Celery 在将我的查询集 obj 作为参数传递时引发错误

Question

我试图执行一个周期性任务，所以我使用 celery 与 Django 1.8 和 Django Rest Framework 和 Postgres 作为数据库。当我尝试将 obj 发送到任务时，我得到TypeError: foreign_model_obj is not JSON serializable. 如何将我的查询集对象传递给我的任务。

视图.py：

class MyModelCreateApiView(generics.CreateAPIView):
    queryset = MyModel.objects.all()
    serializer_class = MyModelSerializer
    authentication_classes = (TokenAuthentication,)

    def create(self, request, *args, **kwargs):
        data = dict()
        data['foreign_model_id'] = kwargs['pk']
        foreign_model_obj = MyForeignModel.objects.get(id=data['foreign_model_id'])

        obj = MyModel.objects.create(**data)
        result = serialize_query(MyModel, {"id": obj.id})
        local_time = foreign_model_obj.time
        my_celery_task.apply_async([foreign_model_obj], eta=local_time)
        return Response(result)

任务.py：

@celery_app.task(name="my_celery_task")
def my_first_celery_task(mymodel_obj):
    # ... updating obj attributes
    mymodel_obj.save()

Answer 1

您只需发送id实例的 并检索任务中的对象。传递实例是一种不好的做法，因为它可以同时更改，特别是您正在执行任务时似乎延迟了。

视图.py：

class MyModelCreateApiView(generics.CreateAPIView):
    queryset = MyModel.objects.all()
    serializer_class = MyModelSerializer
    authentication_classes = (TokenAuthentication,)

    def create(self, request, *args, **kwargs):
        data = dict()
        data['foreign_model_id'] = kwargs['pk']
        foreign_model_obj = MyForeignModel.objects.get(id=data['foreign_model_id'])

        obj = MyModel.objects.create(**data)
        result = serialize_query(MyModel, {"id": obj.id})
        local_time = foreign_model_obj.time
        my_celery_task.apply_async([foreign_model_obj.id], eta=local_time) # send only the obj id
        return Response(result)

任务.py：

@celery_app.task(name="my_celery_task")
def my_celery_task(mymodel_obj_id):
    my_model_obj = MyModel.objects.get(id=mymodel_obj_id) # retrieve your object here
    # ... updating obj attributes
    mymodel_obj.save()

Answer 2

实际上，恕我直言，最好的方法是获取查询集的可腌制组件，然后在任务中重新生成查询集（https://docs.djangoproject.com/en/1.9/ref/models/querysets/）：

import pickle
query = pickle.loads(s)     # Assuming 's' is the pickled string.
qs = MyModel.objects.filter(a__in=[1,2,3]) # whatever you want here...
querystr = pickle.dumps(qs.query)      # pickle the queryset
my_celery_task.apply_async(querystr, eta=local_time) # send only the string...

任务：

@celery_app.task(name="my_celery_task")
def my_celery_task(querystr):
    my_model_objs = MyModel.objects.all()
    my_model_objs.query = pickle.loads(querystr) # Restore the queryset
    # ... updating obj attributes
    item=my_model_objs[0]

我认为这是最好的方法，因为查询将在任务中执行（可能是第一次），防止各种计时问题，它不需要在调用者中执行（因此查询不会加倍）。