cppcms-users

[Cppcms-users] Can't catch signals after service::run()

[CN <cn...@fa...>]

I have read
http://cppcms.com/wikipp/en/page/cppcms_1x_reload_application and the
mailing list archive dated 2013-11-19 08:05:50 GMT before posting my
question.

I need to do some clean-up tasks after the program is killed (kill
<pid>), but before it terminates. As shown below, version 1 can catch
signals.

Version 2 fails to catch signals. I wonder this is because
cppcms::service::run() overwrites all the signal specifications set
before it. How do I circumvent it?
//------------------------------------------------
#include <iostream>
#include <signal.h>
#include <unistd.h>
#include <cstring>
#include <cppcms/service.h>
#include <cppcms/thread_pool.h>
void signal_handler(int signal_number);
static bool got_signal=false;

int main(int argc,char ** argv)
{
	struct sigaction actions;
	memset(&actions,0,sizeof(actions));
	sigemptyset(&actions.sa_mask);
	actions.sa_flags=0;
	actions.sa_handler=signal_handler;
	int sig_result;
	if((sig_result=sigaction(SIGHUP,&actions,NULL)) != 0){
		std::cerr << "sigaction(SIGHUP) failed: " << sig_result
		<< std::endl;
		std::terminate();
	}
	if((sig_result=sigaction(SIGTERM,&actions,NULL)) != 0){
		std::cerr << "sigaction(SIGTERM) failed: " << sig_result
		<< std::endl;
		std::terminate();
	}
	/*
	//version 1: signal can be caught.
	while(true){
		std::cerr<<"loop"<<std::endl;
		if(got_signal){
			std::cerr<<"caught signal"<<std::endl;
			break;
		}
		sleep(1);
	}
	*/
	//version 2: signal can NOT be caught.
	cppcms::thread_pool pool(3);
	//pools.post(...);
	cppcms::service s(argc,argv);
	s.run();
	if(got_signal){
		pool.stop(); //Stop all in progress threads before
		program terminates.
		std::cerr<<"caught signal"<<std::endl;
	}
	return 0;
}

void signal_handler(int signal_number)
{
	got_signal=true;
}
//------------------------------------------------
Regards,
CN

-- 
http://www.fastmail.fm - mmm... Fastmail...

Re: [Cppcms-users] Can't catch signals after service::run()

[Artyom B. <art...@ya...>]

By default, when the SIGTERM is sent to the application, the cppcms::run() exits and you can perform any cleanup you need.

No need to do anything special.

The situation described in wiki is for reloading (i.e. restarting the service rather than exiting the full program)

Artyom Beilis
--------------
CppCMS - C++ Web Framework:   http://cppcms.com/
CppDB - C++ SQL Connectivity: http://cppcms.com/sql/cppdb/

--------------------------------------------
On Mon, 12/9/13, CN <cn...@fa...> wrote:

 Subject: [Cppcms-users] Can't catch signals after service::run()
 To: cpp...@li...
 Date: Monday, December 9, 2013, 5:35 PM
 
 I have read
 http://cppcms.com/wikipp/en/page/cppcms_1x_reload_application
 and the
 mailing list archive dated 2013-11-19 08:05:50 GMT before
 posting my
 question.
 
 I need to do some clean-up tasks after the program is killed
 (kill
 <pid>), but before it terminates. As shown below,
 version 1 can catch
 signals.
 
 Version 2 fails to catch signals. I wonder this is because
 cppcms::service::run() overwrites all the signal
 specifications set
 before it. How do I circumvent it?
 //------------------------------------------------
 #include <iostream>
 #include <signal.h>
 #include <unistd.h>
 #include <cstring>
 #include <cppcms/service.h>
 #include <cppcms/thread_pool.h>
 void signal_handler(int signal_number);
 static bool got_signal=false;
 
 int main(int argc,char ** argv)
 {
     struct sigaction actions;
     memset(&actions,0,sizeof(actions));
     sigemptyset(&actions.sa_mask);
     actions.sa_flags=0;
     actions.sa_handler=signal_handler;
     int sig_result;
    
 if((sig_result=sigaction(SIGHUP,&actions,NULL)) != 0){
         std::cerr <<
 "sigaction(SIGHUP) failed: " << sig_result
         << std::endl;
         std::terminate();
     }
    
 if((sig_result=sigaction(SIGTERM,&actions,NULL)) != 0){
         std::cerr <<
 "sigaction(SIGTERM) failed: " << sig_result
         << std::endl;
         std::terminate();
     }
     /*
     //version 1: signal can be caught.
     while(true){
        
 std::cerr<<"loop"<<std::endl;
         if(got_signal){
            
 std::cerr<<"caught signal"<<std::endl;
            
 break;
         }
         sleep(1);
     }
     */
     //version 2: signal can NOT be caught.
     cppcms::thread_pool pool(3);
     //pools.post(...);
     cppcms::service s(argc,argv);
     s.run();
     if(got_signal){
         pool.stop(); //Stop
 all in progress threads before
         program terminates.
        
 std::cerr<<"caught signal"<<std::endl;
     }
     return 0;
 }
 
 void signal_handler(int signal_number)
 {
     got_signal=true;
 }
 //------------------------------------------------
 Regards,
 CN
 
 -- 
 http://www.fastmail.fm - mmm... Fastmail...
 
 
 ------------------------------------------------------------------------------
 Sponsored by Intel(R) XDK 
 Develop, test and display web and hybrid apps with a single
 code base.
 Download it for free now!
 http://pubads.g.doubleclick.net/gampad/clk?id=111408631&iu=/4140/ostg.clktrk
 _______________________________________________
 Cppcms-users mailing list
 Cpp...@li...
 https://lists.sourceforge.net/lists/listinfo/cppcms-users
 

Re: [Cppcms-users] Can't catch signals after service::run()

[CN <cn...@fa...>]

Hmm... Below simple example shows that statements following run() never
get executed once the program is killed:

#include <iostream>
#include <cstring>
#include <cppcms/service.h>
#include <cppcms/thread_pool.h>
int main(int argc,char ** argv)
{
	cppcms::thread_pool pool(3);
	//pool.post(...); //Launch long running threads.
	cppcms::service s(argc,argv);
	s.run();
	//Program counter never reaches here.
	std::cerr<<"after run()"<<std::endl;
	pool.stop(); //Wait until all threads complete.
	return 0;
}

Can incorrect configuration content in config.js cause programs to
malfunction?

Best regards,
CN

On Tue, Dec 10, 2013, at 12:08 AM, Artyom Beilis wrote:
> By default, when the SIGTERM is sent to the application, the
> cppcms::run() exits and you can perform any cleanup you need.
> 
> No need to do anything special.
> 
> The situation described in wiki is for reloading (i.e. restarting the
> service rather than exiting the full program)
> 
> Artyom Beilis

-- 
http://www.fastmail.fm - Access all of your messages and folders
                          wherever you are