def test_manifestbackfillworker_repeat_digest(clear_rows, initialized_db):
"""
Tests that a manifest with a shared digest will be properly linked.
"""
# Delete existing tag manifest so we can reuse the tag.
TagManifestLabel.delete().execute()
TagManifest.delete().execute()
repo = model.repository.get_repository("devtable", "gargantuan")
tag_v30 = model.tag.get_active_tag("devtable", "gargantuan", "v3.0")
tag_v50 = model.tag.get_active_tag("devtable", "gargantuan", "v5.0")
# Build a manifest and assign it to both tags (this is allowed in the old model).
builder = DockerSchema1ManifestBuilder("devtable", "gargantuan", "sometag")
builder.add_layer("sha256:deadbeef", '{"id": "foo"}')
manifest = builder.build(docker_v2_signing_key)
manifest_1 = TagManifest.create(json_data=manifest.bytes.as_encoded_str(),
digest=manifest.digest,
tag=tag_v30)
manifest_2 = TagManifest.create(json_data=manifest.bytes.as_encoded_str(),
digest=manifest.digest,
tag=tag_v50)
# Backfill "both" manifests and ensure both are pointed to by a single resulting row.
assert _backfill_manifest(manifest_1)
assert _backfill_manifest(manifest_2)
map_row1 = TagManifestToManifest.get(tag_manifest=manifest_1)
map_row2 = TagManifestToManifest.get(tag_manifest=manifest_2)
assert map_row1.manifest == map_row2.manifest
def delete_manifests():
ManifestLegacyImage.delete().execute()
ManifestBlob.delete().execute()
Manifest.delete().execute()
TagManifestToManifest.delete().execute()
TagManifest.delete().execute()
return "OK"
def _associate_manifest(tag, oci_manifest):
with db_transaction():
tag_manifest = TagManifest.create(
tag=tag,
digest=oci_manifest.digest,
json_data=oci_manifest.manifest_bytes)
TagManifestToManifest.create(tag_manifest=tag_manifest,
manifest=oci_manifest)
return tag_manifest
def clear_rows(initialized_db):
# Remove all new-style rows so we can backfill.
TagToRepositoryTag.delete().execute()
Tag.delete().execute()
TagManifestLabelMap.delete().execute()
ManifestLabel.delete().execute()
ManifestBlob.delete().execute()
ManifestLegacyImage.delete().execute()
TagManifestToManifest.delete().execute()
Manifest.delete().execute()
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 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 test_store_tag_manifest(get_storages, initialized_db):
# Create a manifest with some layers.
builder = DockerSchema1ManifestBuilder('devtable', 'simple', 'sometag')
storages = get_storages()
assert storages
repo = model.repository.get_repository('devtable', 'simple')
storage_id_map = {}
for index, storage in enumerate(storages):
image_id = 'someimage%s' % index
builder.add_layer(storage.content_checksum, json.dumps({'id': image_id}))
find_create_or_link_image(image_id, repo, 'devtable', {}, 'local_us')
storage_id_map[storage.content_checksum] = storage.id
manifest = builder.build(docker_v2_signing_key)
tag_manifest, _ = store_tag_manifest_for_testing('devtable', 'simple', 'sometag', manifest,
manifest.leaf_layer_v1_image_id, storage_id_map)
# Ensure we have the new-model expected rows.
mapping_row = TagManifestToManifest.get(tag_manifest=tag_manifest)
assert mapping_row.manifest is not None
assert mapping_row.manifest.manifest_bytes == manifest.bytes.as_encoded_str()
assert mapping_row.manifest.digest == str(manifest.digest)
blob_rows = {m.blob_id for m in
ManifestBlob.select().where(ManifestBlob.manifest == mapping_row.manifest)}
assert blob_rows == {s.id for s in storages}
assert ManifestLegacyImage.get(manifest=mapping_row.manifest).image == tag_manifest.tag.image
def test_manifestbackfillworker_mislinked_manifest(clear_rows, initialized_db):
""" Tests that a manifest whose image is mislinked will have its storages relinked properly. """
