cfgutils Module

Henning Westerholt

1und1 Internet AG

Carsten Bock

BASIS AudioNet GmbH

Elena-Ramona Modroiu

rosdev.ro

Jiri Kuthan

Olle E. Johansson

Juha Heinanen

Edited by

Daniel-Constantin Mierla


Table of Contents

1. Admin Guide
1. Overview
2. Dependencies
3. Parameters
3.1. initial_probability (string)
3.2. hash_file (string)
3.3. initial_gflags (integer)
3.4. lock_set_size (integer)
4. Functions
4.1. rand_event()
4.2. rand_set_prob(probability)
4.3. rand_reset_prob()
4.4. rand_get_prob()
4.5. sleep(time)
4.6. usleep(time)
4.7. abort()
4.8. pkg_status()
4.9. pkg_summary()
4.10. shm_status()
4.11. shm_summary()
4.12. set_gflag(flag)
4.13. reset_gflag(flag)
4.14. is_gflag(flag)
4.15. lock(key [, key2])
4.16. trylock(key [, key2])
4.17. unlock(key [, key2])
4.18. check_route_exists(route)
4.19. route_if_exists(route)
4.20. core_hash(string1, string2, size)
5. RPC Commands
5.1. cfgutils.rand_set_prop
5.2. cfgutils.rand_reset_prob
5.3. cfgutils.rand_get_prob
5.4. cfgutils.check_config_hash
5.5. cfgutils.get_config_hash
5.6. cfgutils.set_gflag
5.7. cfgutils.reset_gflag
5.8. cfgutils.is_gflag
5.9. cfgutils.get_gflags
6. Exported pseudo-variables
6.1. $RANDOM

List of Examples

1.1. initial_probability parameter usage
1.2. hash_file parameter usage
1.3. initial parameter usage
1.4. lock_set_size parameter usage
1.5. rand_event() usage
1.6. rand_set_prob() usage
1.7. rand_reset_prob() usage
1.8. rand_get_prob() usage
1.9. sleep usage
1.10. usleep usage
1.11. abort usage
1.12. pkg_status usage
1.13. pkg_summary usage
1.14. shm_status usage
1.15. shm_summary usage
1.16. set_gflag() usage
1.17. reset_gflag() usage
1.18. is_gflag() usage
1.19. lock() usage
1.20. trylock() usage
1.21. unlock() usage
1.22. check_route_exists() usage
1.23. route_if_exists() usage
1.24. core_hash() usage
1.25. cfgutils.rand_set_prob usage
1.26. cfgutils.rand_reset_prob usage
1.27. cfgutils.rand_get_prob usage
1.28. cfgutils.check_config_hash usage
1.29. cfgutils.get_config_hash usage
1.30. cfgutils.set_gflag usage
1.31. cfgutils.reset_gflag usage
1.32. cfgutils.is_gflag usage
1.33. cfgutils.get_gflags usage
1.34. RANDOM pseudo-variable usage

Chapter 1. Admin Guide

1. Overview

Useful extensions for the server configuration.

The cfgutils module can be used to introduce randomness to the behaviour of the server. It provides setup functions and the rand_event function. This function return either true or false, depending on a random value and a specified probability. E.g. if you set via fifo or script a probability value of 5%, then 5% of all calls to rand_event will return true. The pseudovariable $RANDOM could be used to introduce random values e.g. into a SIP reply.

The benefit of this module is the probability of the decision can be manipulated by external applications such as web interface or command line tools. The probability must be specified as percent value, ranging from 0 to 100.

The module exports commands to FIFO server that can be used to change the global settings via FIFO interface. The FIFO commands are: set_prob, reset_prob and get_prob.

This module can be used for simple load-shedding, e.g. reply 5% of the Invites with a 503 error and an adequate random Retry-After value.

The module provides as well functions to delay the execution of the server. The functions sleep and usleep could be used to let the server wait a specific time interval.

It can also hash the config file used from the server with a (weak) cryptographic hash function on startup. This value is saved and can be later compared to the actual hash, to detect modifications of this file after the server start. This functions are available as the FIFO commands check_config_hash and get_config_hash.

