Make sure GC checks new Blob table as well before deleting CAS storage

test/test_gc.py

@@ -8,7 +8,7 @@ from playhouse.test_utils import assert_query_count
 from app import app, storage
 from initdb import setup_database_for_testing, finished_database_for_testing
 from data import model, database
-from data.database import Image, ImageStorage, DerivedStorageForImage, Label, TagManifestLabel
+from data.database import Image, ImageStorage, DerivedStorageForImage, Label, TagManifestLabel, Blob
 
 
 ADMIN_ACCESS_USER = 'devtable'
@@ -190,6 +190,9 @@ class TestGarbageCollection(unittest.TestCase):
       if storage_row.cas_path:
         storage.get_content({preferred}, storage.blob_path(storage_row.content_checksum))
 
+    for blob_row in Blob.select():
+        storage.get_content({preferred}, storage.blob_path(blob_row.digest))
+
   def test_has_garbage(self):
     """ Remove all existing repositories, then add one without garbage, check, then add one with
         garbage, and check again.
@@ -502,7 +505,48 @@ class TestGarbageCollection(unittest.TestCase):
       # Ensure the CAS path still exists.
       self.assertTrue(storage.exists({preferred}, storage.blob_path(digest)))
 
+  def test_images_shared_cas_with_new_blob_table(self):
+    """ A repository with a tag and image that shares its CAS path with a record in the new Blob
+        table. Deleting the first tag should delete the first image, and its storage, but not the
+        file in storage, as it shares its CAS path with the blob row.
+    """
+    with self.assert_gc_integrity(expect_storage_removed=True):
+      repository = self.createRepository()
 
+      # Create two image storage records with the same content checksum.
+      content = 'hello world'
+      digest = 'sha256:' + hashlib.sha256(content).hexdigest()
+      preferred = storage.preferred_locations[0]
+      storage.put_content({preferred}, storage.blob_path(digest), content)
+
+      media_type = database.MediaType.get(name='text/plain')
+
+      is1 = database.ImageStorage.create(content_checksum=digest, uploading=False)
+      is2 = database.Blob.create(digest=digest, size=0, media_type=media_type)
+
+      location = database.ImageStorageLocation.get(name=preferred)
+      database.ImageStoragePlacement.create(location=location, storage=is1)
+
+      # Ensure the CAS path exists.
+      self.assertTrue(storage.exists({preferred}, storage.blob_path(digest)))
+
+      # Create the image in the repository, and the tag.
+      first_image = Image.create(docker_image_id='i1',
+                                 repository=repository, storage=is1,
+                                 ancestors='/')
+
+      model.tag.store_tag_manifest(repository.namespace_user.username, repository.name,
+                                   'first', first_image.docker_image_id,
+                                   'sha:someshahere1', '{}')
+
+      self.assertNotDeleted(repository, 'i1')
+
+      # Delete the tag.
+      self.deleteTag(repository, 'first')
+      self.assertDeleted(repository, 'i1')
+
+      # Ensure the CAS path still exists, as it is referenced by the Blob table.
+      self.assertTrue(storage.exists({preferred}, storage.blob_path(digest)))
 
 if __name__ == '__main__':
   unittest.main()