# Delete existing tag manifest so we can reuse the tag.
TagManifestLabel.delete().execute()
TagManifest.delete().execute()
repo = model.repository.get_repository('devtable', 'complex')
tag_v30 = model.tag.get_active_tag('devtable', 'gargantuan', 'v3.0')
tag_v50 = model.tag.get_active_tag('devtable', 'gargantuan', 'v5.0')
# Add a mislinked manifest, by having its layer point to a blob in v3.0 but its image
# be the v5.0 image.
builder = DockerSchema1ManifestBuilder('devtable', 'gargantuan', 'sometag')
builder.add_layer(tag_v30.image.storage.content_checksum, '{"id": "foo"}')
manifest = builder.build(docker_v2_signing_key)
mislinked_manifest = TagManifest.create(json_data=manifest.bytes.as_encoded_str(),
digest=manifest.digest,
tag=tag_v50)
# Backfill the manifest and ensure its proper content checksum was linked.
assert _backfill_manifest(mislinked_manifest)
map_row = TagManifestToManifest.get(tag_manifest=mislinked_manifest)
assert not map_row.broken
manifest_row = map_row.manifest
legacy_image = ManifestLegacyImage.get(manifest=manifest_row).image
assert legacy_image == tag_v50.image
manifest_blobs = list(ManifestBlob.select().where(ManifestBlob.manifest == manifest_row))
assert len(manifest_blobs) == 1
assert manifest_blobs[0].blob.content_checksum == tag_v30.image.storage.content_checksum
def test_manifestbackfillworker_mislinked_invalid_manifest(clear_rows, initialized_db):
""" Tests that a manifest whose image is mislinked will attempt to have its storages relinked
properly. """
# Delete existing tag manifest so we can reuse the tag.
TagManifestLabel.delete().execute()
TagManifest.delete().execute()
repo = model.repository.get_repository("devtable", "complex")
tag_v50 = model.tag.get_active_tag("devtable", "gargantuan", "v5.0")
# Add a mislinked manifest, by having its layer point to an invalid blob but its image
# be the v5.0 image.
builder = DockerSchema1ManifestBuilder("devtable", "gargantuan", "sometag")
builder.add_layer("sha256:deadbeef", '{"id": "foo"}')
manifest = builder.build(docker_v2_signing_key)
broken_manifest = TagManifest.create(
json_data=manifest.bytes.as_encoded_str(), digest=manifest.digest, tag=tag_v50
)
# Backfill the manifest and ensure it is marked as broken.
assert _backfill_manifest(broken_manifest)
map_row = TagManifestToManifest.get(tag_manifest=broken_manifest)
assert map_row.broken
manifest_row = map_row.manifest
legacy_image = ManifestLegacyImage.get(manifest=manifest_row).image
assert legacy_image == tag_v50.image
manifest_blobs = list(ManifestBlob.select().where(ManifestBlob.manifest == manifest_row))
assert len(manifest_blobs) == 0
def create_manifest_label(tag_manifest,
key,
value,
source_type_name,
media_type_name=None):
"""
Creates a new manifest label on a specific tag manifest.
"""
if not key:
raise InvalidLabelKeyException("Missing key on label")
# Note that we don't prevent invalid label names coming from the manifest to be stored, as Docker
# does not currently prevent them from being put into said manifests.
if not validate_label_key(key) and source_type_name != "manifest":
raise InvalidLabelKeyException(
"Label key `%s` is invalid or reserved" % key)
# Find the matching media type. If none specified, we infer.
if media_type_name is None:
media_type_name = "text/plain"
if is_json(value):
media_type_name = "application/json"
media_type_id = _get_media_type_id(media_type_name)
if media_type_id is None:
raise InvalidMediaTypeException()
source_type_id = _get_label_source_type_id(source_type_name)
with db_transaction():
label = Label.create(key=key,
value=value,
source_type=source_type_id,
media_type=media_type_id)
tag_manifest_label = TagManifestLabel.create(
annotated=tag_manifest,
label=label,
repository=tag_manifest.tag.repository)
try:
mapping_row = TagManifestToManifest.get(tag_manifest=tag_manifest)
if mapping_row.manifest:
manifest_label = ManifestLabel.create(
manifest=mapping_row.manifest,
label=label,
repository=tag_manifest.tag.repository,
)
TagManifestLabelMap.create(
manifest_label=manifest_label,
tag_manifest_label=tag_manifest_label,
label=label,
manifest=mapping_row.manifest,
tag_manifest=tag_manifest,
)
except TagManifestToManifest.DoesNotExist:
pass
return label
def _garbage_collect_legacy_manifest(legacy_manifest_id, context):
assert legacy_manifest_id is not None
# Add the labels to be GCed.
query = TagManifestLabel.select().where(
TagManifestLabel.annotated == legacy_manifest_id)
for manifest_label in query:
context.add_label_id(manifest_label.label_id)
# Delete the tag manifest.
with db_transaction():
try:
tag_manifest = TagManifest.select().where(
TagManifest.id == legacy_manifest_id).get()
except TagManifest.DoesNotExist:
return False
assert tag_manifest.id == legacy_manifest_id
assert tag_manifest.tag.repository_id == context.repository.id
# Delete any label mapping rows.
(TagManifestLabelMap.delete().where(
TagManifestLabelMap.tag_manifest == legacy_manifest_id).execute())
# Delete the label rows.
TagManifestLabel.delete().where(
TagManifestLabel.annotated == legacy_manifest_id).execute()
# Delete the mapping row if it exists.
try:
tmt = (TagManifestToManifest.select().where(
TagManifestToManifest.tag_manifest == tag_manifest).get())
context.add_manifest_id(tmt.manifest_id)
tmt_deleted = tmt.delete_instance()
if tmt_deleted:
gc_table_rows_deleted.labels(
table="TagManifestToManifest").inc()
except TagManifestToManifest.DoesNotExist:
pass
# Delete the tag manifest.
tag_manifest_deleted = tag_manifest.delete_instance()
if tag_manifest_deleted:
gc_table_rows_deleted.labels(table="TagManifest").inc()
return True
def backfill_label(tag_manifest_label):
logger.info("Backfilling label %s", tag_manifest_label.id)
# Ensure that a mapping row doesn't already exist. If it does, we've been preempted.
if lookup_map_row(tag_manifest_label):
return False
# Ensure the tag manifest has been backfilled into the manifest table.
try:
tmt = TagManifestToManifest.get(
tag_manifest=tag_manifest_label.annotated)
except TagManifestToManifest.DoesNotExist:
# We'll come back to this later.
logger.debug(
"Tag Manifest %s for label %s has not yet been backfilled",
tag_manifest_label.annotated.id,
tag_manifest_label.id,
)
return True
repository = tag_manifest_label.repository
# Create the new mapping entry and label.
with db_transaction():
if lookup_map_row(tag_manifest_label):
return False
label = tag_manifest_label.label
if tmt.manifest:
try:
manifest_label = ManifestLabel.create(manifest=tmt.manifest,
label=label,
repository=repository)
TagManifestLabelMap.create(
manifest_label=manifest_label,
tag_manifest_label=tag_manifest_label,
label=label,
manifest=tmt.manifest,
tag_manifest=tag_manifest_label.annotated,
)
except IntegrityError:
return False
logger.info("Backfilled label %s", tag_manifest_label.id)
return True
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 test_manifestbackfillworker_broken_manifest(clear_rows, initialized_db):
# Delete existing tag manifest so we can reuse the tag.
TagManifestLabel.delete().execute()
TagManifest.delete().execute()
# Add a broken manifest.
broken_manifest = TagManifest.create(json_data='wat?', digest='sha256:foobar',
tag=RepositoryTag.get())
# Ensure the backfill works.
assert _backfill_manifest(broken_manifest)
# Ensure the mapping is marked as broken.
map_row = TagManifestToManifest.get(tag_manifest=broken_manifest)
assert map_row.broken
manifest_row = map_row.manifest
assert manifest_row.manifest_bytes == broken_manifest.json_data
assert manifest_row.digest == broken_manifest.digest
assert manifest_row.repository == broken_manifest.tag.repository
legacy_image = ManifestLegacyImage.get(manifest=manifest_row).image
assert broken_manifest.tag.image == legacy_image
def _garbage_collect_manifest(manifest_id, context):
assert manifest_id is not None
# Make sure the manifest isn't referenced.
if _check_manifest_used(manifest_id):
return False
# Add the manifest's blobs to the context to be GCed.
for manifest_blob in ManifestBlob.select().where(ManifestBlob.manifest == manifest_id):
context.add_blob_id(manifest_blob.blob_id)
# Retrieve the manifest's associated image, if any.
try:
legacy_image_id = ManifestLegacyImage.get(manifest=manifest_id).image_id
context.add_legacy_image_id(legacy_image_id)
except ManifestLegacyImage.DoesNotExist:
