0、目的
在开发项目中，经常有一些操作时间比较长（生产环境中超过了nginx的timeout时间），或者是间隔一段时间就要执行的任务。
在这种情况下，使用celery就是一个很好的选择。
celery是一个异步任务队列/基于分布式消息传递的作业队列。
Celery通过消息（message）进行通信，使用代理（broker）在客户端和工作执行者之间进行交互。
当开始一个任务时，客户端发送消息到队列并由代理将其发往响应的工作执行者处。
准备使用redis作为消息代理（broker），Django数据库作为结果存储（ResultStore）。
1、安装
redis：
windows：
linux：
yum install redis-server
PS：需要在cmd中运行，不能再powercmd。很奇怪。
pip install celery
pip install celery-with-redis
pip install django-celery
2、django代码（whthas_home为project，portal为app）
修改代码，whthas_home/__init__.py
1 from __future__ import absolute_import
2
3 from .celery import app as celery_app
修改代码，whthas_home/setting.py
1 # Celery settings
2 import djcelery
3 djcelery.setup_loader()
4
5 BROKER_URL = 'redis://127.0.0.1:6379/0'
6 CELERY_RESULT_BACKEND = 'redis://127.0.0.1:6379/0' ##加密方式CELERY_RESULT_BACKEND = 'redis://:密码@127.0.0.1:6379/0'
7 CELERY_ACCEPT_CONTENT = ['json']
8 CELERY_TASK_SERIALIZER = 'json'
9 CELERY_RESULT_SERIALIZER = 'json'
14
15 INSTALLED_APPS = [
16 'django.contrib.auth',
17 'django.contrib.contenttypes',
18 'django.contrib.sessions',
19 'django.contrib.messages',
20 'django.contrib.staticfiles',
21 'suit',
22 'django.contrib.admin',
23 'DjangoUeditor',
24 'portal',
25 'djcelery',
26 ]
增加文件， whthas_home/celery.py
1 from __future__ import absolute_import
2
3 import os
4 from celery import Celery
5 from django.conf import settings
6
7 os.environ.setdefault('DJANGO_SETTINGS_MODULE', 'whthas_home.settings')
8
9 app = Celery('portal')
10
11 app.config_from_object('django.conf:settings')
12 app.autodiscover_tasks(lambda: settings.INSTALLED_APPS)
13
14
15 @app.task(bind=True)
16 def debug_task(self):
17 print('Request: {0!r}'.format(self.request))
增加文件，portal/tasks.py
1 from celery import task
2 from time import sleep
3
4
5 @task()
6 def Task_A(message):
7 Task_A.update_state(state='PROGRESS', meta={'progress': 0})
8 sleep(10)
9 Task_A.update_state(state='PROGRESS', meta={'progress': 30})
10 sleep(10)
11 return message
12
13
14 def get_task_status(task_id):
15 task = Task_A.AsyncResult(task_id)
16
17 status = task.state
18 progress = 0
19
20 if status == u'SUCCESS':
21 progress = 100
22 elif status == u'FAILURE':
23 progress = 0
24 elif status == 'PROGRESS':
25 progress = task.info['progress']
26
27 return {'status': status, 'progress': progress}
3、测试
启动broker：python manage.py celeryd -l info
进入shell：python manage.py shell，测试程序是否正常。
>>> from portal.tasks import *
>>> t = TaskA.delay("heel2")
>>> get_task_status(t.id)
{'status': u'PROGRESS', 'progress': 0}
>>> get_task_status(t.id)
{'status': u'PROGRESS', 'progress': 0}
>>> get_task_status(t.id)
{'status': u'PROGRESS', 'progress': 30}
>>> get_task_status(t.id)
{'status': u'PROGRESS', 'progress': 30}
>>> get_task_status(t.id)
{'status': u'SUCCESS', 'progress': 100}
>>>
同时broker侧能看到：
[2017-04-21 16:38:47,023: INFO/MainProcess] Received task: portal.tasks.Task_A[da948495-c64b-4ff9-882b-876721cd5017]
