added policyOID option in trunk docs, fixed typo in several docs

[libradsec.git] / radsecproxy.conf.5

'\" -*- coding: us-ascii -*-
.if \n(.g .ds T< \\FC
.if \n(.g .ds T> \\F[\n[.fam]]
.de URL
\\$2 \(la\\$1\(ra\\$3
..
.if \n(.g .mso www.tmac
.TH "radsecproxy.conf " 5 2008-10-06 "radsecproxy 1.2" ""
.SH NAME
radsecproxy.conf
\- Radsec proxy configuration file 
.SH DESCRIPTION
When the proxy server starts, it will first check the command line arguments,
and then read the configuration file. Normally radsecproxy will read the
configuration file \*(T<\fI/etc/radsecproxy.conf\fR\*(T>. The command
line \*(T<\fB\-c\fR\*(T> option can be used to instead read an alternate
file (see
\fBradsecproxy\fR(1)
for details).
.PP
If the configuration file can not be found, the proxy will exit with an
error message. Note that there is also an include facility so that any
configuration file may include other configuration files. The proxy will
also exit on configuration errors.
.SH "CONFIGURATION SYNTAX"
When the configuration file is processed, whitespace (spaces and tabs) are
generally ignored. For each line, leading and trailing whitespace are ignored.
A line is ignored if it is empty, only consists of whitespace, or if the first 
non-whitespace character is a \*(T<#\*(T>. The configuration is
generally case insensitive, but in some cases the option values (see below)
are not.
.PP
There are two types of configuration structures than can be used. The first
and simplest are lines of the format \fIoption value\fR. That
is, an option name, see below for a list of valid options, followed by
whitespace (at least one space or tab character), followed by a value. Note
that if the value contains whitespace, then it must be quoted using
\*(T<""\*(T> or \*(T<''\*(T>. Any whitespace
in front of the option or after the value will be ignored.
.PP
The other type of structure is a block. A block spans at least two lines, and
has the format:
.RS 
.nf

blocktype name {
    option value
    option value
    ...
}
.fi
.RE

That is, some blocktype, see below for a list of the different block types, and
then enclosed in braces you have zero or more lines that each have the
previously described \fIoption value\fR format. Different block
types have different rules for which options can be specified, they are listed
below. The rules regarding white space, comments and quotes are as above. Hence
you may do things like:
.RS 
.nf

blocktype name {
#    option value
    option "value with space"
    ...
}
.fi
.RE
.PP
Option value characters can also be written in hex. This is done by writing the
character \*(T<%\*(T> followed by two hexadecimal digits. If a
\*(T<%\*(T> is used without two following hexadecimal digits, the
\*(T<%\*(T> and the following characters are used as written. If you
want to write a \*(T<%\*(T> and not use this decoding, you may of
course write \*(T<%\*(T> in hex; i.e., \*(T<%25\*(T>.
.PP
There is one special option that can be used both as a basic option and inside
all blocks. That is the option \*(T<include\*(T> where the value
specifies files to be included. The value can be a single file, or it can use
normal shell globbing to specify multiple files, e.g.:
.RS 
include /etc/radsecproxy.conf.d/*.conf
.RE

The files are sorted alphabetically. Included files are read in the order they
are specified, when reaching the end of a file, the next file is read. When
reaching the end of the last included file, the proxy returns to read the next
line following the \*(T<include\*(T> option. Included files may again
include other files.
.SH "BASIC OPTIONS"
The following basic options may be specified in the configuration file. Note
that blocktypes and options inside blocks are discussed later. Note that none
of these options are required, and indeed in many cases they are not needed.
Note that you should specify each at most once. The behaviour with multiple
occurences is undefined.
.TP 
\*(T<logLevel\*(T>
This option specifies the debug level. It must be set to 1, 2, 3 or 4, where 1
logs only serious errors, and 4 logs everything. The default is 3 which logs
errors, warnings and some informational messages. Note that the command line
option \*(T<\fB\-d\fR\*(T> overrides this.
.TP 
\*(T<logDestination\*(T>
This specifies where the log messages should go. By default the messages go to
syslog with facility \*(T<LOG_DAEMON\*(T>. Using this option you can
specify another syslog facility, or you may specify that logging should be to
a particular file, not using syslog. The value must be either a file or
syslog URL. The file URL is the standard one, specifying a local file that
should be used. For syslog, you must use the syntax:
\*(T<x\-syslog:///FACILITY\*(T> where \*(T<FACILITY\*(T> must
be one of \*(T<LOG_DAEMON\*(T>, \*(T<LOG_MAIL\*(T>,
\*(T<LOG_USER\*(T>, \*(T<LOG_LOCAL0\*(T>,
\*(T<LOG_LOCAL1\*(T>, \*(T<LOG_LOCAL2\*(T>,
\*(T<LOG_LOCAL3\*(T>, \*(T<LOG_LOCAL4\*(T>,
\*(T<LOG_LOCAL5\*(T>, \*(T<LOG_LOCAL6\*(T> or
\*(T<LOG_LOCAL7\*(T>. You may omit the facility from the URL to
specify logging to the default facility, but this is not very useful since
this is the default log destination. Note that this option is ignored if
\*(T<\fB\-f\fR\*(T> is specified on the command line.
.TP 
\*(T<listenUDP\*(T>
Normally the proxy will listen to the standard RADIUS UDP port
\*(T<1812\*(T> if configured to handle UDP clients. On most systems it
will do this for all of the system's IP addresses (both IPv4 and IPv6). On some
systems however, it may respond to only IPv4 or only IPv6. To specify an
alternate port you may use a value of the form \*(T<*:port\*(T> where
port is any valid port number. If you also want to specify a specific address
you can do e.g. \*(T<192.168.1.1:1812\*(T> or
\*(T<[2001:db8::1]:1812\*(T>. The port may be omitted if you want the
default one (like in these examples). These examples are equivalent to
\*(T<192.168.1.1\*(T> and \*(T<2001:db8::1\*(T>. Note that
you must use brackets around the IPv6 address if you specify port number.
.TP 
\*(T<listenTLS\*(T>
This is similar to the \*(T<listenUDP\*(T> option, except that it is
used for receiving connections from TLS clients. The default port number is
\*(T<2083\*(T>. Note that this option was previously called
\*(T<listenTCP\*(T>.
.TP 
\*(T<listenTCP\*(T>
This option is deprecated. \*(T<listenTLS\*(T> should be used
instead. In future versions \*(T<listenTCP\*(T> will be used for
RADIUS over TCP (not TLS encrypted).
.TP 
\*(T<listenAccountingUDP\*(T>
This is similar to the \*(T<listenUDP\*(T> option, except that it is
used for specifying port and optionally the address to receive UDP Accounting
messages on. 
.TP 
\*(T<sourceUDP\*(T>
This can be used to specify source address and/or source port that the proxy
will use for sending UDP client messages (e.g. Access Request).
.TP 
\*(T<sourceTLS\*(T>
This can be used to specify source address and/or source port that the proxy
will use for TLS connections. Note that this option was previously called
\*(T<sourceTCP\*(T>.
.TP 
\*(T<sourceTCP\*(T>
This option is deprecated. \*(T<sourceTLS\*(T> should be used
instead. In future versions \*(T<sourceTCP\*(T> will be used for
RADIUS over TCP (not TLS encrypted).
.TP 
\*(T<loopPrevention\*(T>
This can be set to \*(T<on\*(T> or \*(T<off\*(T> with
\*(T<off\*(T> being the default. When this is enabled, a request
will never be sent to a server named the same as the client it was received
from. I.e., the names of the client block and the server block are compared.
Note that this only gives limited protection against loops.
.TP 
\*(T<include\*(T>
This is not a normal configuration option; it can be specified multiple times.
It can both be used as a basic option and inside blocks. For the full
description, see the configuration syntax section above.
.SH BLOCKS
There are five types of blocks, they are \*(T<client\*(T>,
\*(T<server\*(T>, \*(T<realm\*(T>, \*(T<tls\*(T>
and \*(T<rewrite\*(T>. At least one instance of each of
\*(T<client\*(T> and \*(T<realm\*(T> is required. This is
necessary for the proxy to do anything useful, and it will exit if not. The
\*(T<tls\*(T> block is required if at least one TLS client or server
is configured. Note that there can be multiple blocks for each type. For each
type, the block names should be unique. The behaviour with multiple
occurences of the same name for the same block type is undefined. Also note
that some block option values may reference a block by name, in which case
the block name must be previously defined. Hence the order of the blocks may
be significant.
