fix how status is checked to avoid piling-on

oyster/client.py

@@ -22,6 +22,11 @@ class Client(object):
                                       size=mongo_log_maxsize)
         except pymongo.errors.CollectionInvalid:
             pass
+
+        # create status document if it doesn't exist
+        if self.db.status.count() == 0:
+            self.db.status.insert({'update_queue': 0})
+
         self._collection_name = 'fs'
         self.fs = gridfs.GridFS(self.db, self._collection_name)
         self.scraper = scrapelib.Scraper(user_agent=user_agent,
@@ -40,6 +45,7 @@ class Client(object):
         self.db.drop_collection('%s.chunks' % self._collection_name)
         self.db.drop_collection('%s.files' % self._collection_name)
         self.db.drop_collection('logs')
+        self.db.drop_collection('status')
 
 
     def log(self, action, url, error=False, **kwargs):

oyster/tasks.py

@@ -6,8 +6,6 @@ from celery.execute import send_task
 from oyster.client import Client
 
 client = Client()
-client.db.status.drop()
-client.db.status.insert({'update_queue': 0})
 
 
 @task(ignore_result=True)
@@ -31,7 +29,7 @@ class UpdateTaskScheduler(PeriodicTask):
         # (currently the only way we avoid duplicates)
         # alternate option would be to set a _queued flag on documents
         if client.db.status.find_one()['update_queue']:
-            pass
+            return
 
         next_set = client.get_update_queue()
         for doc in next_set: