def test_manifest_backfill_broken_tag(clear_rows, initialized_db):
"""
Tests backfilling a broken tag.
"""
# Delete existing tag manifest so we can reuse the tag.
TagManifestLabel.delete().execute()
TagManifest.delete().execute()
# Create a tag with an image referenced missing parent images.
repo = model.repository.get_repository("devtable", "gargantuan")
broken_image = Image.create(
docker_image_id="foo",
repository=repo,
ancestors="/348723847234/",
storage=ImageStorage.get(),
)
broken_image_tag = RepositoryTag.create(repository=repo,
image=broken_image,
name="broken")
# Backfill the tag.
assert backfill_tag(broken_image_tag)
# Ensure we backfilled, even though we reference a broken manifest.
tag_manifest = TagManifest.get(tag=broken_image_tag)
map_row = TagManifestToManifest.get(tag_manifest=tag_manifest)
manifest = map_row.manifest
assert manifest.manifest_bytes == tag_manifest.json_data
tag = TagToRepositoryTag.get(repository_tag=broken_image_tag).tag
assert tag.name == "broken"
assert tag.manifest == manifest
def store_tag_manifest_for_repo(repository_id,
tag_name,
manifest,
leaf_layer_id,
storage_id_map,
reversion=False):
""" Stores a tag manifest for a specific tag name in the database. Returns the TagManifest
object, as well as a boolean indicating whether the TagManifest was created.
"""
# Create the new-style OCI manifest and its blobs.
oci_manifest = _populate_manifest_and_blobs(repository_id,
manifest,
storage_id_map,
leaf_layer_id=leaf_layer_id)
# Create the tag for the tag manifest.
tag = create_or_update_tag_for_repo(repository_id,
tag_name,
leaf_layer_id,
reversion=reversion,
oci_manifest=oci_manifest)
# Add a tag manifest pointing to that tag.
try:
manifest = TagManifest.get(digest=manifest.digest)
manifest.tag = tag
manifest.save()
return manifest, False
except TagManifest.DoesNotExist:
created = _associate_manifest(tag, oci_manifest)
return created, True
def _namespace_from_kwargs(self, args_dict):
if "namespace_name" in args_dict:
return args_dict["namespace_name"]
if "repository_ref" in args_dict:
return args_dict["repository_ref"].namespace_name
if "tag" in args_dict:
return args_dict["tag"].repository.namespace_name
if "manifest" in args_dict:
manifest = args_dict["manifest"]
if manifest._is_tag_manifest:
return TagManifest.get(
id=manifest._db_id).tag.repository.namespace_user.username
else:
return Manifest.get(
id=manifest._db_id).repository.namespace_user.username
if "manifest_or_legacy_image" in args_dict:
manifest_or_legacy_image = args_dict["manifest_or_legacy_image"]
if isinstance(manifest_or_legacy_image, LegacyImage):
return Image.get(id=manifest_or_legacy_image._db_id
).repository.namespace_user.username
else:
manifest = manifest_or_legacy_image
if manifest._is_tag_manifest:
return TagManifest.get(
id=manifest._db_id
).tag.repository.namespace_user.username
else:
return Manifest.get(
id=manifest._db_id).repository.namespace_user.username
if "derived_image" in args_dict:
return DerivedStorageForImage.get(
id=args_dict["derived_image"]._db_id
).source_image.repository.namespace_user.username
if "blob" in args_dict:
return "" # Blob functions are shared, so no need to do anything.
if "blob_upload" in args_dict:
return "" # Blob functions are shared, so no need to do anything.
raise Exception("Unknown namespace for dict `%s`" % args_dict)
def _namespace_from_kwargs(self, args_dict):
if 'namespace_name' in args_dict:
return args_dict['namespace_name']
if 'repository_ref' in args_dict:
return args_dict['repository_ref'].namespace_name
if 'tag' in args_dict:
return args_dict['tag'].repository.namespace_name
if 'manifest' in args_dict:
manifest = args_dict['manifest']
if manifest._is_tag_manifest:
return TagManifest.get(id=manifest._db_id).tag.repository.namespace_user.username
else:
return Manifest.get(id=manifest._db_id).repository.namespace_user.username
if 'manifest_or_legacy_image' in args_dict:
manifest_or_legacy_image = args_dict['manifest_or_legacy_image']
if isinstance(manifest_or_legacy_image, LegacyImage):
return Image.get(id=manifest_or_legacy_image._db_id).repository.namespace_user.username
else:
manifest = manifest_or_legacy_image
if manifest._is_tag_manifest:
return TagManifest.get(id=manifest._db_id).tag.repository.namespace_user.username
else:
return Manifest.get(id=manifest._db_id).repository.namespace_user.username
if 'derived_image' in args_dict:
return (DerivedStorageForImage
.get(id=args_dict['derived_image']._db_id)
.source_image
.repository
.namespace_user
.username)
if 'blob' in args_dict:
return '' # Blob functions are shared, so no need to do anything.