.SH "CLIENT BLOCK"
The client block is used to configure a client. That is, tell the proxy about a
client, and what parameters should be used for that client. The name of the
client block must (with one exception, see below) be either the IP address
(IPv4 or IPv6) of the client, an IP prefix (IPv4 or IPv6) of the form
IpAddress/PrefixLength, or a domain name (FQDN).
.PP
If a domain name is specified, then this will be resolved immediately to all
the addresses associated with the name, and the proxy will not care about any
possible DNS changes that might occur later. Hence there is no dependency on
DNS after startup.
.PP
When some client later sends a request to the proxy, the proxy will look at the
IP address the request comes from, and then go through all the addresses of
each of the configured clients (in the order they are defined), to determine
which (if any) of the clients this is.
.PP
In the case of TLS, the name of the client must match the FQDN or IP address in
the client certificate. Note that this is not required when the client name is
an IP prefix.
.PP
Alternatively one may use the \*(T<host\*(T> option inside a client
block. In that case, the value of the \*(T<host\*(T> option is used as
above, while the name of the block is only used as a descriptive name for the
administrator.
.PP
The allowed options in a client block are \*(T<host\*(T>,
\*(T<type\*(T>, \*(T<secret\*(T>, \*(T<tls\*(T>,
\*(T<certificateNameCheck\*(T>,
\*(T<matchCertificateAttribute\*(T>, \*(T<rewrite\*(T>,
\*(T<rewriteIn\*(T>, \*(T<rewriteOut\*(T> and
\*(T<rewriteAttribute\*(T>. We already discussed the
\*(T<host\*(T> option. The value of \*(T<type\*(T> must be
either \*(T<udp\*(T> or \*(T<tls\*(T>. The value of
\*(T<secret\*(T> is the shared RADIUS key used with this client. If
the secret contains whitespace, the value must be quoted. This option is
optional for TLS.
.PP
For a TLS client you may also specify the \*(T<tls\*(T> option. The
option value must be the name of a previously defined TLS block. If this
option is not specified, the TLS block with the name
\*(T<defaultClient\*(T> will be used if defined. If not defined, it
will try to use the TLS block named \*(T<default\*(T>. If the
specified TLS block name does not exist, or the option is not specified and
none of the defaults exist, the proxy will exit with an error.
.PP
For a TLS client, the option \*(T<certificateNameCheck\*(T> can be set
to \*(T<off\*(T>, to disable the default behaviour of matching CN or
SubjectAltName against the specified hostname or IP address.
.PP
Additional validation of certificate attributes can be done by use of the
\*(T<matchCertificateAttribute\*(T> option. Currently one can only do
some matching of CN and SubjectAltName. For regexp matching on CN, one can use
the value \*(T<CN:/regexp/\*(T>. For SubjectAltName one can only do
regexp matching of the URI, this is specified as
\*(T<SubjectAltName:URI:/regexp/\*(T>. Note that currently this option
can only be specified once in a client block.
.PP
The \*(T<rewrite\*(T> option is deprecated. Use
\*(T<rewriteIn\*(T> instead.
.PP
The \*(T<rewriteIn\*(T> option can be used to refer to a rewrite block
that specifies certain rewrite operations that should be performed on incoming
messages from the client. The rewriting is done before other processing.
For details, see the rewrite block text below. Similarly to
\*(T<tls\*(T> discussed above, if this option is not used, there is a
fallback to using the \*(T<rewrite\*(T> block named
\*(T<defaultClient\*(T> if it exists; and if not, a fallback to a
block named \*(T<default\*(T>.
.PP
The \*(T<rewriteOut\*(T> option is used in the same way as
\*(T<rewriteIn\*(T>, except that it specifies rewrite operations that
should be performed on outgoing messages to the client. The rewriting is done
after other processing. Also, there is no rewrite fallback if this option is
not used.
.PP
The \*(T<rewriteAttribute\*(T> option currently makes it possible to
specify that the User-Name attribute in a client request shall be rewritten in
the request sent by the proxy. The User-Name attribute is written back to the
original value if a matching response is later sent back to the client. The
value must be of the form User-Name:/regexpmatch/replacement/. Example usage:
.RS 
rewriteAttribute User-Name:/^(.*)@local$/$1@example.com/
.RE
.SH "SERVER BLOCK"
The server block is used to configure a server. That is, tell the proxy about a
server, and what parameters should be used when communicating with that server.
The name of the server block must (with one exception, see below) be either the
IP address (IPv4 or IPv6) of the server, or a domain name (FQDN). If a domain
name is specified, then this will be resolved immediately to all the addresses
associated with the name, and the proxy will not care about any possible DNS
changes that might occur later. Hence there is no dependency on DNS after
startup. If the domain name resolves to multiple addresses, then for UDP the
first address is used. For TLS, the proxy will loop through the addresses until
it can connect to one of them. In the case of TLS, the name of the server must
match the FQDN or IP address in the server certificate.
.PP
Alternatively one may use the \*(T<host\*(T> option inside a server
block. In that case, the value of the \*(T<host\*(T> option is used as
above, while the name of the block is only used as a descriptive name for the
administrator.
.PP
The allowed options in a server block are \*(T<host\*(T>,
\*(T<port\*(T>, \*(T<type\*(T>, \*(T<secret\*(T>,
\*(T<tls\*(T>, \*(T<certificateNameCheck\*(T>,
\*(T<matchCertificateAttribute\*(T>, \*(T<rewrite\*(T>,
\*(T<rewriteIn\*(T>, \*(T<rewriteOut\*(T>,
\*(T<statusServer\*(T>, \*(T<retryCount\*(T> and
\*(T<retryInterval\*(T>.
.PP
We already discussed the \*(T<host\*(T> option. The
\*(T<port\*(T> option allows you to specify which port number the
server uses. The usage of \*(T<type\*(T>, \*(T<secret\*(T>,
\*(T<tls\*(T>, \*(T<certificateNameCheck\*(T>,
\*(T<matchCertificateAttribute\*(T>, \*(T<rewrite\*(T>,
\*(T<rewriteIn\*(T> and \*(T<rewriteOut\*(T> are just as
specified for the \*(T<client block\*(T> above, except that
\*(T<defaultServer\*(T> (and not \*(T<defaultClient\*(T>)
is the fallback for the \*(T<tls\*(T>, \*(T<rewrite\*(T>
and \*(T<rewriteIn\*(T> options.
.PP
\*(T<statusServer\*(T> can be specified to enable the use of
status-server messages for this server. The value must be either
\*(T<on\*(T> or \*(T<off\*(T>. The default when not
specified, is \*(T<off\*(T>. If statusserver is enabled, the proxy
will during idle periods send regular status-server messages to the server
to verify that it is alive. This should only be enabled if the server
supports it.
.PP
The options \*(T<retryCount\*(T> and
\*(T<retryInterval\*(T> can be used to specify how many times the
proxy should retry sending a request and how long it should wait between each
retry. The defaults are 2 retries and an interval of 5s.
.SH "REALM BLOCK"
When the proxy receives an Access-Request it needs to figure out to which
server it should be forwarded. This is done by looking at the Username attribute
in the request, and matching that against the names of the defined realm blocks.
The proxy will match against the blocks in the order they are specified, using
the first match if any. If no realm matches, the proxy will simply ignore the
request. Each realm block specifies what the server should do when a match is
found. A realm block may contain none, one or multiple \*(T<server\*(T>
options, and similarly \*(T<accountingServer\*(T> options. There are
also \*(T<replyMessage\*(T> and \*(T<accountingResponse\*(T>
options. We will discuss these later.
.SS "REALM BLOCK NAMES AND MATCHING"
In the general case the proxy will look for a \*(T<@\*(T> in the
username attribute, and try to do an exact case insensitive match between what
comes after the \*(T<@\*(T> and the name of the realm block. So if you
get a request with the attribute value \*(T<anonymous@example.com\*(T>,
the proxy will go through the realm names in the order they are specified,
looking for a realm block named \*(T<example.com\*(T>.
