Django中使用Celery

• 安装

  pipenv install django-celery

• 主配置文件导入celery配置

  # settings.py
  
  from .celeryconfig import *
  BROKER_BECKEND = 'redis'
  BROKER_URL = 'redis://192.168.2.128:6379/1'
  CELERY_RESULT_BACKEND = 'redis://192.168.2.128:6379/2'

• celery配置文件

  #celeryconfig.py
  
  from datetime import timedelta
  
  import djcelery
  # setup_loader用于扫描所有app下tasks.py文件中的task
  djcelery.setup_loader()
  
  # 设置不同的任务队列，将普通任务和定时任务分开,redis中实际key如下
  # 1) "_kombu.binding.work_queue"
  # 2) "_kombu.binding.beat_tasks"
  CELERY_QUEUES = {
      'beat_tasks': {
          'exchange': 'beat_tasks',
          'exchange_type': 'direct',
          'binding_key': 'beat_tasks'
      },
      'work_queue': {
          'exchange': 'work_queue',
          'exchange_type': 'direct',
          'binding_key': 'work_queue'
      },
  }
  
  # 设置默认队列
  CELERY_DEFAULT_QUEUE = 'work_queue'
  
  CELERY_IMPORTS = (
      'course.tasks',
  )
  
  # 有些情况下可以防止死锁
  CELERY_FORCE_EXECV = True
  
  # 设置并发的worker数量
  CELERY_CONCURRENCY = 4
  
  # 允许重试
  CELERY_ACKS_LATE = True
  
  # 每个worker最多执行100个任务被销毁，可以防止内存泄漏
  CELERY_MAX_TASKS_PER_CHILD = 100
  
  # 单个任务的最大运行时间
  CELERY_TASK_TIME_LIMIT = 12 * 30
  
  # 定时任务配置
  CELERYBEAT_SCHEDULE = {
      'task1': {
          'task': 'course-task',
          'schedule': timedelta(seconds=5),
          'options': {
              'queue': 'beat_tasks'
          }
      }
  }
  

• app中创建tasks.py并创建task

  # (app)course->tasks.py
  
  import time
  from celery.task import Task
  
  
  class CourseTask(Task):
      name = 'course-task'
  
      def run(self, *args, **kwargs):
          print('start course task')
          time.sleep(4)
          print('args={}, kwargs={}'.format(args, kwargs))
          print('end course task')

• app视图函数调用task

  # (app)course->views.py
  
  from django.http import JsonResponse
  from course.tasks import CourseTask
  
  
  def do(request):
      # 执行异步任务
      print('start do request')
      # CourseTask.delay()
      CourseTask.apply_async(args=('hello',), queue='work_queue')
      print('end do request')
      return JsonResponse({'result': 'ok'})

• 启动命令

  python manage.py celery worker -l INFO
  python manage.py celery beat -l INFO