2 The software in this package is distributed under the GNU General
3 Public License version 2 (with a special exception described below).
5 A copy of GNU General Public License (GPL) is included in this distribution,
6 in the file COPYING.GPL.
8 As a special exception, if other files instantiate templates or use macros
9 or inline functions from this file, or you compile this file and link it
10 with other works to produce a work based on this file, this file
11 does not by itself cause the resulting work to be covered
12 by the GNU General Public License.
14 However the source code for this file must still be made available
15 in accordance with section (3) of the GNU General Public License.
17 This exception does not invalidate any other reasons why a work based
18 on this file might be covered by the GNU General Public License.
26 #include <boost/asio.hpp>
27 #include <boost/foreach.hpp>
28 #include <boost/shared_ptr.hpp>
29 #include <boost/math/special_functions/round.hpp>
30 #include <boost/numeric/conversion/cast.hpp>
31 #include <boost/date_time/posix_time/posix_time_types.hpp>
33 #include <daemonfunc.hpp>
34 #include <logfunc.hpp>
36 #include "boost_assert_handler.h"
37 #include "config/configurationreader.h"
38 #include "config/host.h"
39 #include "link/linkstatus.h"
40 #include "host/loglevel.h"
41 #include "host/pingprotocol.h"
42 #include "host/pingscheduler.h"
43 #include "icmp/icmppinger.h" // contains IcmpPacketDistributor
44 #include "dns/dnsmaster.h"
48 using boost::shared_ptr;
49 using boost::posix_time::time_duration;
50 using I2n::Logger::GlobalLogger;
52 // a map from interval (in seconds) to delay (in seconds)
53 typedef std::pair<int, float> IntervalCountPair;
54 typedef std::map<int, float> DelayMap;
55 typedef shared_ptr<boost::asio::deadline_timer> TimerItem;
57 const boost::posix_time::time_duration SIGNAL_CHECK_INTERVAL = boost::posix_time::seconds(1);
59 //-----------------------------------------------------------------------------
61 //-----------------------------------------------------------------------------
63 typedef std::pair<bool, ConfigurationItem> GetConfigReturnType;
64 GetConfigReturnType get_configuration(int, const char**);
65 LinkStatusItem get_status_notifier(const ConfigurationItem&);
67 void set_log_output(const ConfigurationItem &);
68 DelayMap calc_pinger_delays(const HostList &hosts);
69 void init_pingers(const IoServiceItem, const ConfigurationItem&,
70 const LinkStatusItem&, PingSchedulerList*);
71 void start_pingers(const PingSchedulerList&);
72 void stop_pingers(const PingSchedulerList&);
74 void signal_handler_int(int param);
75 void signal_handler_term(int param);
76 void signal_handler_usr1(int param);
77 void signal_handler_usr2(int param);
78 void signal_checker( const boost::system::error_code &error );
79 void install_signal_handlers( const IoServiceItem io_service, const int config_log_level );
80 void reset_signal_handlers();
82 // data required for signal handling (SIGINT, SIGTERM, ... )
83 struct signal_data_struct
85 volatile sig_atomic_t signaled_flag_int;
86 volatile sig_atomic_t signaled_flag_term;
87 volatile sig_atomic_t signaled_flag_usr1;
88 volatile sig_atomic_t signaled_flag_usr2;
89 IoServiceItem io_service;
90 void (*old_handler_int )(int);
91 void (*old_handler_term)(int);
92 void (*old_handler_usr1)(int);
93 void (*old_handler_usr2)(int);
95 TimerItem check_timer;
99 signaled_flag_int( 0 ),
100 signaled_flag_term( 0 ),
101 signaled_flag_usr1( 0 ),
102 signaled_flag_usr2( 0 ),
104 old_handler_int( 0 ),
105 old_handler_term( 0 ),
106 old_handler_usr1( 0 ),
107 old_handler_usr2( 0 ),
110 config_log_level( I2n::Logger::LogLevel::Notice )
114 //-----------------------------------------------------------------------------
116 //-----------------------------------------------------------------------------
118 GetConfigReturnType get_configuration(
123 ConfigurationReader config_reader;
124 bool parsed_success = config_reader.parse( argc, argv );
125 Configuration config_obj = config_reader.get_configuration();
127 ConfigurationItem configuration( new Configuration( config_obj ) );
