changes that
484 might occur later. Hence there is no dependency on DNS after
488 When some client later sends a request to the proxy, the proxy
489 will look at the IP address the request comes from, and then go
490 through all the addresses of each of the configured clients (in
491 the order they are defined), to determine which (if any) of the
495 In the case of TLS/DTLS, the name of the client must match the
496 FQDN or IP address in the client certificate. Note that this is
497 not required when the client name is an IP prefix.
500 Alternatively one may use the <literal>host</literal> option
501 inside a client block. In that case, the value of the
502 <literal>host</literal> option is used as above, while the name
503 of the block is only used as a descriptive name for the
504 administrator. The host option may be used multiple times, and
505 can be a mix of addresses, FQDNs and prefixes.
508 The allowed options in a client block are
509 <literal>host</literal>, <literal>IPv4Only</literal>,
510 <literal>IPv6Only</literal>, <literal>type</literal>,
511 <literal>secret</literal>, <literal>tls</literal>,
512 <literal>certificateNameCheck</literal>,
513 <literal>matchCertificateAttribute</literal>,
514 <literal>duplicateInterval</literal>, <literal>AddTTL</literal>,
515 <literal>fticksVISCOUNTRY</literal>,
516 <literal>fticksVISINST</literal>, <literal>rewrite</literal>,
517 <literal>rewriteIn</literal>, <literal>rewriteOut</literal>, and
518 <literal>rewriteAttribute</literal>.
520 We already discussed the <literal>host</literal> option. To
521 specify how radsecproxy should resolve a <literal>host</literal>
522 given as a DNS name, the <literal>IPv4Only</literal> or the
523 <literal>IPv6Only</literal> can be set to <literal>on</literal>.
524 At most one of these options can be enabled. Enabling
525 <literal>IPv4Only</literal> or <literal>IPv6Only</literal> here
526 overrides any basic settings set at the top level.
528 The value of <literal>type</literal> must be one of
529 <literal>udp</literal>, <literal>tcp</literal>,
530 <literal>tls</literal> or <literal>dtls</literal>. The value of
531 <literal>secret</literal> is the shared RADIUS key used with
532 this client. If the secret contains whitespace, the value must
533 be quoted. This option is optional for TLS/DTLS and if omitted
534 will default to "radsec". (Note that using a secret other than
535 "radsec" for TLS is a violation of the standard (RFC 6614) and
536 that the proposed standard for DTLS stipulates that the secret
537 must be "radius/dtls".)
540 For a TLS/DTLS client you may also specify the
541 <literal>tls</literal> option. The option value must be the
542 name of a previously defined TLS block. If this option is not
543 specified, the TLS block with the name
544 <literal>defaultClient</literal> will be used if defined. If not
545 defined, it will try to use the TLS block named
546 <literal>default</literal>. If the specified TLS block name does
547 not exist, or the option is not specified and none of the
548 defaults exist, the proxy will exit with an error.
550 NOTE: All versions of radsecproxy up to and including 1.6
551 erroneously verify client certificate chains using the CA in the
552 very first matching client block regardless of which block is
553 used for the final decision. This was changed in version 1.6.1
554 so that a client block with a different <literal>tls</literal>
555 option than the first matching client block is no longer
556 considered for verification of clients.
560 For a TLS/DTLS client, the option
561 <literal>certificateNameCheck</literal> can be set to
562 <literal>off</literal>, to disable the default behaviour of
563 matching CN or SubjectAltName against the specified hostname or
567 Additional validation of certificate attributes can be done by
568 use of the <literal>matchCertificateAttribute</literal>
569 option. Currently one can only do some matching of CN and
570 SubjectAltName. For regexp matching on CN, one can use the value
571 <literal>CN:/regexp/</literal>. For SubjectAltName one can only
572 do regexp matching of the URI, this is specified as
573 <literal>SubjectAltName:URI:/regexp/</literal>. Note that
574 currently this option can only be specified once in a client
578 The <literal>duplicateInterval</literal> option can be used to
579 specify for how many seconds duplicate checking should be
580 done. If a proxy receives a new request within a few seconds of
581 a previous one, it may be treated the same if from the same
582 client, with the same authenticator etc. The proxy will then
583 ignore the new request (if it is still processing the previous
584 one), or returned a copy of the previous reply.