[2017-04-21 16:39:07,035: INFO/MainProcess] Task portal.tasks.Task_A[da948495-c64b-4ff9-882b-876721cd5017] succeeded in 20.0099999905s: heel
表示代码能正常运行。
使用redis-client看到任务状态：
127.0.0.1:6379> get "celery-task-meta-da948495-c64b-4ff9-882b-876721cd5017"
"{\"status\": \"SUCCESS\", \"traceback\": null, \"result\": \"heel\", \"children\": []}"
相关redis命令：keys *
4、django后台定义任务
进入后台
设置任务
5、执行任务
启动broker：python manage.py celeryd -l info
因为这里是个定时任务，所以还需要启动心跳 ：python manage.py celery beat
broker侧能看到：
[2017-04-21 16:56:33,216: INFO/MainProcess] Received task: portal.tasks.Task_A[d8a26977-8413-4bf0-b518-b53052af4cee]
[2017-04-21 16:56:53,211: INFO/MainProcess] Received task: portal.tasks.Task_A[00f1cab7-eb56-4cc3-9979-a8a68aaaf2de]
[2017-04-21 16:56:53,220: INFO/MainProcess] Task portal.tasks.Task_A[d8a26977-8413-4bf0-b518-b53052af4cee] succeeded in 20.003000021s: heel2
[2017-04-21 16:57:13,211: INFO/MainProcess] Received task: portal.tasks.Task_A[aa652612-8525-4110-94ea-9010085ec20b]
[2017-04-21 16:57:13,223: INFO/MainProcess] Task portal.tasks.Task_A[00f1cab7-eb56-4cc3-9979-a8a68aaaf2de] succeeded in 20.0080001354s: heel2
[2017-04-21 16:57:33,213: INFO/MainProcess] Received task: portal.tasks.Task_A[9876890e-6a71-4501-bdae-775492ebae88]
[2017-04-21 16:57:33,219: INFO/MainProcess] Task portal.tasks.Task_A[aa652612-8525-4110-94ea-9010085ec20b] succeeded in 20.0050001144s: heel2
[2017-04-21 16:57:53,211: INFO/MainProcess] Received task: portal.tasks.Task_A[c12fcffc-3910-4a22-93b3-0df740910728]
[2017-04-21 16:57:53,221: INFO/MainProcess] Task portal.tasks.Task_A[9876890e-6a71-4501-bdae-775492ebae88] succeeded in 20.0069999695s: heel2
[2017-04-21 16:58:13,211: INFO/MainProcess] Received task: portal.tasks.Task_A[ccfad575-c0b4-48f5-9385-85ff5dac76fc]
[2017-04-21 16:58:13,217: INFO/MainProcess] Task portal.tasks.Task_A[c12fcffc-3910-4a22-93b3-0df740910728] succeeded in 20.003000021s: heel2
[2017-04-21 16:58:33,211: INFO/MainProcess] Received task: portal.tasks.Task_A[0d6f77a6-a29c-4ead-9428-e3e758c754e1]
[2017-04-21 16:58:33,221: INFO/MainProcess] Task portal.tasks.Task_A[ccfad575-c0b4-48f5-9385-85ff5dac76fc] succeeded in 20.007999897s: heel2
[2017-04-21 16:58:53,211: INFO/MainProcess] Received task: portal.tasks.Task_A[34b2b3a0-771c-4a28-94a1-92e25763fae1]
[2017-04-21 16:58:53,217: INFO/MainProcess] Task portal.tasks.Task_A[0d6f77a6-a29c-4ead-9428-e3e758c754e1] succeeded in 20.0039999485s: heel2
[2017-04-21 16:59:13,211: INFO/MainProcess] Received task: portal.tasks.Task_A[d4be920f-376a-46a2-9edd-095234d29ef2]
[2017-04-21 16:59:13,217: INFO/MainProcess] Task portal.tasks.Task_A[34b2b3a0-771c-4a28-94a1-92e25763fae1] succeeded in 20.0039999485s: heel2
可见：设置定时任务成功。
参考资料：
1、https://gist.github.com/tyrchen/1436486，Django-celery + Redis notes
2、http://www.cnblogs.com/aguncn/p/4947092.html，django celery redis简单测试
3、https://my.oschina.net/kinegratii/blog/292395?fromerr=2lvw3H0L，djcelery入门：实现运行定时任务
另一个教程： https://www.cnblogs.com/alex3714/p/6351797.html
二：
如果有耗时比较长的任务，也可以用celery先返回一个ID，然后在页面中用ajax拿着id再取结果。