The gflags functionality (global flags) keeps a bitmap of flags in shared memory and may be used to change behaviour of server based on value of the flags. Example:

	if (is_gflag("1")) {
		t_relay("udp:10.0.0.1:5060");
	} else {
		t_relay("udp:10.0.0.2:5060");
	}

The benefit of this is the value of the switch flags can be manipulated by external applications such as web interface or command line tools. The size of bitmap is 32.

The module exports external commands that can be used to change the global flags via Management Interface and RPC. See relevant sections below.

2. Dependencies

The module depends on the following modules (in other words the listed modules must be loaded before this module):

  • none

3. Parameters

3.1. initial_probability (string)

The initial value of the probability.

Default value is 10.

Example 1.1. initial_probability parameter usage

...
modparam("cfgutils", "initial_probability", 15)
...

3.2. hash_file (string)

The config file name for that a hash value should be calculated on startup.

There is no default value, is no parameter is given the hash functionality is disabled.

Example 1.2. hash_file parameter usage

...
modparam("cfgutils", "hash_file", "/etc/kamailio/kamailio.cfg")
...

3.3. initial_gflags (integer)

The initial value of global flags bitmap.

Default value is 0.

Example 1.3. initial parameter usage

...
modparam("cfgutils", "initial_gflags", 15)
...

3.4. lock_set_size (integer)

Size of lock set - the value is used as power of two to compute the size of lock array.

Default value is 0 - no lock set created.

Example 1.4. lock_set_size parameter usage

...
modparam("cfgutils", "lock_set_size", 4)
...

4. Functions

4.1. rand_event()

Return true or false, depending on a random value and a probability value.

Example 1.5. rand_event() usage

...
if (rand_event()) {
  append_to_reply("Retry-After: 120\n");
  sl_send_reply("503", "Try later");
  exit;
};
# normal message processing follows
...

4.2. rand_set_prob(probability)

Set the probability of the decision.

probability can have a value from the range 0..100.

Example 1.6. rand_set_prob() usage

...
rand_set_prob("4");
...

4.3. rand_reset_prob()

Reset the probability back to the initial value.

Example 1.7. rand_reset_prob() usage

...
rand_reset_prob();
...

4.4. rand_get_prob()

Return the current probability setting, e.g. for logging purposes.

Example 1.8. rand_get_prob() usage

...
rand_get_prob();
...

4.5.  sleep(time)

Waits "time" seconds.

Meaning of the parameters is as follows:

  • time - Time to wait in seconds. It can be an integer or a variable holding an integer.

This function can be used from ANY_ROUTE.

Example 1.9. sleep usage

...
sleep("1");
...

4.6.  usleep(time)

Waits "time" micro-seconds.

Meaning of the parameters is as follows:

  • time - Time to wait in micro-seconds (1/1000000 of a second). It can be an integer or a variable holding an integer.

This function can be used from ANY_ROUTE.

Example 1.10. usleep usage

...
# wait 5 milliseconds
usleep("5000");
...

4.7.  abort()

Debugging function that aborts the server. Depending on the configuration of the server a core dump will be created.

This function can be used from REQUEST_ROUTE, ONREPLY_ROUTE, FAILURE_ROUTE, BRANCH_ROUTE.

Example 1.11. abort usage

...
abort();
...

4.8.  pkg_status()

Debugging function that dumps the status for the private (PKG) memory. This information is logged to the default log facility, depending on the general log level and the memlog setting. You need to compile the server with activated memory debugging to get detailed informations.

This function can be used from REQUEST_ROUTE, ONREPLY_ROUTE, FAILURE_ROUTE, BRANCH_ROUTE.

Example 1.12. pkg_status usage

...
pkg_status();
...

4.9.  pkg_summary()

Debugging function that dumps the summary for the private (PKG) memory usage. This information is logged to the default log facility, depending on the general log level and the memlog setting. You need to compile the server with activated memory debugging to get detailed informations.

This function can be used from ANY_ROUTE.

Example 1.13. pkg_summary usage

...
pkg_summary();
...

