Celery--任务调度利器

Celery文档: http://docs.jinkan.org/docs/celery/getting-started/first-steps-with-celery.html

安装celery和celery-with-redis

pip install Celery
pip install celery-with-redis

开始编写task.py

# tasks.py
import time
from celery import Celery

celery = Celery('tasks', broker='redis://localhost:6379/0')

@celery.task
def sendmail(mail):
    print('sending mail to %s...' % mail['to'])
    time.sleep(2.0)
    print('mail sent.')

启动Celery处理任务

$ celery -A tasks worker --loglevel=info

上面的命令行实际上启动的是Worker，如果要放到后台运行，可以扔给supervisor。

发送任务

>>> from tasks import sendmail
>>> sendmail.delay(dict(to='[email protected]'))
<AsyncResult: 1a0a9262-7858-4192-9981-b7bf0ea7483b>

可以看到，Celery的API设计真的非常简单。然后，在Worker里就可以看到任务处理的消息：

[2013-08-27 19:20:23,363: WARNING/MainProcess] [email protected] ready.
[2013-08-27 19:20:23,367: INFO/MainProcess] consumer: Connected to redis://localhost:6379/0.
[2013-08-27 19:20:45,618: INFO/MainProcess] Got task from broker: tasks.sendmail[1a0a9262-7858-4192-9981-b7bf0ea7483b]
[2013-08-27 19:20:45,655: WARNING/PoolWorker-4] sending mail to [email protected]...
[2013-08-27 19:20:47,657: WARNING/PoolWorker-4] mail sent.
[2013-08-27 19:20:47,658: INFO/MainProcess] Task tasks.sendmail[1a0a9262-7858-4192-9981-b7bf0ea7483b] succeeded in 2.00266814232s: None

Celery默认设置就能满足基本要求。Worker以Pool模式启动，默认大小为CPU核心数量，缺省序列化机制是pickle，但可以指定为json。由于Python调用UNIX/Linux程序实在太容易，所以，用Celery作为异步任务框架非常合适