128 GetConfigReturnType return_val( parsed_success, configuration );
132 LinkStatusItem get_status_notifier(
133 const ConfigurationItem &configuration
136 int hosts_down_limit = configuration->get_hosts_down_limit();
137 int link_up_interval_in_min = configuration->get_link_up_interval_in_min();
138 int link_down_interval_in_min = configuration->get_link_down_interval_in_min();
139 string status_notifier_cmd = configuration->get_status_notifier_cmd();
140 LinkStatusItem link_analyzer(
143 link_up_interval_in_min,
144 link_down_interval_in_min,
149 return link_analyzer;
154 // set default: log at level NOTICE to syslog and stderr
155 // to ensure that in case of faulty config, the error is noticed
156 I2n::Logger::enable_syslog( I2n::Logger::Facility::User );
157 I2n::Logger::enable_stderr_log( true );
158 I2n::Logger::set_log_level( I2n::Logger::LogLevel::Notice );
162 const ConfigurationItem &configuration
165 LogOutput log_output = configuration->get_log_output();
168 case LogOutput_SYSLOG:
169 GlobalLogger.info() << "Setting log output target to syslog" << endl;
170 I2n::Logger::enable_syslog(true);
171 I2n::Logger::enable_stderr_log(false);
172 I2n::Logger::enable_log_file(false);
173 GlobalLogger.info() << "Set log output target to syslog" << endl;
175 case LogOutput_TERMINAL:
176 GlobalLogger.info() << "Setting log output target to terminal" << endl;
177 I2n::Logger::enable_syslog(false);
178 I2n::Logger::enable_stderr_log(true);
179 I2n::Logger::enable_log_file(false);
180 GlobalLogger.info() << "Set log output target to terminal" << endl;
181 GlobalLogger.info() << "(check syslog for earlier messages)" << endl;
184 GlobalLogger.error() << "Unknown log output target!" << endl;
190 * @brief calculate delay between pingers to evenly distribute them in time
192 * If there are many pingers with same interval, will get bursts of pings
193 * and none in-between. This function calculates delays for large numbers
194 * of hosts with same ping intervals, to distribute them as evenly as
195 * possible, right from the start (might diverge over time, anyway).
197 * Will not do much good for pingers with many different intervals, but
198 * then is not required anyway and does no(t much) harm.
200 * Called by init_pingers with
201 * @param hosts list of hosts as obtained from configuration
202 * @returns a map from ping interval to delay between pingers of that interval
204 DelayMap calc_pinger_delays(const HostList &hosts)
206 // first step: count number of hosts with same intervals
207 DelayMap delay_shifts;
209 BOOST_FOREACH( const HostItem &host, hosts )
211 curr_interval = host->get_interval_in_sec();
213 delay_shifts[curr_interval] = 1.0f;
215 delay_shifts[curr_interval] += 1.0f;
218 // second step: divide intervals by counts, round to int
219 // --> for 18 pingers with a 30s interval, get 30s/18 = 1.66667
220 BOOST_FOREACH( IntervalCountPair interval_and_count, delay_shifts )
221 delay_shifts[interval_and_count.first] =
222 boost::numeric_cast<float>(interval_and_count.first) /
223 interval_and_count.second;
229 const IoServiceItem io_service,
230 const ConfigurationItem &configuration,
231 const LinkStatusItem &status_notifier,
232 PingSchedulerList *scheduler_list
235 string default_network_interface = configuration->get_source_network_interface();
236 int ping_fail_limit = configuration->get_ping_fail_limit();
237 int ping_reply_timeout = configuration->get_ping_reply_timeout();
239 // remove some hosts at random
240 configuration->randomize_hosts();
242 // calculate delays between pingers of same interval
243 DelayMap delay_shifts = calc_pinger_delays(configuration->get_hosts());
245 // setup memory for assigned delays
247 BOOST_FOREACH( IntervalCountPair interval_and_delay, delay_shifts )
248 delays[interval_and_delay.first] = 0.0f;
250 HostList hosts = configuration->get_hosts();
251 BOOST_FOREACH( const HostItem &host, hosts )
253 string destination_address = host->get_address();
254 uint16_t destination_port = host->get_port();
255 string host_network_interface = host->get_source_network_interface();
256 string network_interface = ( host_network_interface == "default" ) ?
