Change lookup of blobs to not use a subquery, which is murder on MySQL

data/model/blob.py

@@ -16,16 +16,16 @@ def get_repository_blob_by_digest(repository, blob_digest):
   """ Find the content-addressable blob linked to the specified repository.
   """
   try:
-    storage_id_query = (ImageStorage
-                        .select(ImageStorage.id)
-                        .join(Image)
-                        .where(Image.repository == repository,
-                               ImageStorage.content_checksum == blob_digest,
-                               ImageStorage.uploading == False)
-                        .limit(1))
+    storage = (ImageStorage
+               .select(ImageStorage.uuid)
+               .join(Image)
+               .where(Image.repository == repository,
+                      ImageStorage.content_checksum == blob_digest,
+                      ImageStorage.uploading == False)
+               .get())
 
-    return storage_model.get_storage_by_subquery(storage_id_query)
-  except InvalidImageException:
+    return storage_model.get_storage_by_uuid(storage.uuid)
+  except (ImageStorage.DoesNotExist, InvalidImageException):
     raise BlobDoesNotExist('Blob does not exist with digest: {0}'.format(blob_digest))
 
 
@@ -33,18 +33,18 @@ def get_repo_blob_by_digest(namespace, repo_name, blob_digest):
   """ Find the content-addressable blob linked to the specified repository.
   """
   try:
-    storage_id_query = (ImageStorage
-                        .select(ImageStorage.id)
-                        .join(Image)
-                        .join(Repository)
-                        .join(Namespace, on=(Namespace.id == Repository.namespace_user))
-                        .where(Repository.name == repo_name, Namespace.username == namespace,
-                               ImageStorage.content_checksum == blob_digest,
-                               ImageStorage.uploading == False)
-                        .limit(1))
+    storage = (ImageStorage
+               .select(ImageStorage.uuid)
+               .join(Image)
+               .join(Repository)
+               .join(Namespace, on=(Namespace.id == Repository.namespace_user))
+               .where(Repository.name == repo_name, Namespace.username == namespace,
+                      ImageStorage.content_checksum == blob_digest,
+                      ImageStorage.uploading == False)
+               .get())
 
-    return storage_model.get_storage_by_subquery(storage_id_query)
-  except InvalidImageException:
+    return storage_model.get_storage_by_uuid(storage.uuid)
+  except (ImageStorage.DoesNotExist, InvalidImageException):
     raise BlobDoesNotExist('Blob does not exist with digest: {0}'.format(blob_digest))
 
 

data/model/oci/blob.py

@@ -1,6 +1,6 @@
 from data.database import ImageStorage, ManifestBlob
 from data.model import BlobDoesNotExist
-from data.model.storage import get_storage_by_subquery, InvalidImageException
+from data.model.storage import get_storage_by_uuid, InvalidImageException
 from data.model.blob import get_repository_blob_by_digest as legacy_get
 
 def get_repository_blob_by_digest(repository, blob_digest):
@@ -8,16 +8,16 @@ def get_repository_blob_by_digest(repository, blob_digest):
       returns it or None if none.
   """
   try:
-    storage_id_query = (ImageStorage
-                        .select(ImageStorage.id)
-                        .join(ManifestBlob)
-                        .where(ManifestBlob.repository == repository,
-                               ImageStorage.content_checksum == blob_digest,
-                               ImageStorage.uploading == False)
-                        .limit(1))
+    storage = (ImageStorage
+               .select(ImageStorage.uuid)
+               .join(ManifestBlob)
+               .where(ManifestBlob.repository == repository,
+                      ImageStorage.content_checksum == blob_digest,
+                      ImageStorage.uploading == False)
+               .get())
 
-    return get_storage_by_subquery(storage_id_query)
-  except InvalidImageException:
+    return get_storage_by_uuid(storage.uuid)
+  except (ImageStorage.DoesNotExist, InvalidImageException):
     # TODO(jschorr): Remove once we are no longer using the legacy tables.
     # Try the legacy call.
     try:

data/model/storage.py

@@ -233,15 +233,6 @@ def _get_storage(query_modifier):
   return found
 
 
-def get_storage_by_subquery(subquery):
-  """ Returns the storage (and its locations) for the storage id returned by the subquery. The
-      subquery must return at most 1 result, which is a storage ID. """
-  def filter_by_subquery(query):
-    return query.where(ImageStorage.id == subquery)
-
-  return _get_storage(filter_by_subquery)
-
-
 def get_storage_by_uuid(storage_uuid):
   def filter_to_uuid(query):
     return query.where(ImageStorage.uuid == storage_uuid)

endpoints/v2/test/test_blob.py

@@ -50,7 +50,7 @@ def test_blob_caching(method, endpoint, client, app):
 
   with patch('endpoints.v2.blob.model_cache', InMemoryDataModelCache()):
     # First request should make a DB query to retrieve the blob.
-    with assert_query_count(3):
+    with assert_query_count(4):
       conduct_call(client, 'v2.' + endpoint, url_for, method, params, expected_code=200,
                    headers=headers)