legacy_image_id = None
# Add child manifests to be GCed.
for connector in ManifestChild.select().where(ManifestChild.manifest == manifest_id):
context.add_manifest_id(connector.child_manifest_id)
# Add the labels to be GCed.
for manifest_label in ManifestLabel.select().where(ManifestLabel.manifest == manifest_id):
context.add_label_id(manifest_label.label_id)
# Delete the manifest.
with db_transaction():
try:
manifest = Manifest.select().where(Manifest.id == manifest_id).get()
except Manifest.DoesNotExist:
return False
assert manifest.id == manifest_id
assert manifest.repository_id == context.repository.id
if _check_manifest_used(manifest_id):
return False
# Delete any label mappings.
(TagManifestLabelMap
.delete()
.where(TagManifestLabelMap.manifest == manifest_id)
.execute())
# Delete any mapping rows for the manifest.
TagManifestToManifest.delete().where(TagManifestToManifest.manifest == manifest_id).execute()
# Delete any label rows.
ManifestLabel.delete().where(ManifestLabel.manifest == manifest_id,
ManifestLabel.repository == context.repository).execute()
# Delete any child manifest rows.
ManifestChild.delete().where(ManifestChild.manifest == manifest_id,
ManifestChild.repository == context.repository).execute()
# Delete the manifest blobs for the manifest.
ManifestBlob.delete().where(ManifestBlob.manifest == manifest_id,
ManifestBlob.repository == context.repository).execute()
# Delete the manifest legacy image row.
if legacy_image_id:
(ManifestLegacyImage
.delete()
.where(ManifestLegacyImage.manifest == manifest_id,
ManifestLegacyImage.repository == context.repository)
.execute())
# Delete the manifest.
manifest.delete_instance()
context.mark_manifest_removed(manifest)
return True
def _get_dangling_manifest_count():
manifest_ids = set([current.id for current in Manifest.select()])
referenced_by_tag_manifest = set(
[tmt.manifest_id for tmt in TagManifestToManifest.select()])
return len(manifest_ids - referenced_by_tag_manifest)
def retarget_tag(tag_name,
manifest_id,
is_reversion=False,
now_ms=None,
adjust_old_model=True):
"""
Creates or updates a tag with the specified name to point to the given manifest under its
repository.
If this action is a reversion to a previous manifest, is_reversion should be set to True.
Returns the newly created tag row or None on error.
"""
try:
manifest = (Manifest.select(
Manifest,
MediaType).join(MediaType).where(Manifest.id == manifest_id).get())
except Manifest.DoesNotExist:
return None
# CHECK: Make sure that we are not mistargeting a schema 1 manifest to a tag with a different
# name.
if manifest.media_type.name in DOCKER_SCHEMA1_CONTENT_TYPES:
try:
parsed = DockerSchema1Manifest(Bytes.for_string_or_unicode(
manifest.manifest_bytes),
validate=False)
if parsed.tag != tag_name:
logger.error(
"Tried to re-target schema1 manifest with tag `%s` to tag `%s",
parsed.tag,
tag_name,
)
return None
except MalformedSchema1Manifest:
logger.exception("Could not parse schema1 manifest")
return None
legacy_image = get_legacy_image_for_manifest(manifest)
now_ms = now_ms or get_epoch_timestamp_ms()
now_ts = int(now_ms / 1000)
with db_transaction():
# Lookup an existing tag in the repository with the same name and, if present, mark it
# as expired.
existing_tag = get_tag(manifest.repository_id, tag_name)
if existing_tag is not None:
_, okay = set_tag_end_ms(existing_tag, now_ms)
# TODO: should we retry here and/or use a for-update?
if not okay:
return None
# Create a new tag pointing to the manifest with a lifetime start of now.
created = Tag.create(
name=tag_name,
repository=manifest.repository_id,
lifetime_start_ms=now_ms,
reversion=is_reversion,
manifest=manifest,
tag_kind=Tag.tag_kind.get_id("tag"),
)