587 The <literal>AddTTL</literal> option is similar to the
588 <literal>AddTTL</literal> option used in the basic config. See
589 that for details. Any value configured here overrides the basic
590 one when sending messages to this client.
593 The <literal>fticksVISCOUNTRY</literal> option configures
594 clients eligible to F-Ticks logging as defined by the
595 <literal>FTicksReporting</literal> basic option.
598 The <literal>fticksVISINST</literal> option overwrites
599 the default <literal>VISINST</literal> value taken from the client
603 The <literal>rewrite</literal> option is deprecated. Use
604 <literal>rewriteIn</literal> instead.
607 The <literal>rewriteIn</literal> option can be used to refer to
608 a rewrite block that specifies certain rewrite operations that
609 should be performed on incoming messages from the client. The
610 rewriting is done before other processing. For details, see the
611 rewrite block text below. Similarly to <literal>tls</literal>
612 discussed above, if this option is not used, there is a fallback
613 to using the <literal>rewrite</literal> block named
614 <literal>defaultClient</literal> if it exists; and if not, a
615 fallback to a block named <literal>default</literal>.
618 The <literal>rewriteOut</literal> option is used in the same way
619 as <literal>rewriteIn</literal>, except that it specifies
620 rewrite operations that should be performed on outgoing messages
621 to the client. The rewriting is done after other
622 processing. Also, there is no rewrite fallback if this option is
626 The <literal>rewriteAttribute</literal> option currently makes
627 it possible to specify that the User-Name attribute in a client
628 request shall be rewritten in the request sent by the proxy. The
629 User-Name attribute is written back to the original value if a
630 matching response is later sent back to the client. The value
631 must be on the form User-Name:/regexpmatch/replacement/. Example
635 rewriteAttribute User-Name:/^(.*)@local$/\1@example.com/
641 <title>Server Block</title>
643 The server block is used to configure a server. That is, tell
644 the proxy about a server, and what parameters should be used
645 when communicating with that server. The name of the server
646 block must (with one exception, see below) be either the IP
647 address (IPv4 or IPv6) of the server, or a domain name
648 (FQDN). If a domain name is specified, then this will be
649 resolved immediately to all the addresses associated with the
650 name, and the proxy will not care about any possible DNS changes
651 that might occur later. Hence there is no dependency on DNS
652 after startup. If the domain name resolves to multiple
653 addresses, then for UDP/DTLS the first address is used. For
654 TCP/TLS, the proxy will loop through the addresses until it can
655 connect to one of them. The way an FQDN is resolved into an IP
656 address may be influenced by the use of the
657 <literal>IPv4Only</literal> and <literal>IPv6Only</literal>
658 options. In the case of TLS/DTLS, the name of the server must
659 match the FQDN or IP address in the server certificate.
662 Alternatively one may use the <literal>host</literal> option
663 inside a server block. In that case, the value of the
664 <literal>host</literal> option is used as above, while the name
665 of the block is only used as a descriptive name for the
666 administrator. Note that multiple host options may be used. This
667 will then be treated as multiple names/addresses for the same
668 server. When initiating a TCP/TLS connection, all addresses of
669 all names may be attempted, but there is no failover between the
670 different host values. For failover one must use separate server
674 Note that the name of the block, or values of host options may
675 include a port number (separated with a column). This port
676 number will then override the default port or a port option in
677 the server block. Also note that literal IPv6 addresses must be
678 enclosed in brackets.
681 The allowed options in a server block are
682 <literal>host</literal>, <literal>port</literal>,
683 <literal>IPv4Only</literal>, <literal>IPv6Only</literal>,
684 <literal>type</literal>, <literal>secret</literal>,
685 <literal>tls</literal>, <literal>certificateNameCheck</literal>,
686 <literal>matchCertificateAttribute</literal>,
687 <literal>AddTTL</literal>, <literal>rewrite</literal>,
688 <literal>rewriteIn</literal>, <literal>rewriteOut</literal>,
689 <literal>statusServer</literal>, <literal>retryCount</literal>,
690 <literal>dynamicLookupCommand</literal> and
691 <literal>retryInterval</literal> and
692 <literal>LoopPrevention</literal>.
696 We already discussed the <literal>host</literal> option. To
697 specify how radsecproxy should resolve a <literal>host</literal>
698 given as a DNS name, the <literal>IPv4Only</literal> or the
699 <literal>IPv6Only</literal> can be set to <literal>on</literal>.
