First stab at time machine using fixed two week expiration policy.

config.py

@@ -185,5 +185,8 @@ class DefaultConfig(object):
   LOG_ARCHIVE_LOCATION = 'local_us'
   LOG_ARCHIVE_PATH = 'logarchive/'
 
+  # Number of revisions to keep expired tags
+  TIME_MACHINE_DELTA_SECONDS = 14 * 24 * 60 * 60
+
   # For enterprise:
   MAXIMUM_REPOSITORY_USAGE = 20

data/database.py

@@ -456,12 +456,14 @@ class RepositoryTag(BaseModel):
   name = CharField()
   image = ForeignKeyField(Image)
   repository = ForeignKeyField(Repository)
+  lifetime_start = DateTimeField(default=datetime.utcnow)
+  lifetime_end = DateTimeField(null=True)
 
   class Meta:
     database = db
     read_slaves = (read_slave,)
     indexes = (
-      (('repository', 'name'), True),
+      (('repository', 'name'), False),
     )
 
 

data/model/legacy.py

@@ -106,12 +106,15 @@ class TooManyLoginAttemptsException(Exception):
     self.retry_after = retry_after
 
 
-def _get_repository(namespace_name, repository_name):
-  return (Repository
-          .select(Repository, Namespace)
-          .join(Namespace, on=(Repository.namespace_user == Namespace.id))
-          .where(Namespace.username == namespace_name, Repository.name == repository_name)
-          .get())
+def _get_repository(namespace_name, repository_name, for_update=False):
+  query = (Repository
+           .select(Repository, Namespace)
+           .join(Namespace, on=(Repository.namespace_user == Namespace.id))
+           .where(Namespace.username == namespace_name, Repository.name == repository_name))
+  if for_update:
+    query = db_for_update(query)
+
+  return query.get()
 
 
 def hash_password(password, salt=None):
@@ -1534,12 +1537,27 @@ def list_repository_tags(namespace_name, repository_name):
           .join(Namespace, on=(Repository.namespace_user == Namespace.id))
           .switch(RepositoryTag)
           .join(Image)
-          .where(Repository.name == repository_name, Namespace.username == namespace_name))
+          .where(Repository.name == repository_name, Namespace.username == namespace_name,
+                 RepositoryTag.lifetime_end >> None))
+
+
+def _garbage_collect_tags(namespace_name, repository_name):
+  with config.app_config['DB_TRANSACTION_FACTORY'](db):
+    repo = _get_repository(namespace_name, repository_name)
+    collect_time = (datetime.utcnow() -
+                    timedelta(seconds=config.app_config['TIME_MACHINE_DELTA_SECONDS']))
+
+    (RepositoryTag
+     .delete()
+     .where(RepositoryTag.repository == repo, RepositoryTag.lifetime_end < collect_time)
+     .execute())
 
 
 def garbage_collect_repository(namespace_name, repository_name):
   storage_id_whitelist = {}
 
+  _garbage_collect_tags(namespace_name, repository_name)
+
   with config.app_config['DB_TRANSACTION_FACTORY'](db):
     # TODO (jake): We could probably select this and all the images in a single query using
     # a different kind of join.
@@ -1573,12 +1591,12 @@ def garbage_collect_repository(namespace_name, repository_name):
 
   if len(to_remove) > 0:
     logger.info('Garbage collecting storage for images: %s', to_remove)
-    garbage_collect_storage(storage_id_whitelist)
+    _garbage_collect_storage(storage_id_whitelist)
 
   return len(to_remove)
 
 
-def garbage_collect_storage(storage_id_whitelist):
+def _garbage_collect_storage(storage_id_whitelist):
   if len(storage_id_whitelist) == 0:
     return
 
@@ -1710,45 +1728,57 @@ def get_parent_images(namespace_name, repository_name, image_obj):
 
 def create_or_update_tag(namespace_name, repository_name, tag_name,
                          tag_docker_image_id):
-  try:
-    repo = _get_repository(namespace_name, repository_name)
-  except Repository.DoesNotExist:
-    raise DataModelException('Invalid repository %s/%s' % (namespace_name, repository_name))
 
-  try:
-    image = Image.get(Image.docker_image_id == tag_docker_image_id, Image.repository == repo)
-  except Image.DoesNotExist:
-    raise DataModelException('Invalid image with id: %s' % tag_docker_image_id)
+  with config.app_config['DB_TRANSACTION_FACTORY'](db):
+    try:
+      repo = _get_repository(namespace_name, repository_name)
+    except Repository.DoesNotExist:
+      raise DataModelException('Invalid repository %s/%s' % (namespace_name, repository_name))
 
