http://musicforprogramming.net/?one 不谢

import asyncio

async def slow_operation(n):
await asyncio.sleep(n)
print('Slow operation {} complete'.format(n))
return n


loop = asyncio.get_event_loop()
done, _ = loop.run_until_complete(
asyncio.wait([
slow_operation(1),
slow_operation(2),
slow_operation(9),
slow_operation(2),
slow_operation(1),
slow_operation(2),
slow_operation(3),
]))
for fut in done:
print("return value is {}".format(fut.result()))

然后用 uvloop

y worker -A app.tasks.celery -l INFO -Q default -c 20 （每个队列多搞几个 worker ） -n default_worker.%%i

from time import sleep
from concurrent.futures import ThreadPoolExecutor\ ProcessPoolExecutor
def child_1():
sleep(9)
print(1)


def child_2():
sleep(2)
print(2)



def child_3():
sleep(3)
print(3)



def child_4():
sleep(1)
print(4)


def child_5():
sleep(2)
print(5)



with ThreadPoolExecutor\ProcessPoolExecutor(max_workers=5) as executor:
executor.submit(child_1)
executor.submit(child_2)
executor.submit(child_3)
executor.submit(child_4)
executor.submit(child_5)


t1= executor.submit(child_1)
t2=executor.submit(child_2)
t3=executor.submit(child_3)
t4=executor.submit(child_4)
t5=executor.submit(child_5)

print(t1.result())

作死的头像

销售 副业 写代码 ？？？ 6666

引起不适，block

有毛用，看还不如 google 快

mark

@zhoudaiyu

可以使用 celery 提供的信号接口，监测每个任务的执行时间，然后发给 statsd 或者 promethus 之类的

https://github.com/getredash/redash/blob/master/redash/metrics/celery.py

这个是代码例子

@mercury233 捞一波是一波

插眼

In [1]: import asyncio
...:
...: try:
...: # pylint: disable=invalid-name
...: asyncio_run = asyncio.run # type: ignore
...: except AttributeError:
...:
...: def asyncio_run(main, debug=False):
...: """Minimal re-implementation of asyncio.run (since 3.7)."""
...: loop = asyncio.new_event_loop()
...: asyncio.set_event_loop(loop)
...: loop.set_debug(debug)
...: try:
...: return loop.run_until_complete(main)
...: finally:
...: asyncio.set_event_loop(None)
...: loop.close()
...:
...: async def slow_operation_1(m):
...: await asyncio.sleep(m)
...: print('Slow operation {} complete'.format(m))
...: return m
...:
...:
...:
...: async def slow_operation_2(n):
...: await asyncio.sleep(n)
...: print('Slow operation {} complete'.format(n))
...:
...: return await slow_operation_1(m=n+1)
...:
...: result = asyncio_run(slow_operation_2(2))
...:
...: print(result)
...:

Slow operation 2 complete
Slow operation 3 complete
3

RabbitMQ is feature-complete, stable, durable and easy to install. It ’ s an excellent choice for a production environment.

官方文档推荐的~
@zhoudaiyu

建议用 rabittmq。
还算稳定,分享下我的配置



CELERY_IGNORE_RESULT = True
CELERY_BROKER_URL = '‘ xxxx ’
CELERY_TIMEZONE = 'xxx'
CELERY_DEFAULT_QUEUE = 'default'
CELERY_CREATE_MISSING_QUEUES = 'default'

BROKER_HEARTBEAT = 24*60*60*2

CELERY_ENABLE_UTC = False
CELERY_DISABLE_RATE_LIMITS = True
CELERYD_MAX_TASKS_PER_CHILD = 10
BROKER_TRANSPORT_OPTIONS = {'visibility_timeout': 1800}
CELERYD_FORCE_EXECV = True
BROKER_POOL_LIMIT = None

今天打算把你顶上去，这么欢乐的事情居然发生在 2019 年

看到较强的抗压能力就发抖

虽然你们的前端确实挺菜的，但是直接辞退怕是有点过分。你们的技术 leader 连前端忽悠你们都分辨不出来，他也有很大的责任