Fixed critical issue in janitor

fractalis/cleanup.py

@@ -9,21 +9,24 @@ def janitor():
     file system while Fractalis is running.
     """
     data_dir = os.path.join(app.config['FRACTALIS_TMP_DIR'], 'data')
+    tracked_ids = [key.split(':')[1] for key in redis.scan_iter('data:*')]
     if not os.path.exists(data_dir):
-        for key in redis.scan_iter('data:*'):
-            redis.delete(key)
+        for task_id in tracked_ids:
+            async_result = celery.AsyncResult(task_id)
+            if async_result.state == 'SUCCESS':
+                redis.delete('data:{}'.format(task_id))
         return
+
     cached_files = [f for f in os.listdir(data_dir)
                     if os.path.isfile(os.path.join(data_dir, f))]
-    tracked_files = [key.split(':')[1] for key in redis.scan_iter('data:*')]
 
     # clean cached files
     for cached_file in cached_files:
-        if cached_file not in tracked_files:
+        if cached_file not in tracked_ids:
             sync.remove_file(os.path.join(data_dir, cached_file))
 
     # clean tracked files
-    for task_id in tracked_files:
+    for task_id in tracked_ids:
         path = os.path.join(data_dir, task_id)
         async_result = celery.AsyncResult(task_id)
         if async_result.state == 'SUCCESS' and not os.path.exists(path):

fractalis/data/controller.py

@@ -79,8 +79,7 @@ def get_all_data() -> Tuple[Response, int]:
     for task_id in session['data_tasks']:
         data_state = get_data_state_for_task_id(task_id, wait)
         if data_state is None:
-            warning = "Data state with task_id '{}' expired. " \
-                      "Discarding...".format(task_id)
+            warning = "Data state with task_id '{}' expired.".format(task_id)
             logger.warning(warning)
             continue
         # remove internal information from response