[ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
9.1 Rule Tracing | Tracing rules. | |
9.2 Debugging | Enabling full debugging information. | |
9.3 Test Mode | Running radius in test mode. |
[ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
If you have more than one entry in your ‘users’ file it is not always obvious which of the entries were used for authentication. The authentication data flow becomes even harder to understand if there are some complex rules in the ‘hints’ and ‘huntgroups’ files.
The rule tracing mode is intended to help you find out the exact
order of the rules that each request matched during processing.
The mode is toggled by trace-rules
statement in auth
or acct
block of your ‘config’ file. When rule tracing
mode is on for a given type of requests, radiusd
will
display the data flow diagram for each processed request of this
type. The diagram is output on info
logging category,
it represents the list of rules in reverse chronological order.
Each rule is represented by its location in the form
filename:line. To make the output more compact, if
several rules appear in the same configuration file, their locations
are listed as a comma-separated list of numbers after the file name.
Furthermore, if the configuration files have the same path prefix,
then only the first file name appears with the full prefix.
Here is an example of trace rule diagram:
|
This diagram means, that the authentication request from server ‘foo’ for user ‘bar’ with ID 170 matched the following rules
File name | Line number |
‘/etc/raddb/hints’ | 34 |
‘/etc/raddb/huntgroups’ | 72 |
‘/etc/raddb/users’ | 3 |
‘/etc/raddb/users’ | 22 |
‘/etc/raddb/users’ | 157 |
As a practical example, let's suppose you have the following setup. There are three classes of users:
In addition, users from the first two classes are accounted using
custom Scheme procedure staff-acct
.
The configuration files for this setup are showed below:
Contents of ‘hints’:
DEFAULT Group = "root" Scheme-Acct-Procedure = "staff-acct", Hint = "admin" DEFAULT Group = "staff" Scheme-Acct-Procedure = "staff-acct", Hint = "staff" |
Contents of file ‘users’:
DEFAULT Auth-Type = SQL, Simultaneous-Use = 1 Service-Type = Framed-User, Framed-Protocol = PPP DEFAULT Hint = "admin", Auth-Type = System Service-Type = Login-User, Login-IP-Host = 192.168.0.1, Login-Service = Rlogin DEFAULT Hint = "staff", Auth-Type = System, Simultaneous-Use = 1 Service-Type = Login-User, Login-IP-Host = 192.168.0.2, Login-Service = Telnet |
Now, let's suppose that user ‘svp’ is in the group
‘staff’ and is trying to log in. However, he fails to do so and
in radiusd
logs you see:
|
Why? To answer this question, you add to auth
block of your
‘config’ the statement
trace-rules yes; |
and ask user ‘svp’ to retry his attempt. Now you see in your logs:
|
This means that the request for ‘svp’ has first matched rule on the line 1 of file ‘hints’, then the rule on line 1 of file ‘users’. Now you see the error: the entries in ‘users’ appear in wrong order! After fixing it your ‘users’ looks like:
DEFAULT Hint = "admin", Auth-Type = System Service-Type = Login-User, Login-IP-Host = 192.168.0.1, Login-Service = Rlogin DEFAULT Hint = "staff", Auth-Type = System, Simultaneous-Use = 1 Service-Type = Login-User, Login-IP-Host = 192.168.0.2, Login-Service = Telnet DEFAULT Auth-Type = SQL, Simultaneous-Use = 1 Service-Type = Framed-User, Framed-Protocol = PPP |
Now, you ask ‘svp’ to log in again, and see:
|
Let's also suppose that user ‘plog’ is not listed in groups “root” and “staff”, so he is supposed to authenticate using SQL. When he logs in, you see in your logs:
|
[ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
GNU Radius provides extensive debugging features. These are enabled
either by the ‘--debug’ (‘-x’) command line option to
radiusd
(see section How to Start the Daemon.), or by the level
statement in the debug category (see section logging statement).
Both cases require as an argument a valid debug specification.
