Add a check that will fail if we try to mislink V1 layers

Also logs some useful information

data/model/image.py

@@ -133,7 +133,8 @@ def invert_placement_query_results(placement_query):
 
 def lookup_repository_images(repo, docker_image_ids):
   return (Image
-          .select()
+          .select(Image, ImageStorage)
+          .join(ImageStorage) # TODO(jschorr): Remove once no longer needed in v2/manifest.py.
           .where(Image.repository == repo, Image.docker_image_id << docker_image_ids))
 
 

endpoints/v2/manifest.py

@@ -4,7 +4,7 @@ import json
 import features
 
 from peewee import IntegrityError
-from flask import make_response, request, url_for
+from flask import make_response, request, url_for, abort
 from collections import namedtuple, OrderedDict
 from jwkest.jws import SIGNER_ALGS, keyrep
 from datetime import datetime
@@ -373,6 +373,13 @@ def _write_manifest(namespace, repo_name, manifest):
 
     # If there is already a V1 image for this layer, nothing more to do.
     if v1_mdata.docker_id in images_map:
+      # Ensure that the V1 image's storage matches the V2 blob. If not, we've found
+      # a data inconsistency and need to fail.
+      v1_image = images_map[v1_mdata.docker_id]
+      if v1_image.storage.content_checksum != digest_str:
+        logger.error('Checksum mismatch on V1 layer %s (#%s): Expected digest %s, found %s',
+                     v1_mdata.docker_id, v1_image.id, digest_str, v1_image.storage.content_checksum)
+        abort(500)
       continue
 
     # Lookup the parent image for the layer, if any.