Join（）调用正在冻结执行，而我使用multyprocessing modu创建进程

Question

我正在使用多处理函数进行一些测试，在调用连接函数后，我面临一个冻结问题

import time
import math
import pandas as pd
import numpy as np
from multiprocessing import Process, Queue, Lock, Value

def writeDF(start,end,td,lock,q):
    print('write from ',start,' ',end)
    for  i in range(start,end):
        lock.acquire()
        q.put(td.iloc[i,:])
        lock.release()
    print('function writeDF completed')

if __name__ == '__main__':
    td=pd.read_csv(r'C:\Users\dorian\Desktop\Analyzer.txt', encoding = "ISO-8859-1", index_col="Start", parse_dates=True, sep=',')
    td=td[0:10] 
    jobs=[]
    q = Queue()
    lock = Lock()


    start = time.time()

    begin=0
    stop=td.shape[0]/2

    for i in range(0,2):
        if len(jobs) < 10:
            t = Process(target=writeDF, args=(int(begin),int(stop),td, lock,q))
            t.start()
            jobs.append(t)
            begin=stop
            stop=stop+(td.shape[0]/2)
        else:    
            t = jobs.pop(0)
            t.join()

    for p in jobs: 
        print('try to join element:',p)
        p.join()
        print('element is join')

    l=[]    
    while not q.empty():
        l.append(q.get())

    df=pd.DataFrame(l)
    end = time.time()  
    print('Value:',df.shape)
    print('2 process Time taken in seconds -', end - start)

实际结果是程序在join（）语句上冻结，但是由于start（）在join（）之前，所以不应该阻止它

Answer 1

multiprocessing.Queue队列的容量受到用于实现它们的底层操作系统级管道的限制。这意味着你不能等到作者写完后再阅读。写入程序和读取器必须并发执行才能安全地使用multiprocessing.Queue。你没那么做