.PP
There are two exceptions to this, one is the realm name \*(T<*\*(T>
which means match everything. Hence if you have a realm block named
\*(T<*\*(T>, then it will always match. This should then be the last
realm block defined, since any blocks after this would never be checked. This
is useful for having a default.
.PP
The other exception is regular expression matching. If the realm name starts
with a \*(T</\*(T>, the name is treated as an regular expression. A
case insensitive regexp match will then be done using this regexp on the value
of the entire Username attribute. Optionally you may also have a trailing
\*(T</\*(T> after the regexp. So as an example, if you want to use
regexp matching the domain \*(T<example.com\*(T> you could have a
realm block named \*(T</@example\e\e.com$\*(T>. Optinally this can also
be written \*(T</@example\e\e.com$/\*(T>. If you want to match all
domains under the \*(T<.com\*(T> top domain, you could do
\*(T</@.*\e\e.com$\*(T>. Note that since the matching is done on the
entire attribute value, you can also use rules like
\*(T</^[a\-k].*@example\e\e.com$/\*(T> to get some of the users in this
domain to use one server, while other users could be matched by another realm
block and use another server.
.SS "REALM BLOCK OPTIONS"
A realm block may contain none, one or multiple \*(T<server\*(T>
options. If defined, the values of the \*(T<server\*(T> options must
be the names of previously defined server blocks. Normally requests will be
forwarded to the first server option defined. If there are multiple server
options, the proxy will do fail-over and use the second server if the first
is down. If the two first are down, it will try the third etc. If say the
first server comes back up, it will go back to using that one. Currently
detection of servers being up or down is based on the use of StatusServer (if
enabled), and that TLS connections are up.
.PP
A realm block may also contain none, one or multiple
\*(T<accountingServer\*(T> options. This is used exactly like the
\*(T<server\*(T> option, except that it is used for specifying where
to send matching accounting requests. The values must be the names of
previously defined server blocks. When multiple accounting servers are
defined, there is a failover mechanism similar to the one for the
\*(T<server\*(T> option.
.PP
If there is no \*(T<server\*(T> option, the proxy will if
\*(T<replyMessage\*(T> is specified, reply back to the client with
an Access Reject message. The message contains a replyMessage attribute with
the value as specified by the \*(T<replyMessage\*(T> option. Note
that this is different from having no match since then the request is simply
ignored. You may wonder why this is useful. One example is if you handle say
all domains under say \*(T<.bv\*(T>. Then you may have several realm
blocks matching the domains that exists, while for other domains under
\*(T<\&.bv\*(T> you want to send a reject. At the same time you might
want to send all other requests to some default server. After the realms for
the subdomains, you would then have two realm definitions. One with the name
\*(T</@.*\e\e.bv$\*(T> with no servers, followed by one with the name
\*(T<*\*(T> with the default server defined. This may also be useful
for blocking particular usernames.
.PP
If there is no \*(T<accountingServer\*(T> option, the proxy will
normally do nothing, ignoring accounting requests. There is however an option
called \*(T<accountingResponse\*(T>. If this is set to
\*(T<on\*(T>, the proxy will log some of the accounting information
and send an Accounting-Response back. This is useful if you do not care much
about accounting, but want to stop clients from retransmitting accounting
requests. By default this option is set to \*(T<off\*(T>.
.SH "TLS BLOCK"
The TLS block specifies TLS configuration options and you need at least one
of these if you have clients or servers using TLS. As discussed in the client
and server block descriptions, a client or server block may reference a
particular TLS block by name. There are also however the special TLS block
names \*(T<default\*(T>, \*(T<defaultClient\*(T> and
\*(T<defaultServer\*(T> which are used as defaults if the client or
server block does not reference a TLS block. Also note that a TLS block must
be defined before the client or server block that would use it. If you want
the same TLS configuration for all TLS clients and servers, you need just a
single tls block named \*(T<default\*(T>, and the client and servers
need not refer to it. If you want all TLS clients to use one config, and all
TLS servers to use another, then you would be fine only defining two TLS
blocks named \*(T<defaultClient\*(T> and
\*(T<defaultServer\*(T>. If you want different clients (or different
servers) to have different TLS parameters, then you may need to create other
TLS blocks with other names, and reference those from the client or server
definitions. Note that you could also have say a client block refer to a
default, even \*(T<defaultServer\*(T> if you really want to.