4.10.  shm_status()

Debugging function that dumps the status for the shared (SHM) memory. This information is logged to the default log facility, depending on the general log level and the memlog setting. You need to compile the server with activated memory debugging to get detailed informations.

This function can be used from REQUEST_ROUTE, ONREPLY_ROUTE, FAILURE_ROUTE, BRANCH_ROUTE.

Example 1.14. shm_status usage

...
shm_status();
...

4.11.  shm_summary()

Debugging function that dumps the summary for the shared (SHM) memory usage. This information is logged to the default log facility, depending on the general log level and the memlog setting. You need to compile the server with activated memory debugging to get detailed informations.

This function can be used from ANY_ROUTE.

Example 1.15. shm_summary usage

...
shm_summary();
...

4.12. set_gflag(flag)

Set the bit at the position flag in global flags.

flag can have a value in the range of 0..31.

This function can be used from REQUEST_ROUTE, FAILURE_ROUTE, ONREPLY_ROUTE, BRANCH_ROUTE and LOCAL_ROUTE.

Example 1.16. set_gflag() usage

...
set_gflag("4");
...

4.13. reset_gflag(flag)

Reset the bit at the position flag in global flags.

flag can have a value in the range of 0..31.

This function can be used from REQUEST_ROUTE, FAILURE_ROUTE, ONREPLY_ROUTE, BRANCH_ROUTE and LOCAL_ROUTE.

Example 1.17. reset_gflag() usage

...
reset_gflag("4");
...

4.14. is_gflag(flag)

Check if bit at the position flag in global flags is set.

flag can have a value in the range of 0..31.

This function can be used from REQUEST_ROUTE, FAILURE_ROUTE, ONREPLY_ROUTE, BRANCH_ROUTE and LOCAL_ROUTE.

Example 1.18. is_gflag() usage

...
if(is_gflag("4"))
{
	log("global flag 4 is set\n");
} else {
	log("global flag 4 is not set\n");
};
...

4.15. lock(key [, key2])

Lock the key. Can be used to synchronize operations in config file, a hash id is computed over the keys and appropriate lock is set in the lock array controlled by parameter "lock_set_size". Do not use lock() after another lock() unless you are sure the keys hit different array entries.

key can be static string or string with PVs.

key2 is optional and can be static string or string with PVs.

This function can be used from REQUEST_ROUTE, FAILURE_ROUTE, ONREPLY_ROUTE, BRANCH_ROUTE and LOCAL_ROUTE.

Example 1.19. lock() usage

...
lock("$rU");
...

4.16. trylock(key [, key2])

Try to lock the key. If the lock can not be obtained (possibly already locked), the function returns an error and script execution continues.

key can be static string or string with PVs.

key2 is optional and can be static string or string with PVs.

This function can be used from REQUEST_ROUTE, FAILURE_ROUTE, ONREPLY_ROUTE, BRANCH_ROUTE and LOCAL_ROUTE.

Example 1.20. trylock() usage

...
if (trylock("$rU")) {
	xlog("L_INFO", "Doing some cool stuff\n");
	unlock("$rU");
}
...

4.17. unlock(key [, key2])

Unlock the key.

key can be static string or string with PVs.

key2 is optional and can be static string or string with PVs.

This function can be used from REQUEST_ROUTE, FAILURE_ROUTE, ONREPLY_ROUTE, BRANCH_ROUTE and LOCAL_ROUTE.

Example 1.21. unlock() usage

...
unlock("$rU");
...

4.18. check_route_exists(route)

Check if a route block exists. It returns true (1) on found and false on not found or error.

Parameters:

name of a route block in the config file, like route[FROGJUMP]

This function can be used from any route. You can only check for route[] blocks, not reply, event or other routes.

Example 1.22. check_route_exists() usage

...
if(check_route_exists("JUMP") {
	$var(jumping_frogs) = 1;
};
...

4.19. route_if_exists(route)

Execute a routing block only if it is defined. If it's not defined, silently move to the next action in the configuration script.

It returns the code of last action in the route block, if that exists, or false if the route doesn't exists or was an error executing it.