700 At most one of these options can be enabled. Enabling
701 <literal>IPv4Only</literal> or <literal>IPv6Only</literal> here
702 overrides any basic settings set at the top level.
704 The <literal>port</literal> option allows you to specify which
705 port number the server uses. The usage of
706 <literal>type</literal>, <literal>secret</literal>,
707 <literal>tls</literal>, <literal>certificateNameCheck</literal>,
708 <literal>matchCertificateAttribute</literal>,
709 <literal>AddTTL</literal>, <literal>rewrite</literal>,
710 <literal>rewriteIn</literal> and <literal>rewriteOut</literal>
711 are just as specified for the <literal>client block</literal>
712 above, except that <literal>defaultServer</literal> (and not
713 <literal>defaultClient</literal>) is the fallback for the
714 <literal>tls</literal>, <literal>rewrite</literal> and
715 <literal>rewriteIn</literal> options.
718 <literal>statusServer</literal> can be specified to enable the
719 use of status-server messages for this server. The value must be
720 either <literal>on</literal> or <literal>off</literal>. The
721 default when not specified, is <literal>off</literal>. If
722 statusserver is enabled, the proxy will during idle periods send
723 regular status-server messages to the server to verify that it
724 is alive. This should only be enabled if the server supports it.
727 The options <literal>retryCount</literal> and
728 <literal>retryInterval</literal> can be used to specify how many
729 times the proxy should retry sending a request and how long it
730 should wait between each retry. The defaults are 2 retries and
734 The option <literal>dynamicLookupCommand</literal> can be used
735 to specify a command that should be executed to dynamically
736 configure a server. The executable file should be given with
737 full path and will be invoked with the name of the realm as its
738 first and only argument. It should either print a valid
739 <literal>server</literal> option on stdout and exit with a code
740 of 0 or print nothing and exit with a non-zero exit code. An
741 example of a shell script resolving the DNS NAPTR records for
742 the realm and then the SRV records for each NAPTR matching
743 'x-eduroam:radius.tls' is provided in
744 <literal>tools/naptr-eduroam.sh</literal>. This option was
745 added in radsecproxy-1.3 but tends to crash radsecproxy versions
749 Using the <literal>LoopPrevention</literal> option here
750 overrides any basic setting of this option. See section
751 <literal>BASIC OPTIONS</literal> for details on this option.
755 <title>Realm Block</title>
757 When the proxy receives an Access-Request it needs to figure out
758 to which server it should be forwarded. This is done by looking
759 at the Username attribute in the request, and matching that
760 against the names of the defined realm blocks. The proxy will
761 match against the blocks in the order they are specified, using
762 the first match if any. If no realm matches, the proxy will
763 simply ignore the request. Each realm block specifies what the
764 server should do when a match is found. A realm block may
765 contain none, one or multiple <literal>server</literal> options,
766 and similarly <literal>accountingServer</literal> options. There
767 are also <literal>replyMessage</literal> and
768 <literal>accountingResponse</literal> options. We will discuss
772 <title>Realm block names and matching</title>
774 In the general case the proxy will look for a
775 <literal>@</literal> in the username attribute, and try to do
776 an exact case insensitive match between what comes after the
777 <literal>@</literal> and the name of the realm block. So if
778 you get a request with the attribute value
779 <literal>anonymous@example.com</literal>, the proxy will go
780 through the realm names in the order they are specified,
781 looking for a realm block named
782 <literal>example.com</literal>.
785 There are two exceptions to this, one is the realm name
786 <literal>*</literal> which means match everything. Hence if
787 you have a realm block named <literal>*</literal>, then it
788 will always match. This should then be the last realm block
789 defined, since any blocks after this would never be
790 checked. This is useful for having a default.