257 default_network_interface :
258 host_network_interface;
259 PingProtocolList protocol_list = host->get_ping_protocol_list();
260 int ping_interval_in_sec = host->get_interval_in_sec();
262 // get delay for this scheduler and update assigned delays
263 int current_delay = boost::math::iround(delays[ping_interval_in_sec]);
264 delays[ping_interval_in_sec] += delay_shifts[ping_interval_in_sec];
266 PingSchedulerItem scheduler(
273 ping_interval_in_sec,
280 scheduler_list->push_back( scheduler );
285 const PingSchedulerList &scheduler_list
288 // start each ping scheduler
289 BOOST_FOREACH( const PingSchedulerItem &scheduler, scheduler_list )
290 scheduler->start_pinging();
294 const PingSchedulerList &scheduler_list
297 // Stop each ping scheduler
298 GlobalLogger.info() << "Telling all pingers to stop";
299 BOOST_FOREACH( const PingSchedulerItem &scheduler, scheduler_list )
301 scheduler->stop_pinging();
304 IcmpPacketDistributor::clean_up_all();
308 // the one instance of signal_data_struct
309 signal_data_struct signal_data;
312 /// registered as signal handler; just sets signal_data.signaled_flag
313 void signal_handler_int(int param)
315 signal_data.signaled_flag_int = 1;
317 void signal_handler_term(int param)
319 signal_data.signaled_flag_term = 1;
321 void signal_handler_usr1(int param)
323 signal_data.signaled_flag_usr1 = 1;
325 void signal_handler_usr2(int param)
327 signal_data.signaled_flag_usr2 = 1;
331 /// called regularly from io_service; checks signal_data.signal_flag
332 void signal_checker( const boost::system::error_code &error )
334 bool want_stop = false;
337 { // there was an error in the timer
338 if ( error == boost::asio::error::operation_aborted )
340 GlobalLogger.error() << "Signal check timer was cancelled! Stopping io_service" << endl;
345 GlobalLogger.error() << "Signal check timer handler received error code " << error
346 << "! Stopping io_service" << endl;
351 if ( signal_data.signaled_flag_int )
353 signal_data.signaled_flag_int = 0;
354 GlobalLogger.notice() << "Received signal SIGINT --> will stop" << endl;
357 else if ( signal_data.signaled_flag_term )
359 signal_data.signaled_flag_term = 0;
360 GlobalLogger.notice() << "Received signal SIGTERM --> will stop" << endl;
363 else if ( signal_data.signaled_flag_usr1 )
365 signal_data.signaled_flag_usr1 = 0;
366 int new_log_level = I2n::Logger::get_log_level()+1;
367 I2n::Logger::set_log_level( new_log_level );
368 GlobalLogger.info() << "Received SIGUSR1 -- increased log level to "
369 << I2n::Logger::get_log_level_string();
371 else if ( signal_data.signaled_flag_usr2 )
373 signal_data.signaled_flag_usr2 = 0;
374 I2n::Logger::set_log_level( signal_data.config_log_level );
375 GlobalLogger.info() << "Received SIGUSR2 -- reset log level to normal ("
376 << I2n::Logger::get_log_level_string() << ")";
381 { // interrupt infinite loop in main and asio event loop
382 signal_data.stopped = true;
383 signal_data.io_service->stop();
386 { // re-schedule timer
387 signal_data.check_timer->expires_from_now( SIGNAL_CHECK_INTERVAL );
388 signal_data.check_timer->async_wait( signal_checker );
392 /// register own signal handlers; see reset_signal_handlers for undo
393 void install_signal_handlers( const IoServiceItem io_service, const int config_log_level )
395 signal_data.signaled_flag_int = 0;
396 signal_data.signaled_flag_term = 0;
397 signal_data.signaled_flag_usr1 = 0;
398 signal_data.signaled_flag_usr2 = 0;
399 signal_data.config_log_level = config_log_level;
401 // install own signal handlers
402 signal_data.old_handler_int = signal(SIGINT, signal_handler_int);
403 signal_data.old_handler_term = signal(SIGTERM, signal_handler_term);
404 signal_data.old_handler_usr1 = signal(SIGUSR1, signal_handler_usr1);
405 signal_data.old_handler_usr2 = signal(SIGUSR2, signal_handler_usr2);
406 if ( signal_data.old_handler_int == SIG_ERR ||
407 signal_data.old_handler_term == SIG_ERR ||
408 signal_data.old_handler_usr1 == SIG_ERR ||
409 signal_data.old_handler_usr2 == SIG_ERR )
410 throw runtime_error( string("Failed to install signal handler: ") + string(strerror(errno)) );
412 // create a timer and a shared pointer to it, so it does not get out of scope
413 TimerItem check_timer( new boost::asio::deadline_timer( *io_service ) );
415 // remember the io_service and the timer
416 signal_data.io_service = io_service;
417 signal_data.check_timer = check_timer;
420 check_timer->expires_from_now( SIGNAL_CHECK_INTERVAL );
421 check_timer->async_wait( signal_checker );
