From fe45797490fcb28c1b878b36184919a50a20af40 Mon Sep 17 00:00:00 2001
From: Joseph Schorr <joseph.schorr@coreos.com>
Date: Mon, 14 Jan 2019 13:34:25 -0500
Subject: [PATCH] Optimize the list active tags query

Some customers are hitting this endpoint rapidly for repositories with many, many tags. This change drops the unnecessary joins, which should reduce database load somewhat.
---
 data/model/tag.py                             | 24 ++++++++++++-------
 data/registry_model/registry_pre_oci_model.py |  8 ++++---
 2 files changed, 20 insertions(+), 12 deletions(-)

diff --git a/data/model/tag.py b/data/model/tag.py
index b3ed24adb..db9a3e9f1 100644
--- a/data/model/tag.py
+++ b/data/model/tag.py
@@ -204,18 +204,24 @@ def get_tag_manifest_digests(tags):
   return {manifest.tag_id: manifest.digest for manifest in manifests}
 
 
-def list_active_repo_tags(repo, start_id=None, limit=None):
+def list_active_repo_tags(repo, start_id=None, limit=None, include_images=True):
   """ Returns all of the active, non-hidden tags in a repository, joined to they images
       and (if present), their manifest.
   """
-  query = _tag_alive(RepositoryTag
-                     .select(RepositoryTag, Image, ImageStorage, TagManifest.digest)
-                     .join(Image)
-                     .join(ImageStorage)
-                     .where(RepositoryTag.repository == repo, RepositoryTag.hidden == False)
-                     .switch(RepositoryTag)
-                     .join(TagManifest, JOIN.LEFT_OUTER)
-                     .order_by(RepositoryTag.id))
+  if include_images:
+    query = _tag_alive(RepositoryTag
+                       .select(RepositoryTag, Image, ImageStorage, TagManifest.digest)
+                       .join(Image)
+                       .join(ImageStorage)
+                       .where(RepositoryTag.repository == repo, RepositoryTag.hidden == False)
+                       .switch(RepositoryTag)
+                       .join(TagManifest, JOIN.LEFT_OUTER)
+                       .order_by(RepositoryTag.id))
+  else:
+    query = _tag_alive(RepositoryTag
+                       .select(RepositoryTag)
+                       .where(RepositoryTag.repository == repo, RepositoryTag.hidden == False)
+                       .order_by(RepositoryTag.id))
 
   if start_id is not None:
     query = query.where(RepositoryTag.id >= start_id)
diff --git a/data/registry_model/registry_pre_oci_model.py b/data/registry_model/registry_pre_oci_model.py
index 4f64cce82..747f2d928 100644
--- a/data/registry_model/registry_pre_oci_model.py
+++ b/data/registry_model/registry_pre_oci_model.py
@@ -271,9 +271,11 @@ class PreOCIModel(SharedModel, RegistryDataInterface):
     operation on repositories with a lot of tags, and should be avoided for more targetted
     operations wherever possible.
     """
-    # NOTE: include_legacy_images isn't used here because `list_active_repo_tags` includes the
-    # information already, so we might as well just use it. However, the new model classes will
-    # *not* include it by default, so we make it a parameter now.
+    if not include_legacy_images:
+      tags = model.tag.list_active_repo_tags(repository_ref._db_id, start_pagination_id, limit,
+                                             include_images=False)
+      return [Tag.for_repository_tag(tag) for tag in tags]
+
     tags = model.tag.list_active_repo_tags(repository_ref._db_id, start_pagination_id, limit)
     return [Tag.for_repository_tag(tag,
                                    legacy_image=LegacyImage.for_image(tag.image),