if 'blob_upload' in args_dict:
return '' # Blob functions are shared, so no need to do anything.
raise Exception('Unknown namespace for dict `%s`' % args_dict)
def _get_manifest_id(repositorytag):
repository_tag_datatype = TagDataType.for_repository_tag(repositorytag)
# Retrieve the TagManifest for the RepositoryTag, backfilling if necessary.
with db_transaction():
manifest_datatype = None
try:
manifest_datatype = pre_oci_model.get_manifest_for_tag(
repository_tag_datatype, backfill_if_necessary=True)
except MalformedSchema1Manifest:
logger.exception('Error backfilling manifest for tag `%s`',
repositorytag.id)
if manifest_datatype is None:
logger.error('Could not load or backfill manifest for tag `%s`',
repositorytag.id)
# Create a broken manifest for the tag.
tag_manifest = TagManifest.create(tag=repositorytag,
digest='BROKEN-%s' %
repositorytag.id,
json_data='{}')
else:
# Retrieve the new-style Manifest for the TagManifest, if any.
try:
tag_manifest = TagManifest.get(id=manifest_datatype._db_id)
except TagManifest.DoesNotExist:
logger.exception('Could not find tag manifest')
return None
try:
found = TagManifestToManifest.get(tag_manifest=tag_manifest).manifest
# Verify that the new-style manifest has the same contents as the old-style manifest.
# If not, update and then return. This is an extra check put in place to ensure unicode
# manifests have been correctly copied.
if found.manifest_bytes != tag_manifest.json_data:
logger.warning('Fixing manifest `%s`', found.id)
found.manifest_bytes = tag_manifest.json_data
found.save()
return found.id
except TagManifestToManifest.DoesNotExist:
# Could not find the new style manifest, so backfill.
_backfill_manifest(tag_manifest)
# Try to retrieve the manifest again, since we've performed a backfill.
try:
return TagManifestToManifest.get(tag_manifest=tag_manifest).manifest_id
except TagManifestToManifest.DoesNotExist:
return None
def verify_backfill(namespace_name):
logger.info('Checking namespace %s', namespace_name)
namespace_user = model.user.get_namespace_user(namespace_name)
assert namespace_user
repo_tags = (RepositoryTag
.select()
.join(Repository)
.where(Repository.namespace_user == namespace_user)
.where(RepositoryTag.hidden == False))
repo_tags = list(repo_tags)
logger.info('Found %s tags', len(repo_tags))
for index, repo_tag in enumerate(repo_tags):
logger.info('Checking tag %s under repository %s (%s/%s)', repo_tag.name,
repo_tag.repository.name, index + 1, len(repo_tags))
tag = TagToRepositoryTag.get(repository_tag=repo_tag).tag
assert not tag.hidden
assert tag.repository == repo_tag.repository
assert tag.name == repo_tag.name, _vs(tag.name, repo_tag.name)
assert tag.repository == repo_tag.repository, _vs(tag.repository_id, repo_tag.repository_id)
assert tag.reversion == repo_tag.reversion, _vs(tag.reversion, repo_tag.reversion)
start_check = int(tag.lifetime_start_ms / 1000) == repo_tag.lifetime_start_ts
assert start_check, _vs(tag.lifetime_start_ms, repo_tag.lifetime_start_ts)
if repo_tag.lifetime_end_ts is not None:
end_check = int(tag.lifetime_end_ms / 1000) == repo_tag.lifetime_end_ts
assert end_check, _vs(tag.lifetime_end_ms, repo_tag.lifetime_end_ts)
else:
assert tag.lifetime_end_ms is None
try:
tag_manifest = tag.manifest
repo_tag_manifest = TagManifest.get(tag=repo_tag)
digest_check = tag_manifest.digest == repo_tag_manifest.digest
assert digest_check, _vs(tag_manifest.digest, repo_tag_manifest.digest)
bytes_check = tag_manifest.manifest_bytes == repo_tag_manifest.json_data
assert bytes_check, _vs(tag_manifest.manifest_bytes, repo_tag_manifest.json_data)
except TagManifest.DoesNotExist:
logger.info('No tag manifest found for repository tag %s', repo_tag.id)
mli = ManifestLegacyImage.get(manifest=tag_manifest)
assert mli.repository == repo_tag.repository
manifest_legacy_image = mli.image
assert manifest_legacy_image == repo_tag.image, _vs(manifest_legacy_image.id, repo_tag.image_id)
def compute_layer_id(layer):
""" Returns the ID for the layer in the security scanner. """
