【问题标题】:Django Celery tasks run twiceDjango Celery 任务运行两次
【发布时间】:2024-01-24 04:39:01
【问题描述】:

我的项目目录结构如下:

rss_reader  
 - reader  
    - __init__.py   
    - models.py  
    - views.py  
    - ...  
 - rss_reader  
    - __init__.py   
    - settings.py  
    - urls.py  
    - wsgi.py  
 - rss_contents  
    - __init__.py   
    - celery.py  
    - tasks.py  
    - config.py  
 - ...  

tasks.py

from __future__ import absolute_import
from rss_contents.celery import app

@app.task
def processArticle():
    print "100"

芹菜.py

from __future__ import absolute_import
from celery import Celery

app = Celery('rss_contents', include=['rss_contents.tasks'])
app.config_from_object('rss_contents.config')

config.py

from __future__ import absolute_import

CELERY_RESULT_BACKEND = 'redis://127.0.0.1:6379/5'
BROKER_URL = 'redis://127.0.0.1:6379/6'

阅读器> __init__.py

from rss_contents import tasks
tasks.processArticle.delay()

我的步骤如下:

  1. celery -A rss_contents worker -l 信息

控制台显示:

[2017-01-11 10:34:33,829: INFO/MainProcess] Connected to redis://127.0.0.1:6379/6  
[2017-01-11 10:34:33,855: INFO/MainProcess] mingle: searching for neighbors  
[2017-01-11 10:34:34,861: INFO/MainProcess] mingle: all alone  
[2017-01-11 10:34:34,892: WARNING/MainProcess] celery@DESKTOP-6KAT7MF ready.
  1. 运行服务器,在http://127.0.0.1:8000/启动开发服务器

但是celery控制台显示任务运行了两次:

[2017-01-11 10:41:20,910: INFO/MainProcess] Received task: rss_contents.tasks.processArticle[aa355c77-4ee8-4208-9e8c-915b110c7bbd]  
[2017-01-11 10:41:20,911: WARNING/Worker-1] 100
[2017-01-11 10:41:20,917: INFO/MainProcess] Task rss_contents.tasks.processArticle[aa355c77-4ee8-4208-9e8c-915b110c7bbd] succeeded in 0.00600004196167s: None  

[2017-01-11 10:41:22,430: INFO/MainProcess] Received task: rss_contents.tasks.processArticle[d92a151c-f0f9-4e8f-921a-fff2c1eb64c6]  
[2017-01-11 10:41:22,431: WARNING/Worker-1] 100   
[2017-01-11 10:41:22,447: INFO/MainProcess] Task rss_contents.tasks.processArticle[d92a151c-f0f9-4e8f-921a-fff2c1eb64c6] succeeded in 0.0160000324249s: None

为什么会运行两次?一次怎么办?

提前致谢。

【问题讨论】:

    标签: python django celery django-celery


    【解决方案1】:

    那是因为您在__init__.py 文件中定义了任务调用。

    我已经开始使用您的代码进行测试,并像这样在__init__.py 中插入一个 pdb 中断

    from celery_proj import tasks
    import pdb; pdb.set_trace()  # <-- make a break point here
    print __file__
    tasks.processArticle.delay()
    

    并获得以下兴趣发现:

    当您运行python manage.py runserver 时,会触发一次断点,并使用以下堆栈跟踪:

    (Pdb) where
      /Users/enix/Source/python/*/sotest/manage.py(10)<module>()
    -> execute_from_command_line(sys.argv)
      /Users/enix/Source/python/bae/lib/python2.7/site-packages/django/core/management/__init__.py(353)execute_from_command_line()
    -> utility.execute()
      /Users/enix/Source/python/bae/lib/python2.7/site-packages/django/core/management/__init__.py(316)execute()
    -> autoreload.check_errors(django.setup)()
      /Users/enix/Source/python/bae/lib/python2.7/site-packages/django/utils/autoreload.py(226)wrapper()
    -> fn(*args, **kwargs)
      /Users/enix/Source/python/bae/lib/python2.7/site-packages/django/__init__.py(18)setup()
    -> apps.populate(settings.INSTALLED_APPS)
      /Users/enix/Source/python/bae/lib/python2.7/site-packages/django/apps/registry.py(85)populate()
    -> app_config = AppConfig.create(entry)
      /Users/enix/Source/python/bae/lib/python2.7/site-packages/django/apps/config.py(90)create()
    -> module = import_module(entry)
      /System/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/importlib/__init__.py(37)import_module()
    -> __import__(name)
    > /Users/enix/Source/python/*/sotest/app/__init__.py(4)<module>()
    -> print __file__
    

    当我键入c 继续时,断点会再次触发并使用相同的堆栈跟踪。

    所以你的任务被调用了两次。但是我不太清楚为什么django会导入两次应用包。

    (bae)bogon:sotest enix$ python manage.py runserver
    /Users/enix/Source/python/*/sotest/app/__init__.py  # <-- print __file__
    /Users/enix/Source/python/*/sotest/app/__init__.pyc # <-- print __file__
    Performing system checks...
    System check identified no issues (0 silenced).
    January 10, 2017 - 21:44:12
    Django version 1.9.12, using settings 'sotest.settings'
    Starting development server at http://127.0.0.1:8000/
    Quit the server with CONTROL-C.
    

    也许你应该把你的任务调用移到其他地方再试一次。

    【讨论】:

    • 服务开始运行时可以在哪里调用task?
    【解决方案2】:

    @enix 把它钉在了头上。因为reader是一个app,所以在django初始化应用注册表的时候会调用一次。但是,由于此调用是动态进行的,因此阅读器应用程序可能会再次初始化,因为在某处全局使用阅读器包。在任何情况下,解决此问题的最简单方法就是将您的调用移动到一个类并从一个单例在任何 python 文件中调用它

    class AppInitializer(object):
        initialized = false
    
        @classmethod 
        def initialize(cls):
            if not cls.initialized:
                cls.initialized = True
                from rss_contents import tasks
                tasks.processArticle.delay()
    
    AppInitializer.initialize()
    

    如果您的 appengine 是多线程的,您需要在 initialize 周围添加一个方法锁。

    【讨论】: