Fork 多进程 模拟并行访问web service获取响应时间差

#include <ros/ros.h>
#include <iostream>
#include <string>
#include <cstring>
// 名称空间映射表
#include "k8s_sum/SumServiceImplPortBinding.nsmap"
#include "k8s_sum/soapSumServiceImplPortBindingProxy.h"
//using namespace std;
#include <stdio.h>
#include <iostream>
#include <unistd.h>
#include <fstream>
#include <time.h>
#include <sys/types.h>
#define NUM 1000
#define FORK_NUM 100

//获取当前时间
double getCurrentTime(){
    struct timeval tv;
    gettimeofday(&tv,NULL);
    //return tv.tv_sec*1000+tv.tv_usec/1000;//毫秒
    return tv.tv_sec*1000+tv.tv_usec/1000+tv.tv_usec%1000*0.001;//毫秒带小数
}
//计算数组内平均值
double getAvgDiff(double *diffs){
    double sum = 0.0;
    for(int i=0;i<NUM;i++){
        sum +=*(diffs+i);
    }
    return sum/NUM;
}

//总处理函数
void handle(){
    //输出到文件
    std::ofstream fout;
    int pid = getpid();
    std::string fileName="/tmp/rosSumfile_"+std::to_string(pid)+".dat";
    fout.open(fileName, std::ios::app);
    double first_time;//记录本次最开始时间
    double last_time;//记录本次最后一次时间
    double avg_diff;//平均响应时间
    double diffs[NUM] ={0.0};
    for(int i=0;i<NUM;i++){

        SumServiceImplPortBindingProxy sumWebservice;
        
        K8S1__getSum sumRequest;
        K8S1__getSumResponse res;
        int a,b;
        //两个1-100的随机数
        a= rand() % 100;
        b= rand() % 100;
        sumRequest.arg0 = a;
        sumRequest.arg1 = b;

        std::cout<<"a="<<a<<",b="<<b<<std::endl;

        time_t tt = time(NULL);//这句返回的只是一个时间戳 精确到秒
        //clock_t start_time,end_time; //这个会精确到毫秒
        //start_time = clock();
        double start_time,end_time;
        start_time = getCurrentTime();
        if(i==0){
            first_time = start_time;//记录第一次时间
        }
        fout<<"Current timestamp(k8s_sum) request(ms)="<<std::to_string(start_time);
        int result = sumWebservice.getSum(&sumRequest, res);
        //sleep(2); 暂停2秒
        if(SOAP_OK == result)
        {
            int sum_value = res.return_;
            //time_t tt1 = time(NULL);
            end_time = getCurrentTime();
            if(i+1==NUM){
                last_time = end_time;
            }
            fout<<",response(ms)="<<std::to_string(end_time);
            fout<<",the diff(ms)="<<end_time-start_time<<std::endl;
            diffs[i]=end_time-start_time;
        }else{
            fout<<",request is error!"<<std::endl;
        }

    }
    //计算平均响应时间
    avg_diff = getAvgDiff(diffs);
    fout<<"Count firstTime="<<std::to_string(first_time)<<",lastTime="<<std::to_string(last_time)<<",avgDiff="<<avg_diff<<std::endl;
    fout.close();
}
 
int main(int argc, char **argv)
{
    ros::init(argc, argv, "k8s_sum_node");
    pid_t pid;
    //创建子进程
    for(int i=0;i<FORK_NUM;i++){
        pid = fork();
        //子进程退出循环，不再创建子进程，全部由主进程创建子进程 ，这里是关键所在
        if(pid ==0 || pid == -1){
            break;
        }
    }
    if(pid == -1){
        ROS_INFO("Fail to Fork!");
        exit(1);
    }
    else if(pid == 0){
        //子进程处理逻辑
        handle();
        sleep(5);
        exit(0);
    }
    else{
        //主进程处理逻辑
        handle();
        exit(0);
    }
    
    ros::spin();
    return 0;
}

采用了ROS 的编程方式

Web Service 模拟了一个 加法运算，计算求和