# NOTE: this is temporary until we switch to Clair V3.
if isinstance(layer, ManifestDataType):
if layer._is_tag_manifest:
layer = TagManifest.get(id=layer._db_id).tag.image
else:
manifest = Manifest.get(id=layer._db_id)
try:
layer = ManifestLegacyImage.get(manifest=manifest).image
except ManifestLegacyImage.DoesNotExist:
return None
elif isinstance(layer, LegacyImage):
layer = Image.get(id=layer._db_id)
assert layer.docker_image_id
assert layer.storage.uuid
return '%s.%s' % (layer.docker_image_id, layer.storage.uuid)
def test_retarget_tag(initialized_db):
repo = get_repository("devtable", "history")
results, _ = list_repository_tag_history(repo,
1,
100,
specific_tag_name="latest")
assert len(results) == 2
assert results[0].lifetime_end_ms is None
assert results[1].lifetime_end_ms is not None
# Revert back to the original manifest.
created = retarget_tag("latest",
results[0].manifest,
is_reversion=True,
now_ms=results[1].lifetime_end_ms + 10000)
assert created.lifetime_end_ms is None
assert created.reversion
assert created.name == "latest"
assert created.manifest == results[0].manifest
# Verify in the history.
results, _ = list_repository_tag_history(repo,
1,
100,
specific_tag_name="latest")
assert len(results) == 3
assert results[0].lifetime_end_ms is None
assert results[1].lifetime_end_ms is not None
assert results[2].lifetime_end_ms is not None
assert results[0] == created
# Verify old-style tables.
repository_tag = TagToRepositoryTag.get(tag=created).repository_tag
assert repository_tag.lifetime_start_ts == int(created.lifetime_start_ms /
1000)
tag_manifest = TagManifest.get(tag=repository_tag)
assert TagManifestToManifest.get(
tag_manifest=tag_manifest).manifest == created.manifest
def get_tag_manifest(tag):
try:
return TagManifest.get(tag=tag)
except TagManifest.DoesNotExist:
return None
def test_tagbackfillworker(clear_all_rows, initialized_db):
# Remove the new-style rows so we can backfill.
TagToRepositoryTag.delete().execute()
Tag.delete().execute()
if clear_all_rows:
TagManifestLabelMap.delete().execute()
ManifestLabel.delete().execute()
ManifestBlob.delete().execute()
ManifestLegacyImage.delete().execute()
TagManifestToManifest.delete().execute()
Manifest.delete().execute()
found_dead_tag = False
for repository_tag in list(RepositoryTag.select()):
# Backfill the tag.
assert backfill_tag(repository_tag)
# Ensure if we try again, the backfill is skipped.
assert not backfill_tag(repository_tag)
# Ensure that we now have the expected tag rows.
tag_to_repo_tag = TagToRepositoryTag.get(repository_tag=repository_tag)
tag = tag_to_repo_tag.tag
assert tag.name == repository_tag.name
assert tag.repository == repository_tag.repository
assert not tag.hidden
assert tag.reversion == repository_tag.reversion
if repository_tag.lifetime_start_ts is None:
assert tag.lifetime_start_ms is None
else:
assert tag.lifetime_start_ms == (repository_tag.lifetime_start_ts *
1000)
if repository_tag.lifetime_end_ts is None:
assert tag.lifetime_end_ms is None
else:
assert tag.lifetime_end_ms == (repository_tag.lifetime_end_ts *
1000)
found_dead_tag = True
assert tag.manifest
# Ensure that we now have the expected manifest rows.
try:
tag_manifest = TagManifest.get(tag=repository_tag)
except TagManifest.DoesNotExist:
continue
map_row = TagManifestToManifest.get(tag_manifest=tag_manifest)
assert not map_row.broken
manifest_row = map_row.manifest
assert manifest_row.manifest_bytes == tag_manifest.json_data
assert manifest_row.digest == tag_manifest.digest
assert manifest_row.repository == tag_manifest.tag.repository
assert tag.manifest == map_row.manifest
legacy_image = ManifestLegacyImage.get(manifest=manifest_row).image
assert tag_manifest.tag.image == legacy_image
expected_storages = {tag_manifest.tag.image.storage.id}
for parent_image_id in tag_manifest.tag.image.ancestor_id_list():
expected_storages.add(Image.get(id=parent_image_id).storage_id)
found_storages = {
manifest_blob.blob_id
for manifest_blob in ManifestBlob.select().where(
ManifestBlob.manifest == manifest_row)