793 The other exception is regular expression matching. If the
794 realm name starts with a <literal>/</literal>, the name is
795 treated as an regular expression. A case insensitive regexp
796 match will then be done using this regexp on the value of the
797 entire Username attribute. Optionally you may also have a
798 trailing <literal>/</literal> after the regexp. So as an
799 example, if you want to use regexp matching the domain
800 <literal>example.com</literal> you could have a realm block
801 named <literal>/@example\\.com$</literal>. Optinally this can
802 also be written <literal>/@example\\.com$/</literal>. If you
803 want to match all domains under the <literal>.com</literal>
804 top domain, you could do <literal>/@.*\\.com$</literal>. Note
805 that since the matching is done on the entire attribute value,
806 you can also use rules like
807 <literal>/^[a-k].*@example\\.com$/</literal> to get some of
808 the users in this domain to use one server, while other users
809 could be matched by another realm block and use another
814 <title>Realm block options</title>
816 A realm block may contain none, one or multiple
817 <literal>server</literal> options. If defined, the values of
818 the <literal>server</literal> options must be the names of
819 previously defined server blocks. Normally requests will be
820 forwarded to the first server option defined. If there are
821 multiple server options, the proxy will do fail-over and use
822 the second server if the first is down. If the two first are
823 down, it will try the third etc. If say the first server comes
824 back up, it will go back to using that one. Currently
825 detection of servers being up or down is based on the use of
826 StatusServer (if enabled), and that TCP/TLS/DTLS connections
830 A realm block may also contain none, one or multiple
831 <literal>accountingServer</literal> options. This is used
832 exactly like the <literal>server</literal> option, except that
833 it is used for specifying where to send matching accounting
834 requests. The values must be the names of previously defined
835 server blocks. When multiple accounting servers are defined,
836 there is a failover mechanism similar to the one for the
837 <literal>server</literal> option.
840 If there is no <literal>server</literal> option, the proxy
841 will if <literal>replyMessage</literal> is specified, reply
842 back to the client with an Access Reject message. The message
843 contains a replyMessage attribute with the value as specified
844 by the <literal>replyMessage</literal> option. Note that this
845 is different from having no match since then the request is
846 simply ignored. You may wonder why this is useful. One example
847 is if you handle say all domains under say
848 <literal>.bv</literal>. Then you may have several realm blocks
849 matching the domains that exists, while for other domains
850 under <literal>.bv</literal> you want to send a reject. At the
851 same time you might want to send all other requests to some
852 default server. After the realms for the subdomains, you would
853 then have two realm definitions. One with the name
854 <literal>/@.*\\.bv$</literal> with no servers, followed by one
855 with the name <literal>*</literal> with the default server
856 defined. This may also be useful for blocking particular
860 If there is no <literal>accountingServer</literal> option, the
861 proxy will normally do nothing, ignoring accounting
862 requests. There is however an option called
863 <literal>accountingResponse</literal>. If this is set to
864 <literal>on</literal>, the proxy will log some of the
865 accounting information and send an Accounting-Response
866 back. This is useful if you do not care much about accounting,
867 but want to stop clients from retransmitting accounting
868 requests. By default this option is set to
869 <literal>off</literal>.
874 <title>TLS Block</title>
876 The TLS block specifies TLS configuration options and you need
877 at least one of these if you have clients or servers using
878 TLS/DTLS. As discussed in the client and server block
879 descriptions, a client or server block may reference a
880 particular TLS block by name. There are also however the special
881 TLS block names <literal>default</literal>,
882 <literal>defaultClient</literal> and
883 <literal>defaultServer</literal> which are used as defaults if
884 the client or server block does not reference a TLS block. Also
885 note that a TLS block must be defined before the client or
886 server block that would use it. If you want the same TLS
887 configuration for all TLS/DTLS clients and servers, you need
888 just a single tls block named <literal>default</literal>, and
889 the client and servers need not refer to it. If you want all
890 TLS/DTLS clients to use one config, and all TLS/DTLS servers to
891 use another, then you would be fine only defining two TLS blocks
892 named <literal>defaultClient</literal> and
893 <literal>defaultServer</literal>. If you want different clients
894 (or different servers) to have different TLS parameters, then
895 you may need to create other TLS blocks with other names, and
896 reference those from the client or server definitions. Note that
897 you could also have say a client block refer to a default, even
898 <literal>defaultServer</literal> if you really want to.
