c++ 进程间心跳检测--socket （UDP）

目录

代码结构如下：

代码解析：

check tread:

receive tread:

send  tread:

完整代码工程下载：

编译代码方式：

---

功能实现：可以检测 进程是否 live 或者 僵死，然后根据具体状态，发signal 处理被检测的进程

 demo 有两个进程 一个是 server 另一个是 client (可以有多个)（send && receive）

代码结构如下：

├── bin
├── include
│   ├── heartbeat.h
│   ├── Log.h
│   └── udpnet.h
├── Makefile
├── make.sh
├── obj
└── src
    ├── heartbeat.cpp
    ├── Log.cpp
    ├── recv
    │   └── heartbeatMonitor.cpp
    ├── send
    │   └── heartbeatMonitor_send.cpp
    └── udpnet.cpp

代码解析：

两个进程，主要有3个线程来实现：send tread  receive tread  check thread

check tread:

void* HeartbeatMonitor::checkThread( void* param )
{
	const int LOST_LIMIT = 5;

	static int call_cnt = 0;

	HeartbeatMonitor* monitor = (HeartbeatMonitor*)param;
	HeartbeartStatistic	&statistic = monitor->statistic; //使用 引用调用 HeartbeatMonitor 

	int	pkg_cnt = 0;
	bool err_found = false;
	Heartbeat new_heartbeat;
	HeartbeatChecker checker;

	hb_usleep(HEARTBEAT_CYCLE_MS * 1000); // 3000 ms
	while(1)
	{ 
		int &pkg_lost_cnt = checker.pkg_lost_cnt;

		if (monitor->pkg_cnt == pkg_cnt) //丢包判断
		{
			pkg_lost_cnt++;
			statistic.pkg_lost_cnt++;
			hbprintf(DEBUG_LOG, DBG_HEAD " YL Pkg lost for %d times.", pkg_lost_cnt); 

			if (LOST_LIMIT < pkg_lost_cnt)
			{	
				checker.recover();

				statistic.pkg_lost_kill_cnt++;			
			}

			goto LOOP_DELAY;
		}
		else
		{
			pkg_cnt = monitor->pkg_cnt;
			pkg_lost_cnt = 0;
		}		
	
		new_heartbeat = monitor->getHeartbeat();

		err_found = false;
		for (int i = 0; i < HEARTBEAT_COUNTER_NUM; ++i) //检测不同模块的进程的 状态counter
		{
			
			if (new_heartbeat.counter[i] == checker.heartbeat.counter[i])
			{
				err_found = true;
				break;
			}
		}

		if (err_found)
		{
			//new_heartbeat.showContent(DBG_HEAD);
		}
	
		for (int i = 0; i < HEARTBEAT_COUNTER_NUM; ++i)
		{
			int &lost_cnt = checker.lost_cnt[i];
			if (new_heartbeat.counter[i] != checker.heartbeat.counter[i])
			{
				lost_cnt++;
				statistic.lost_cnt.counter[i]++;
				hbprintf(DEBUG_LOG, DBG_HEAD "\"%s\" lost for %d times.", Heartbeat::counter_name[i], lost_cnt); 
				
				if (LOST_LIMIT < lost_cnt)
				{	
					checker.recover(); //丢包 超过次数后执行 操作

					statistic.kill_cnt.counter[i]++;
					
					goto LOOP_DELAY;
				}
			}
			else
			{
				lost_cnt = 0;
			}
		}

		checker.heartbeat = new_heartbeat;
		statistic.last_heart = new_heartbeat;

	LOOP_DELAY:

		++call_cnt;
		if (call_cnt % HEARTBEAT_PRINT_CYCLE == 0)
		{
			statistic.showContent("HeartbeatMonitor");
		}
		
		hb_usleep(HEARTBEAT_CYCLE_MS * 1000); // 3000 ms
	}

	return (void *)0;
}

 

receive tread:

/* static */ void* HeartbeatMonitor::recvThread( void* param )
{
#define DBG_HEAD "[Recv] "

	bool 	ret = true;
	UINT32 	peer_ip; //lo: 127.0.0.1   Loopback address
	int 	peer_port;
	double  heartbeat_cycle_sec = 1.0 * HEARTBEAT_CYCLE_MS / 1000;

	HeartbeatMonitor* monitor = (HeartbeatMonitor*)param;
	HeartbeartStatistic	&statistic = monitor->statistic;

	while(1)
	{    	
	    ret = monitor->netAdp.NetRecv(monitor->recv_buff, monitor->recv_size, heartbeat_cycle_sec, &peer_ip, &peer_port);
        if (ret == false)
        {
            if (monitor->recv_size == 0)
            {
            	hbprintf(DEBUG_LOG, DBG_HEAD "Time out at NetRecv.");
				statistic.timeout_cnt++;
			}
            else
            {
            	hbprintf(DEBUG_LOG, DBG_HEAD "ERROR at NetRecv.");
				statistic.err_cnt++;
            }

			continue;
        }
		
		//hbprintf(DEBUG_LOG, DBG_HEAD "Hearbeat recv ok.");
		monitor->pkg_cnt++;
		statistic.ok_cnt++;

		monitor->parseHeartbeatPkg();
	}

	return (void *)0;

#undef DBG_HEAD
}

send  tread:

/* static */ void* HeartbeatSender::sendThread( void* param )
{
#define DBG_HEAD "[ YL -- Send] "

	static int call_cnt = 0;

	int iRet = true;
         bool ret = true;

	HeartbeatSender* sender = (HeartbeatSender*)param;
	HeartbeartStatistic	&statistic = sender->statistic;

	while(1)
	{   
		if (HeartbeatPkgUT_ON())
		{
			goto CYCLE_DELEY; 
		}
		
		sender->generateHeartbeatPkg();
		
		hbprintf(DEBUG_LOG," YL  -- dest_ip = %d ,dest_port = %d,sender->send_size =%d , sender->send_buff =%s",sender->dest_ip,sender->dest_port,sender->send_size,sender->send_buff);

		ret = sender->netAdp.NetSend(sender->dest_ip, 
									sender->dest_port, 
									sender->send_buff, 
									sender->send_size, 
									&iRet);
	    if (ret == false)
	    {
	        if (iRet == 0)
	        {
	            hbprintf(DEBUG_LOG, DBG_HEAD "Time out at NetSend.");  
				statistic.timeout_cnt++;
	        }
	        else
	        {
	            hbprintf(DEBUG_LOG, DBG_HEAD "ERROR at NetSend.");
				statistic.err_cnt++;
	        }

			goto CYCLE_DELEY;
	    }

		// hbprintf(DEBUG_LOG, DBG_HEAD "Hearbeat send ok.");
		statistic.last_heart = sender->getHeartbeat();
		statistic.ok_cnt++;
		
	CYCLE_DELEY:

		++call_cnt;
		if (call_cnt % HEARTBEAT_PRINT_CYCLE == 0)
		{
			statistic.showContent("HeartbeatSender");
		}

		hb_usleep(HEARTBEAT_CYCLE_MS * 1000); // 3000 ms
	}

	return (void *)0;

#undef DBG_HEAD
}

完整代码工程下载：

To click on：core download

编译代码方式：

make all  即可以编译出搜有模块

有兴趣的可以修改 Makefile和代码 编译多个  receive 端