}
assert expected_storages == found_storages
# Ensure the labels were copied over.
tmls = list(TagManifestLabel.select().where(
TagManifestLabel.annotated == tag_manifest))
expected_labels = {tml.label_id for tml in tmls}
found_labels = {
m.label_id
for m in ManifestLabel.select().where(
ManifestLabel.manifest == manifest_row)
}
assert found_labels == expected_labels
# Verify at the repository level.
for repository in list(Repository.select()):
tags = RepositoryTag.select().where(
RepositoryTag.repository == repository,
RepositoryTag.hidden == False)
oci_tags = Tag.select().where(Tag.repository == repository)
assert len(tags) == len(oci_tags)
assert {t.name for t in tags} == {t.name for t in oci_tags}
for tag in tags:
tag_manifest = TagManifest.get(tag=tag)
ttr = TagToRepositoryTag.get(repository_tag=tag)
manifest = ttr.tag.manifest
assert tag_manifest.json_data == manifest.manifest_bytes
assert tag_manifest.digest == manifest.digest
assert tag.image == ManifestLegacyImage.get(
manifest=manifest).image
assert tag.lifetime_start_ts == (ttr.tag.lifetime_start_ms / 1000)
if tag.lifetime_end_ts:
assert tag.lifetime_end_ts == (ttr.tag.lifetime_end_ms / 1000)
else:
assert ttr.tag.lifetime_end_ms is None
assert found_dead_tag
def _backfill_manifest(tag_manifest):
logger.info('Backfilling manifest for tag manifest %s', tag_manifest.id)
# Ensure that a mapping row doesn't already exist. If it does, we've been preempted.
if lookup_manifest_map_row(tag_manifest):
return False
# Parse the manifest. If we cannot parse, then we treat the manifest as broken and just emit it
# without additional rows or data, as it will eventually not be useful.
is_broken = False
try:
manifest = DockerSchema1Manifest(Bytes.for_string_or_unicode(
tag_manifest.json_data),
validate=False)
except ManifestException:
logger.exception('Exception when trying to parse manifest %s',
tag_manifest.id)
manifest = BrokenManifest(tag_manifest.digest, tag_manifest.json_data)
is_broken = True
# Lookup the storages for the digests.
root_image = tag_manifest.tag.image
repository = tag_manifest.tag.repository
image_storage_id_map = {
root_image.storage.content_checksum: root_image.storage.id
}
try:
parent_images = get_parent_images(repository.namespace_user.username,
repository.name, root_image)
except DataModelException:
logger.exception(
'Exception when trying to load parent images for manifest `%s`',
tag_manifest.id)
parent_images = {}
is_broken = True
for parent_image in parent_images:
image_storage_id_map[
parent_image.storage.content_checksum] = parent_image.storage.id
# Ensure that all the expected blobs have been found. If not, we lookup the blob under the repo
# and add its storage ID. If the blob is not found, we mark the manifest as broken.
storage_ids = set()
try:
for blob_digest in manifest.get_blob_digests_for_translation():
if blob_digest in image_storage_id_map:
storage_ids.add(image_storage_id_map[blob_digest])
else:
logger.debug(
'Blob `%s` not found in images for manifest `%s`; checking repo',
blob_digest, tag_manifest.id)
try:
blob_storage = get_repo_blob_by_digest(
repository.namespace_user.username, repository.name,
blob_digest)
storage_ids.add(blob_storage.id)
except BlobDoesNotExist:
logger.debug(
'Blob `%s` not found in repo for manifest `%s`',
blob_digest, tag_manifest.id)
is_broken = True
except MalformedSchema1Manifest:
logger.warning(
'Found malformed schema 1 manifest during blob backfill')
is_broken = True
with db_transaction():
# Re-retrieve the tag manifest to ensure it still exists and we're pointing at the correct tag.
try:
tag_manifest = TagManifest.get(id=tag_manifest.id)
except TagManifest.DoesNotExist:
return True
# Ensure it wasn't already created.
if lookup_manifest_map_row(tag_manifest):
return False
# Check for a pre-existing manifest matching the digest in the repository. This can happen
# if we've already created the manifest row (typically for tag reverision).
try:
manifest_row = Manifest.get(digest=manifest.digest,
repository=tag_manifest.tag.repository)
except Manifest.DoesNotExist:
# Create the new-style rows for the manifest.
try:
manifest_row = populate_manifest(tag_manifest.tag.repository,
manifest,
tag_manifest.tag.image,
storage_ids)
except IntegrityError:
# Pre-empted.
return False
# Create the mapping row. If we find another was created for this tag manifest in the
# meantime, then we've been preempted.
try:
TagManifestToManifest.create(tag_manifest=tag_manifest,
manifest=manifest_row,
broken=is_broken)
except IntegrityError:
return False
# Backfill any labels on the manifest.
_backfill_labels(tag_manifest, manifest_row, repository)
return True