901 The available TLS block options are
902 <literal>CACertificateFile</literal>,
903 <literal>CACertificatePath</literal>,
904 <literal>certificateFile</literal>,
905 <literal>certificateKeyFile</literal>,
906 <literal>certificateKeyPassword</literal>,
907 <literal>cacheExpiry</literal>, <literal>CRLCheck</literal> and
908 <literal>policyOID</literal>. When doing RADIUS over TLS/DTLS,
909 both the client and the server present certificates, and they
910 are both verified by the peer. Hence you must always specify
911 <literal>certificateFile</literal> and
912 <literal>certificateKeyFile</literal> options, as well as
913 <literal>certificateKeyPassword</literal> if a password is
914 needed to decrypt the private key. Note that
915 <literal>CACertificateFile</literal> may be a certificate
916 chain. In order to verify certificates, or send a chain of
917 certificates to a peer, you also always need to specify
918 <literal>CACertificateFile</literal> or
919 <literal>CACertificatePath</literal>. Note that you may specify
920 both, in which case the certificates in
921 <literal>CACertificateFile</literal> are checked first. By
922 default CRLs are not checked. This can be changed by setting
923 <literal>CRLCheck</literal> to <literal>on</literal>. One can
924 require peer certificates to adhere to certain policies by
925 specifying one or multiple policyOIDs using one or multiple
926 <literal>policyOID</literal> options.
929 CA certificates and CRLs are normally cached permanently. That
930 is, once a CA or CRL has been read, the proxy will never attempt
931 to re-read it. CRLs may change relatively often and the proxy
932 should ideally always use the latest CRLs. Rather than
933 restarting the proxy, there is an option
934 <literal>cacheExpiry</literal> that specifies how many seconds
935 the CA and CRL information should be cached. Reasonable values
936 might be say 3600 (1 hour) or 86400 (24 hours), depending on how
937 frequently CRLs are updated and how critical it is to be up to
938 date. This option may be set to zero to disable caching.
942 <title>Rewrite Block</title>
944 The rewrite block specifies rules that may rewrite RADIUS
945 messages. It can be used to add, remove and modify specific
946 attributes from messages received from and sent to clients and
947 servers. As discussed in the client and server block
948 descriptions, a client or server block may reference a
949 particular rewrite block by name. There are however also the
950 special rewrite block names <literal>default</literal>,
951 <literal>defaultClient</literal> and
952 <literal>defaultServer</literal> which are used as defaults if
953 the client or server block does not reference a block. Also note
954 that a rewrite block must be defined before the client or server
955 block that would use it. If you want the same rewrite rules for
956 input from all clients and servers, you need just a single
957 rewrite block named <literal>default</literal>, and the client
958 and servers need not refer to it. If you want all clients to use
959 one config, and all servers to use another, then you would be
960 fine only defining two rewrite blocks named
961 <literal>defaultClient</literal> and
962 <literal>defaultServer</literal>. Note that these defaults are
963 only used for rewrite on input. No rewriting is done on output
964 unless explicitly specified using the
965 <literal>rewriteOut</literal> option.
968 The available rewrite block options are
969 <literal>addAttribute</literal>,
970 <literal>addVendorAttribute</literal>,
971 <literal>removeAttribute</literal>,
972 <literal>removeVendorAttribute</literal> and
973 <literal>modifyAttribute</literal>. They can all be specified
974 none, one or multiple times.
977 <literal>addAttribute</literal> is used to add attributes to a
978 message. The option value must be on the form
979 <literal>attribute:value</literal> where attribute is a
980 numerical value specifying the attribute. Simliarly, the
981 <literal>addVendorAttribute</literal> is used to specify a
982 vendor attribute to be added. The option value must be on the
983 form <literal>vendor:subattribute:value</literal>, where vendor
984 and subattribute are numerical values.
987 The <literal>removeAttribute</literal> option is used to specify
988 an attribute that should be removed from received messages. The
989 option value must be a numerical value specifying which
990 attribute is to be removed. Similarly,
991 <literal>removeVendorAttribute</literal> is used to specify a
992 vendor attribute that is to be removed. The value can be a
993 numerical value for removing all attributes from a given vendor,
994 or on the form <literal>vendor:subattribute</literal>, where
995 vendor and subattribute are numerical values, for removing a
996 specific subattribute for a specific vendor.
999 <literal>modifyAttribute</literal> is used to specify
1000 modification of attributes. The value must be on the form
1001 <literal>attribute:/regexpmatch/replacement/</literal> where
1002 attribute is a numerical attribute type, regexpmatch is regexp
1003 matching rule and replacement specifies how to replace the
1004 matching regexp. Example usage:
1007 modifyAttribute 1:/^(.*)@local$/\1@example.com/
1013 <title>See Also</title>
1016 <refentrytitle>radsecproxy</refentrytitle><manvolnum>1</manvolnum>
1018 <ulink url="http://tools.ietf.org/html/draft-ietf-radext-radsec">
1019 <citetitle>RadSec internet draft</citetitle>