422 GlobalLogger.debug() << "signal timer set" << endl;
425 /// reset handlers to the ones saved in install_signal_handlers
426 void reset_signal_handlers()
428 void (*old_handler_int)(int) = 0;
429 void (*old_handler_term)(int) = 0;
430 void (*old_handler_usr1)(int) = 0;
431 void (*old_handler_usr2)(int) = 0;
432 if (signal_data.old_handler_int != 0 )
433 old_handler_int = signal(SIGINT , signal_data.old_handler_int);
434 if (signal_data.old_handler_term != 0 )
435 old_handler_term = signal(SIGTERM, signal_data.old_handler_term);
436 if (signal_data.old_handler_usr1 != 0 )
437 old_handler_usr1 = signal(SIGUSR1, signal_data.old_handler_usr1);
438 if (signal_data.old_handler_usr2 != 0 )
439 old_handler_usr2 = signal(SIGUSR2, signal_data.old_handler_usr2);
441 if ( old_handler_int == SIG_ERR ||
442 old_handler_term == SIG_ERR ||
443 old_handler_usr1 == SIG_ERR ||
444 old_handler_usr2 == SIG_ERR )
445 throw runtime_error( string("Failed to reset signal handler: ") + string(strerror(errno)) );
449 int main( int argc, const char *argv[] )
452 GlobalLogger.debug() << "logger initiated with default config";
454 PingSchedulerList scheduler_list;
455 IoServiceItem io_service;
457 unsigned n_exceptions = 0;
458 unsigned max_exceptions = 1;
462 GetConfigReturnType success_and_config = get_configuration( argc, argv );
463 ConfigurationItem configuration = success_and_config.second;
465 if ( configuration->get_print_version() ) // do this even if parsing of config failed
467 GlobalLogger.debug() << "Printing version info (" << VERSION_STRING << ") and exit" << endl;
468 cout << PROJECT_NAME << " version " << VERSION_STRING << "."
469 << VERSION_REVISION_STRING << " build " << __DATE__ << endl;
473 if ( ! success_and_config.first )
475 GlobalLogger.error() << "Could not read/parse configuration!";
476 GlobalLogger.debug() << "Return 1 immediately" << endl;
479 GlobalLogger.debug() << "Start setup" << endl;
481 int log_level = configuration->get_log_level();
482 I2n::Logger::set_log_level( log_level );
483 GlobalLogger.info() << "Set LogLevel to " << I2n::Logger::get_log_level_string() << endl;
485 set_log_output( configuration );
486 GlobalLogger.notice() << "started";
488 bool daemon_mode = configuration->get_daemon();
491 I2n::Daemon::daemonize();
494 LinkStatusItem status_notifier = get_status_notifier( configuration );
496 IoServiceItem io_service_temp( new boost::asio::io_service() );
497 io_service_temp.swap( io_service );
498 io_service_temp.reset();
501 boost::asio::ip::address name_server_ip =
502 boost::asio::ip::address::from_string(
503 configuration->get_nameserver() );
505 DnsMaster::create_master(
508 configuration->get_resolved_ip_ttl_threshold(),
509 configuration->get_max_address_resolution_attempts(),
510 configuration->get_dns_cache_file() );
512 init_pingers( io_service, configuration, status_notifier, &scheduler_list );
514 install_signal_handlers( io_service, log_level );
516 start_pingers( scheduler_list );
518 catch ( const std::exception &ex )
520 GlobalLogger.error() << "Uncaught exception. " << ex.what() << endl;
525 GlobalLogger.error() << "Caught unknown exception!" << endl;
532 GlobalLogger.info() << "starting io_service main loop" << endl;
534 if (max_exceptions > 0)
535 GlobalLogger.warning() << "Limited number of acceptable exceptions,"
536 << " this is a debugging option!";
538 // call boost::asio main event loop, catching exceptions
539 while ( !signal_data.stopped )
545 catch ( const std::exception &ex )
548 GlobalLogger.error() << "Caught exception, will continue. " << ex.what() << endl;
552 GlobalLogger.error() << "Caught unknown exception, will continue!" << endl;
555 if (max_exceptions > 0 && n_exceptions >= max_exceptions)
557 GlobalLogger.info() << "reached max number of exceptions allowed in main loop" << endl;
559 signal_data.stopped = true;
562 else if ( signal_data.stopped )
563 GlobalLogger.info() << "exiting io_service main loop" << endl;
564 // ( io_service->stop() has been called already in signal handler)
566 GlobalLogger.info() << "continuing io_service main loop" << endl;
573 GlobalLogger.info() << "Cleaning up" << endl;
574 stop_pingers( scheduler_list );
575 reset_signal_handlers();
577 catch ( const std::exception &ex )
579 GlobalLogger.error() << "Uncaught exception while cleaning up: " << ex.what() << endl;
583 GlobalLogger.error() << "Caught unknown exception while cleaning up!" << endl;
587 GlobalLogger.notice() << "Pingcheck done " << endl;
588 GlobalLogger.debug() << "In main loop, had " << n_exceptions << " exception[s]" << endl;