任务中的芹菜任务

Question

我在我的应用程序中将 Dropbox 与芹菜相结合，这样我允许用户在连接 Dropbox 的情况下存储自己的照片。

我已经写了一段代码，但我担心这可能会导致无限循环，从而杀死系统。

我正在使用的 API 一次只允许 60 张照片，然后为您提供分页。

这是我 tasks.py 文件的副本 - 这实际上工作正常，但我想检查我是否在做正确的事情并且没有对系统产生太大影响。

class DropboxUsers(PeriodicTask):
    run_every = timedelta(hours=4)
    def run(self, **kwargs):
        logger = self.get_logger(**kwargs)
        logger.info("Collecting Dropbox users")
        dropbox_users = UserSocialAuth.objects.filter(provider='dropbox')
        for db in dropbox_users:
            ...
            ...
            ...
            sync_images.delay(first, second, third_argument)
        return True

@task(ignore_result=True)
def sync_images(token, secret, username):
    """docstring for sync_images"""
    logger = sync_images.get_logger()
    logger.info("Syncing images for %s" % username)
    ...
    ...
    ...
    ...
    feed = api.user_recent_media(user_id='self', count=60)
    images = feed[0]
    pagination = feed[1]
    for obj in images:
        ### STORE TO DROPBOX
        ...
        ...
        ...
        response = dropbox.put_file(f, my_picture, overwrite=True)
    ### CLOSE DB SESSION
    sess.unlink()
    if pagination:
        store_images.delay(first, second, third, fourth_argument)
@task(ignore_result=True)
def store_images(token, secret, username, max_id):
    """docstring for sync_images"""
    logger = store_images.get_logger()
    logger.info("Storing images for %s" % username)
    ...
    ...
    ...
    ...
    feed = api.user_recent_media(user_id='self', count=60, max_id=max_id)
    images = feed[0]
    try:
        pagination = feed[1]
    except:
        pagination = None
    for obj in images:
        ### STORE TO DROPBOX
        ...
        ...
        ...
        response = dropbox.put_file(f, my_picture, overwrite=True)
    ### CLOSE DB SESSION
    sess.unlink()
    if pagination:
        ### BASICALLY RESTART THE TASK WITH NEW ARGS
        store_images.delay(first, second, third, fourth_argument)
    return True

非常感谢您的专业知识。

Answer 1

我没有看到任何重大问题。我还实现了一个任务启动另一个任务的系统。

有一段时间，我在服务器重新启动时遇到芹菜复制任务的问题。我写了一个装饰器，它环绕一个任务，该任务使用缓存后端来确保具有相同参数的相同任务不会太频繁地运行。可能对您来说对冲无限循环很有用。

from django.core.cache import cache as _djcache
from django.utils.functional import wraps
class cache_task(object):
    """ Makes sure that a task is only run once over the course of a configurable
    number of seconds. Useful for tasks that get queued multiple times by accident,
    or on service restart, etc. Uses django's cache (memcache) to keep track."""
    def __init__(self, seconds=120, minutes=0, hours=0):
        self.cache_timeout_seconds = seconds + 60 * minutes + 60 * 60 * hours
    def __call__(self, task):
        task.unsynchronized_run = task.run
        @wraps(task.unsynchronized_run)
        def wrapper(*args, **kwargs):
            key = sha1(str(task.__module__) + str(task.__name__) + str(args) + str(kwargs)).hexdigest()
            is_cached = _djcache.get(key)
            if not is_cached:
                # store the cache BEFORE to cut down on race conditions caused by long tasks
                if self.cache_timeout_seconds:
                    _djcache.set(key, True, self.cache_timeout_seconds)
                task.unsynchronized_run(*args, **kwargs)
        task.run = wrapper
        return task

用法：

@cache_task(hours=2)
@task(ignore_result=True)
def store_images(token, secret, username, max_id):
   ...