Add a cleanup worker for blob uploads

data/model/blob.py

@@ -1,9 +1,10 @@
+from datetime import datetime
 from uuid import uuid4
 
 from data.model import (tag, _basequery, BlobDoesNotExist, InvalidBlobUpload, db_transaction,
                         storage as storage_model, InvalidImageException)
 from data.database import (Repository, Namespace, ImageStorage, Image, ImageStoragePlacement,
-                           BlobUpload, ImageStorageLocation)
+                           BlobUpload, ImageStorageLocation, db_random_func)
 
 
 def get_repo_blob_by_digest(namespace, repo_name, blob_digest):
@@ -58,6 +59,35 @@ def store_blob_record_and_temp_link(namespace, repo_name, blob_digest, location_
   return storage
 
 
+def get_stale_blob_upload(stale_timespan):
+  """ Returns a random blob upload which was created before the stale timespan. """
+  stale_threshold = datetime.now() - stale_timespan
+
+  try:
+    candidates = (BlobUpload
+                  .select()
+                  .where(BlobUpload.created <= stale_threshold)
+                  .limit(500)
+                  .distinct()
+                  .alias('candidates'))
+
+    found = (BlobUpload
+             .select(candidates.c.id)
+             .from_(candidates)
+             .order_by(db_random_func())
+             .get())
+    if not found:
+      return None
+
+    return (BlobUpload
+            .select(BlobUpload, ImageStorageLocation)
+            .join(ImageStorageLocation)
+            .where(BlobUpload.id == found.id)
+            .get())
+  except BlobUpload.DoesNotExist:
+    return None
+
+
 def get_blob_upload(namespace, repo_name, upload_uuid):
   """ Load the upload which is already in progress.
   """