.PP
The available TLS block options are \*(T<CACertificateFile\*(T>,
\*(T<CACertificatePath\*(T>, \*(T<certificateFile\*(T>,
\*(T<certificateKeyFile\*(T>,
\*(T<certificateKeyPassword\*(T>, \*(T<cacheExpiry\*(T>
and \*(T<CRLCheck\*(T>. When doing RADIUS over TLS, both the client
and the server present certificates, and they are both verified by the peer.
Hence you must always specify \*(T<certificateFile\*(T> and
\*(T<certificateKeyFile\*(T> options, as well as
\*(T<certificateKeyPassword\*(T> if a password is needed to decrypt
the private key. Note that \*(T<CACertificateFile\*(T> may be a
certificate chain. In order to verify certificates, or send a chain of
certificates to a peer, you also always need to specify
\*(T<CACertificateFile\*(T> or \*(T<CACertificatePath\*(T>.
Note that you may specify both, in which case the certificates in
\*(T<CACertificateFile\*(T> are checked first. By default CRLs are
not checked. This can be changed by setting \*(T<CRLCheck\*(T> to
\*(T<on\*(T>.
.PP
CA certificates and CRLs are normally cached permanently. That is, once a CA
or CRL has been read, the proxy will never attempt to re-read it. CRLs may
change relatively often and the proxy should ideally always use the latest
CRLs. Rather than restarting the proxy, there is an option
\*(T<cacheExpiry\*(T> that specifies how many seconds the CA and
CRL information should be cached. Reasonable values might be say 3600
(1 hour) or 86400 (24 hours), depending on how frequently CRLs are updated
and how critical it is to be up to date. This option may be set to zero to
disable caching.
.SH "REWRITE BLOCK"
The rewrite block specifies rules that may rewrite RADIUS messages. It can be
used to add, remove and modify specific attributes from messages received
from and sent to clients and servers. As discussed in the client and server
block descriptions, a client or server block may reference a particular
rewrite block by name. There are however also the special rewrite block names
\*(T<default\*(T>, \*(T<defaultClient\*(T> and
\*(T<defaultServer\*(T> which are used as defaults if the client or
server block does not reference a block. Also note that a rewrite block must
be defined before the client or server block that would use it. If you want
the same rewrite rules for input from all clients and servers, you need just
a single rewrite block named \*(T<default\*(T>, and the client and
servers need not refer to it. If you want all clients to use one config, and
all servers to use another, then you would be fine only defining two rewrite
blocks named \*(T<defaultClient\*(T> and
\*(T<defaultServer\*(T>. Note that these defaults are only used for
rewrite on input. No rewriting is done on output unless explicitly specifed
using the \*(T<rewriteOut\*(T> option.
.PP
The available rewrite block options are \*(T<addAttribute\*(T>,
\*(T<removeAttribute\*(T>, \*(T<removeVendorAttribute\*(T>
and \*(T<modifyAttribute\*(T>. They can all be specified none, one
or multiple times.
.PP
\*(T<addAttribute\*(T> is used to add attributes to a message. The
option value must be of the form \*(T<attribute:value\*(T> where
attribute is a numerical value specifying the attribute.
.PP
The \*(T<removeAttribute\*(T> option is used to specify an
attribute that should be removed from received messages. The option value
must be a numerical value specifying which attribute is to be removed.
Similarly, \*(T<removeVendorAttribute\*(T> is used to specify a
vendor attribute that is to be removed. The value can be a numerical value
for removing all attributes from a given vendor, or of the form
\*(T<vendor:subattribute\*(T>, where vendor and subattribute are
numerical values, for removing a specific subattribute for a specific
vendor.
.PP
\*(T<modifyAttribute\*(T> is used to specify modification of
attributes. The value must be of the form
\*(T<attribute:/regexpmatch/replacement/\*(T> where attribute is
a numerical attribute type, regexpmatch is regexp matching rule and
replacement specifies how to replace the matching regexp. Example usage:
.RS 
modifyAttribute 1:/^(.*)@local$/$1@example.com/
.RE
.SH "SEE ALSO"
\fBradsecproxy\fR(1),
.URL http://tools.ietf.org/html/draft-ietf-radext-radsec "         RadSec internet draft       "