关于python中协程asynico 与进程池pool冲突的问题

Question

使用Pool进程池并发，然后让将每个进程的任务注册在一个在事件循环中，但Pool没有生效，是哪里出现了问题？

---

from multiprocessing import  Pool
import asyncio
import time

async def fristwork():
    await asyncio.sleep(1)
    print("fristwork take" ,str(time.time()))
    return "Done"
async def secondwork():
    a = await  fristwork()
    print (a)
def task(num):
    coroutine = secondwork()
    loop = asyncio.get_event_loop()
    loop.run_until_complete(coroutine)
    print("task {}".format(num))

async def run_more(num):
    print("start run_more")
    pool = Pool(processes = 2)
    for i in range(num):
        pool.apply_async(task,args=(i,))
        print(i)
    pool.close()
    pool.join()
def main(num):
    coroutine = run_more(num)
    tasks = [
        asyncio.ensure_future(coroutine),
    ]
    loop2 = asyncio.get_event_loop()
    loop2.run_until_complete(asyncio.wait(tasks))
if __name__ == '__main__':
    main(4)

以下为输出:

---

start run_more
0
1
2
3
fristwork take 1509178144.5045896
Done
task 0
fristwork take 1509178144.5145957
Done
task 1
fristwork take 1509178145.506057
Done
task 2
fristwork take 1509178145.517011
Done
task 3

Answer 1

asyncio多进程情况下最好不要使用同一个loop，用new_event_loop为每个子进程分配一个loop，因为asyncio在不同进程之间轮训时，没有进程安全限制，也没有根据进程id区分不同进程的工作任务，asyncio最新的开发计划中将增加这部分处理，可以等等看