Building from Source
We recommend installing from packages if possible. Full instructions on building and installing from source code follow.
The mandatory dependencies
must be installed before FreeRADIUS can be built. These dependencies
are libtalloc
and libkqueue
, which FreeRADIUS uses for memory
management, and platform-independent event handling.
Per-module dependencies that enable support for external services
such as LDAP, SQL, etc, are optional. They must be installed for
any modules that are to be used. The FreeRADIUS ./configure
step
will automatically detect if each module has its dependencies met
and automatically enable support for them. If the features you
require are not enabled you should inspect the configure script
output to figure out which additional development libraries need
to be installed.
The FreeRADIUS source may be obtained from a number of locations:
-
Download the latest version of the FreeRADIUS source from the FreeRADIUS web site; or
-
download directly from the FreeRADIUS FTP site; or
-
download from GitHub.
The file wil be name something like: freeradius-server-4.0.0.tar.gz
.
Later version will be 4.0.1
, or 4.1.0
, etc. PGP signatures are
also provided for official releases from the FTP site; these are
named e.g. freeradius-server-4.0.0.tar.gz.sig
.
Un-tar the file, and change to the FreeRADIUS directory (where
VERSION
below is the version of the server that you have
downloaded).
tar -zxf freeradius-server-VERSION.tar.gz
cd freeradius-server-VERSION
Take the following steps to build and install the server from source:
./configure
make
sudo make install
Custom build
FreeRADIUS has GNU autoconf support. This means you have to run
./configure
, and then run make
. To see which configuration
options are supported, run ./configure --help
, and read its output.
The make install
stage will install the binaries, the "man" pages,
and may install the configuration files. If you have not installed a
RADIUS server before, then the configuration files for FreeRADIUS will
be installed.
If you already have a RADIUS server installed, then FreeRADIUS WILL NOT over-write your current configuration.
The make install
process will warn you about the files it could not
install.
If you see a warning message about files that could not be installed, then you must ensure that the new server is using the new configuration files and not the old configuration files, as this may cause undesired behavior and failure to operate correctly.
The initial output from running in debugging mode (radiusd -X
)
will tell you which configuration files are being used. See
Upgrading for information about
upgrading from older versions. There may be changes in the
dictionary files which are required for a new version of the
software. These files will not be installed over your current
configuration, so you must verify and install any problem files by
hand, for example using diff(1)
to check for changes.
When installing from source, it is extremely helpful to read the
output of ./configure
, make
, and make install
. If a
particular module you expected to be installed was not installed,
then the output will tell you why that module was not installed.
The most likely reason is that required libraries (including their
development header files) are not available.
Please do not post questions to the FreeRADIUS users list without first carefully reading the output of this process as it often contains the information needed to resolve a problem.
Upgrading To A New Minor Release
The installation process will not over-write your existing configuration files. It will, however, warn you about the files it did not install. These will require manual integration with the existing files.
It is generally not possible to re-use configurations between different major versions of the server. (For example - version 2 to version 3, or version 3 to version 4.
For details on what has changed between the version, see the upgrade guide.
We strongly recommend that new major versions be installed in a different location than any existing installations. Any local policies can then be migrated gradually to the configuration format of the new major version. The number of differences in the new configuration mean that is is both simpler and safer to migrate your configurations rather than to try and just get the old configuration to work.
Running the server
If the server builds and installs, but doesn’t run correctly, then
you should first use debugging mode (radiusd -X
) to figure out
the problem.
This is your best hope for understanding the problem. Read all of the messages which are printed to the screen, the answer to your problem will often be in a warning or error message.
We really cannot emphasize that last sentence enough. Configuring a RADIUS server for complex local authentication isn’t a trivial task. Your best and only method for debugging it is to read the debug messages, where the server will tell you exactly what it’s doing, and why. You should then compare its behaviour to what you intended, and edit the configuration files as appropriate.
If you don’t use debugging mode, and ask questions on the mailing
list, then the responses will all tell you to use debugging mode.
The server prints out a lot of information in this mode, including
suggestions for fixes to common problems. Look especially for
WARNING
and ERROR
messages in the output, and read the related
messages.
Since the main developers of FreeRADIUS use debugging mode to track down their configuration problems with the server, it’s a good idea for you to use it, too. If you don’t, there is little hope for you to solve any configuration problem related to the server.
To start the server in debugging mode, do:
radiusd -X
You should see a lot of text printed on the screen as it starts up. If you don’t, or if you see error messages, please read the FAQ:
If the server says Ready to process requests.
, then it is running
properly. From another shell (or another window), type
radtest test test localhost 0 testing123
You should see the server print out more messages as it receives the
request, and responds to it. The radtest
program should receive the
response within a few seconds. It doesn’t matter if the authentication
request is accepted or rejected, what matters is that the server
received the request, and responded to it.
You can now edit the configuration files for your local system. You will
usually want to start with sites-enabled/default
for main
configurations. To set which NASes (clients) can communicate with this
server, edit raddb/clients.conf
. Please read the configuration files
carefully, as many configuration options are only documented in comments
in the file.
Note that is is highly recommended that you use some sort of version control system to manage your configuration, such as git or Subversion. You should then make small changes to the configuration, checking in and testing as you go. When a config change causes the server to stop working, you will be able to easily step back and find out what update broke the configuration.
It is also considered a best practice to maintain a staging or development environment. This allows you to test and integrate your changes without impacting your active production environment. You should make the appropirate investment in order to properly support a critical resource such as your authentication servers.
Configuring and running the server MAY be complicated. Many modules have
man
pages. See man rlm_pap
, or man rlm_*
for information.
Please read the documentation in the doc/ directory. The comments in the
configuration files also contain a lot of documentation.
If you have any additional issues, the FAQ is also a good place to start.