Parameters:

name of a route block in the config file, like route[FROGJUMP]

This function can be used from any route. You can only execute it for route[] blocks, not reply, event or other routes.

Example 1.23. route_if_exists() usage

...
if(route_if_exists("JUMP")) {
    exit;
}
...

4.20. core_hash(string1, string2, size)

Exported function that enables the core_hash() function to be used from the configuration file.

This is a quick and simple hash function and it is not cryptographically secure. This function should not be used for any security related purposes.

Parameters:

  • string1 first string to hash

  • string2 (optional) second string to hash (set to "" if not needed)

  • size size of the hash space (used as a power of 2)

This function can be used from ANY_ROUTE.

Example 1.24. core_hash() usage

...
core_hash("$ci", "", 4);
...

5. RPC Commands

Functions that check or change some global flags accepts one parameter which is the flag bitmap/mask specifying the corresponding flags. It is not possible to specify directly the flag position that should be changed as in the functions available in the routing script.

5.1. cfgutils.rand_set_prop

Set the probability value to the given parameter. The parameter should be a percent value.

The parameter value must be a number from 0 to 100.

Example 1.25. cfgutils.rand_set_prob usage

...
$ kamcmd cfgutils.rand_set_prob 10
...

5.2. cfgutils.rand_reset_prob

Reset the probability value to the initial start value.

This command don't need a parameter.

Example 1.26.  cfgutils.rand_reset_prob usage

...
$ kamcmd cfgutils.rand_reset_prob
...

5.3. cfgutils.rand_get_prob

Return the actual probability setting.

The function return the actual probability value.

Example 1.27. cfgutils.rand_get_prob usage

...
$ kamcmd cfgutils.rand_get_prob
...

5.4. cfgutils.check_config_hash

Check if the actual config file hash is identical to the stored one.

The function returns 200 OK if the hash values are identical, 400 if there are not identical, 404 if no file for hashing has been configured and 500 on errors. Additional a short text message is printed.

Example 1.28. cfgutils.check_config_hash usage

...
$ kamcmd cfgutils.check_config_hash
...

5.5. cfgutils.get_config_hash

Return the stored config file hash.

The function returns 200 OK and the hash value on success or 404 if no file for hashing has been configured.

Example 1.29. cfgutils.get_config_hash usage

...
$ kamcmd cfgutils.get_config_hash
...

5.6. cfgutils.set_gflag

Set the value of some flags (specified by bitmask) to 1.

The parameter value must be a bitmask in decimal or hexadecimal format. The bitmask has a 32 bit size.

Example 1.30. cfgutils.set_gflag usage

...
$ kamcmd cfgutils.set_gflag 1
...

5.7. cfgutils.reset_gflag

Reset the value of some flags to 0.

The parameter value must be a bitmask in decimal or hexadecimal format. The bitmask has a 32 bit size.

Example 1.31.  cfgutils.reset_gflag usage

...
$ kamcmd cfgutils.reset_gflag 1
...

5.8. cfgutils.is_gflag

Returns true if all the flags from the bitmask are set.

The parameter value must be a bitmask in decimal or hexadecimal format. The bitmask has a 32 bit size.

The function returns TRUE if all the flags from the set are set and FALSE if at least one is not set.

Example 1.32. cfgutils.is_gflag usage

...
$ kamcmd cfgutils.set_gflag 1024
$ kamcmd cfgutils.is_gflag 1024
TRUE
...

5.9. cfgutils.get_gflags

Return the bitmap with all flags. The function gets no parameters and returns the bitmap in hexadecimal and decimal format.

Example 1.33.  cfgutils.get_gflags usage

...
$ kamcmd cfgutils.get_gflags
...

6. Exported pseudo-variables

6.1. $RANDOM

Returns a random value from the [0 - 2^31) range.

Example 1.34. RANDOM pseudo-variable usage

...
if (rand_event()) {
  $var(value) = ($RANDOM / 16777216); # 2^24
  if ($var(value) < 10) {
     $var(value) = 10;
  }
  append_to_reply("Retry-After: $var(value)\n");
  sl_send_reply("503", "Try later");
  exit;
};
# normal message processing follows