Move the creation of images to when the JSON is uploaded.

data/model/legacy.py

@@ -1628,8 +1628,6 @@ def garbage_collect_repository(namespace_name, repository_name):
     logger.info('Garbage collecting storage for images: %s', to_remove)
     _garbage_collect_storage(storage_id_whitelist)
 
-  return len(to_remove)
-
 
 def _garbage_collect_storage(storage_id_whitelist):
   if len(storage_id_whitelist) == 0:

endpoints/index.py

@@ -222,32 +222,6 @@ def create_repository(namespace, repository):
     repo = model.create_repository(namespace, repository,
                                    get_authenticated_user())
 
-  logger.debug('Determining already added images')
-  added_images = OrderedDict([(desc['id'], desc) for desc in image_descriptions])
-  new_repo_images = dict(added_images)
-
-  # Optimization: Lookup any existing images in the repository with matching docker IDs and
-  # remove them from the added dict, so we don't need to look them up one-by-one.
-  def chunks(l, n):
-    for i in xrange(0, len(l), n):
-      yield l[i:i+n]
-
-  # Note: We do this in chunks in an effort to not hit the SQL query size limit.
-  for chunk in chunks(new_repo_images.keys(), 50):
-    existing_images = model.lookup_repository_images(namespace, repository, chunk)
-    for existing in existing_images:
-      added_images.pop(existing.docker_image_id)
-
-  logger.debug('Creating/Linking necessary images')
-  username = get_authenticated_user() and get_authenticated_user().username
-  translations = {}
-  for image_description in added_images.values():
-    model.find_create_or_link_image(image_description['id'], repo, username,
-                                    translations, storage.preferred_locations[0])
-
-
-  logger.debug('Created images')
-  track_and_log('push_repo', repo)
   return make_response('Created', 201)
 
 
@@ -280,7 +254,7 @@ def update_images(namespace, repository):
       event.publish_event_data('docker-cli', user_data)
 
     logger.debug('GCing repository')
-    num_removed = model.garbage_collect_repository(namespace, repository)
+    model.garbage_collect_repository(namespace, repository)
 
     # Generate a job for each notification that has been added to this repo
     logger.debug('Adding notifications for repository')
@@ -288,8 +262,8 @@ def update_images(namespace, repository):
     updated_tags = session.get('pushed_tags', {})
     event_data = {
       'updated_tags': updated_tags,
-      'pruned_image_count': num_removed
     }
+    track_and_log('push_repo', repo)
     spawn_notification(repo, 'repo_push', event_data)
     return make_response('Updated', 204)
 

endpoints/registry.py

@@ -9,6 +9,7 @@ from time import time
 
 from app import storage as store, image_diff_queue, app
 from auth.auth import process_auth, extract_namespace_repo_from_session
+from auth.auth_context import get_authenticated_user
 from util import checksums, changes
 from util.http import abort, exact_abort
 from auth.permissions import (ReadRepositoryPermission,
@@ -456,9 +457,15 @@ def put_image_json(namespace, repository, image_id):
   logger.debug('Looking up repo image')
   repo_image = model.get_repo_image_extended(namespace, repository, image_id)
   if not repo_image:
-    logger.debug('Image not found')
-    abort(404, 'Image %(image_id)s not found', issue='unknown-image',
-          image_id=image_id)
+    logger.debug('Image not found, creating image')
+    repo = model.get_repository(namespace, repository)
+    if repo is None:
+      abort(404, 'Repository does not exist: %(namespace)s/%(repository)s', issue='no-repo',
+            namespace=namespace, repository=repository)
+
+    username = get_authenticated_user() and get_authenticated_user().username
+    repo_image = model.find_create_or_link_image(image_id, repo, username, {},
+                                                 store.preferred_locations[0])
 
   uuid = repo_image.storage.uuid