-  try:
-    tag = RepositoryTag.get(RepositoryTag.repository == repo, RepositoryTag.name == tag_name)
-    tag.image = image
-    tag.save()
-  except RepositoryTag.DoesNotExist:
-    tag = RepositoryTag.create(repository=repo, image=image, name=tag_name)
+    try:
+      image = Image.get(Image.docker_image_id == tag_docker_image_id, Image.repository == repo)
+    except Image.DoesNotExist:
+      raise DataModelException('Invalid image with id: %s' % tag_docker_image_id)
 
-  return tag
+    now = datetime.utcnow()
+
+    try:
+      # When we move a tag, we really end the timeline of the old one and create a new one
+      tag = RepositoryTag.get(RepositoryTag.repository == repo, RepositoryTag.name == tag_name,
+                              RepositoryTag.lifetime_end >> None)
+      tag.lifetime_end = now
+      tag.save()
+    except RepositoryTag.DoesNotExist:
+      # No tag that needs to be ended
+      pass
+
+    tag = RepositoryTag.create(repository=repo, image=image, name=tag_name, lifetime_start=now)
+
+    return tag
 
 
 def delete_tag(namespace_name, repository_name, tag_name):
-  try:
-    found = (RepositoryTag
-             .select()
-             .join(Repository)
-             .join(Namespace, on=(Repository.namespace_user == Namespace.id))
-             .where(Repository.name == repository_name, Namespace.username == namespace_name,
-                    RepositoryTag.name == tag_name)
-             .get())
+  with config.app_config['DB_TRANSACTION_FACTORY'](db):
+    try:
+      query = (RepositoryTag
+               .select(RepositoryTag, Repository)
+               .join(Repository)
+               .join(Namespace, on=(Repository.namespace_user == Namespace.id))
+               .where(Repository.name == repository_name, Namespace.username == namespace_name,
+                      RepositoryTag.name == tag_name, RepositoryTag.lifetime_end >> None))
+      found = db_for_update(query).get()
 
-  except RepositoryTag.DoesNotExist:
-    msg = ('Invalid repository tag \'%s\' on repository \'%s/%s\'' %
-           (tag_name, namespace_name, repository_name))
-    raise DataModelException(msg)
+    except RepositoryTag.DoesNotExist:
+      msg = ('Invalid repository tag \'%s\' on repository \'%s/%s\'' %
+             (tag_name, namespace_name, repository_name))
+      raise DataModelException(msg)
 
-  found.delete_instance()
+    found.lifetime_end = datetime.utcnow()
+    found.save()
 
 
-def delete_all_repository_tags(namespace_name, repository_name):
+def purge_all_repository_tags(namespace_name, repository_name):
+  """ Immediately purge all repository tags without respecting the lifeline procedure """
   try:
     repo = _get_repository(namespace_name, repository_name)
   except Repository.DoesNotExist:
@@ -1863,7 +1893,7 @@ def set_team_repo_permission(team_name, namespace_name, repository_name,
 
 def purge_repository(namespace_name, repository_name):
   # Delete all tags to allow gc to reclaim storage
-  delete_all_repository_tags(namespace_name, repository_name)
+  purge_all_repository_tags(namespace_name, repository_name)
 
   # Gc to remove the images and storage
   garbage_collect_repository(namespace_name, repository_name)

endpoints/api/tag.py

@@ -54,8 +54,8 @@ class RepositoryTag(RepositoryParamResource):
 
     username = get_authenticated_user().username
     log_action('move_tag' if original_image_id else 'create_tag', namespace,
-               { 'username': username, 'repo': repository, 'tag': tag,
-                 'image': image_id, 'original_image': original_image_id },
+               {'username': username, 'repo': repository, 'tag': tag,
+                'image': image_id, 'original_image': original_image_id},
                repo=model.get_repository(namespace, repository))
 
     return 'Updated', 201

test/test_gc.py

@@ -207,7 +207,7 @@ class TestGarbageColection(unittest.TestCase):
     self.assertNotDeleted(repository,'i1', 'i2', 'i3', 't1', 't2', 't3', 'f1', 'f2')
 
   def test_gc_storage_empty(self):
-    model.garbage_collect_storage(set())
+    model._garbage_collect_storage(set())
 
 if __name__ == '__main__':
   unittest.main()