A debug specification sets the module for which the debugging should be enabled and the debugging level. The higher the level is, the more detailed information is provided. The module name and level are separated by an equal sign. If the level is omitted, the highest possible level (100) is assumed. The module name may be abbreviated to the first N characters, in which case the first matching module is selected. Several such specifications can be specified, in which case they should be separated by commas. For example, the following is a valid debug specification:
proxy.c=10,files.c,config.y=1 |
It sets debug level 10 for module proxy.c
, 100 for
files.c
, and 1 for config.y
.
The modules and debugging levels are subject to change from release to release.
[ < ] | [ > ] | [ << ] | [ Up ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
Test mode is used to test various aspects of radius configuration, without starting the daemon. To enter test mode, run
radiusd -mt |
You will see usual radiusd
diagnostics and the following two lines:
|
The string ‘** TEST SHELL **’ indicates that radiusd
has entered test mode, the string ‘(radiusd)’ is the shell
prompt, indicating that radiusd
is waiting for your
commands.
The syntax of test shell command resembles that of Bourne shell: each command consists of a list of words separated by any amount of whitespace. Each word is either a sequence of allowed word characters (i.e. alphabetical characters, decimal digits, dashes and underscores), or any sequence of characters enclosed in a pair of double quotes. The very first word is a command verb, the rest of words are arguments to this command verb. A command verb may be used in its full form, in its abbreviated form (i.e. you may type only several first characters of the verb, the only condition being that they do not coincide with another command verb), or in it's short form.
The first command you should know is help
(or, in its short
form, h
). This command takes no arguments and displays
the short summary of all the available commands. Here is an example
of its output:
|
Each line of the output consists of three fields. The first field shows the short command form. The second one lists its full form and its arguments, optional arguments being enclosed in square brackets. The third field contains short textual description of the command.
Queries the given NAS about the session described by its arguments. This command is useful in testing simultaneous login verification (see section Multiple Login Checking. Its arguments are
Specifies the NAS to query. It cn be its short name as defined in ‘raddb/naslist’, or its fully qualified domain name, or its IP address.
Name of the user whose session should be verified.
Session ID.
Port number on the NAS.
Framed IP address, assigned to the user.
The command displays the following result codes:
The session is not active.
The session is active
Some error occurred.
Enter Guile shell. The command is only available if the package has been compiled with Guile support. For more information, See section Guile.
Prints or sets the Rewrite stack size.
Runs given Rewrite function and displays its return value. Function arguments are specified in the usual way, i.e. as a comma-separated list of Rewrite tokens.
If the function being tested operates on request contents
(see section Rewriting Incoming Requests), you may supply the request
using request-define
command (see below).
Reads and compiles (“source”) the given Rewrite file. The command prints ‘0’ if the file was compiled successfully. Otherwise, it prints ‘1’ and any relevant diagnostics.
Checks whether the given time falls within the timespan
interval. Its first argument, timespan, contains the valid
radiusd
timespan specification (see section Login-Time
). Rest of
arguments define the time. If any of these is omitted, the
corresponding value from current local time is used.
Ordinal day of week number, counted from 0. I.e.: Sunday – 0, Monday – 1, etc.
Hours counted from 0 to 24.
Minutes.
The following set of samples illustrates this command:
|
Set debugging level. Level is any valid debug level specification (see section Debugging).
Define a request for testing Rewrite functions. The optional arguments are a comma-separated list of A/V pairs. If they are omitted, the command enters interactive mode, allowing you to enter the desired A/V pairs, as in the following example:
|
Notice that any number of A/V pairs may be specified in a line. To finish entering the request, either type an <EOF> character or enter an empty line.
Prints the request, defined by request-define
.
|
Immediately quits the shell.
[ << ] | [ >> ] | [Top] | [Contents] | [Index] | [ ? ] |
This document was generated by Sergey Poznyakoff on December, 6 2008 using texi2html 1.78.