# TODO: Remove the linkage code once RepositoryTag is gone.
# If this is a schema 1 manifest, then add a TagManifest linkage to it. Otherwise, it will only
# be pullable via the new OCI model.
if adjust_old_model:
if (manifest.media_type.name in DOCKER_SCHEMA1_CONTENT_TYPES
and legacy_image is not None):
old_style_tag = RepositoryTag.create(
repository=manifest.repository_id,
image=legacy_image,
name=tag_name,
lifetime_start_ts=now_ts,
reversion=is_reversion,
)
TagToRepositoryTag.create(tag=created,
repository_tag=old_style_tag,
repository=manifest.repository_id)
tag_manifest = TagManifest.create(
tag=old_style_tag,
digest=manifest.digest,
json_data=manifest.manifest_bytes)
TagManifestToManifest.create(tag_manifest=tag_manifest,
manifest=manifest,
repository=manifest.repository_id)
return created
def _garbage_collect_manifest(manifest_id, context):
assert manifest_id is not None
# Make sure the manifest isn't referenced.
if _check_manifest_used(manifest_id):
return False
# Add the manifest's blobs to the context to be GCed.
for manifest_blob in ManifestBlob.select().where(
ManifestBlob.manifest == manifest_id):
context.add_blob_id(manifest_blob.blob_id)
# Retrieve the manifest's associated image, if any.
try:
legacy_image_id = ManifestLegacyImage.get(
manifest=manifest_id).image_id
context.add_legacy_image_id(legacy_image_id)
except ManifestLegacyImage.DoesNotExist:
legacy_image_id = None
# Add child manifests to be GCed.
for connector in ManifestChild.select().where(
ManifestChild.manifest == manifest_id):
context.add_manifest_id(connector.child_manifest_id)
# Add the labels to be GCed.
for manifest_label in ManifestLabel.select().where(
ManifestLabel.manifest == manifest_id):
context.add_label_id(manifest_label.label_id)
# Delete the manifest.
with db_transaction():
try:
manifest = Manifest.select().where(
Manifest.id == manifest_id).get()
except Manifest.DoesNotExist:
return False
assert manifest.id == manifest_id
assert manifest.repository_id == context.repository.id
if _check_manifest_used(manifest_id):
return False
# Delete any label mappings.
deleted_tag_manifest_label_map = (TagManifestLabelMap.delete().where(
TagManifestLabelMap.manifest == manifest_id).execute())
# Delete any mapping rows for the manifest.
deleted_tag_manifest_to_manifest = (
TagManifestToManifest.delete().where(
TagManifestToManifest.manifest == manifest_id).execute())
# Delete any label rows.
deleted_manifest_label = (ManifestLabel.delete().where(
ManifestLabel.manifest == manifest_id,
ManifestLabel.repository == context.repository,
).execute())
# Delete any child manifest rows.
deleted_manifest_child = (ManifestChild.delete().where(
ManifestChild.manifest == manifest_id,
ManifestChild.repository == context.repository,
).execute())
# Delete the manifest blobs for the manifest.
deleted_manifest_blob = (ManifestBlob.delete().where(
ManifestBlob.manifest == manifest_id,
ManifestBlob.repository == context.repository).execute())
# Delete the security status for the manifest
deleted_manifest_security = (ManifestSecurityStatus.delete().where(
ManifestSecurityStatus.manifest == manifest_id,
ManifestSecurityStatus.repository == context.repository,
).execute())
# Delete the manifest legacy image row.
deleted_manifest_legacy_image = 0
if legacy_image_id:
deleted_manifest_legacy_image = (
ManifestLegacyImage.delete().where(
ManifestLegacyImage.manifest == manifest_id,
ManifestLegacyImage.repository == context.repository,
).execute())
# Delete the manifest.
manifest.delete_instance()
context.mark_manifest_removed(manifest)
gc_table_rows_deleted.labels(
table="TagManifestLabelMap").inc(deleted_tag_manifest_label_map)
gc_table_rows_deleted.labels(
table="TagManifestToManifest").inc(deleted_tag_manifest_to_manifest)
gc_table_rows_deleted.labels(
table="ManifestLabel").inc(deleted_manifest_label)
gc_table_rows_deleted.labels(
table="ManifestChild").inc(deleted_manifest_child)
gc_table_rows_deleted.labels(
table="ManifestBlob").inc(deleted_manifest_blob)
gc_table_rows_deleted.labels(
table="ManifestSecurityStatus").inc(deleted_manifest_security)
if deleted_manifest_legacy_image:
gc_table_rows_deleted.labels(
table="ManifestLegacyImage").inc(deleted_manifest_legacy_image)
gc_table_rows_deleted.labels(table="Manifest").inc()
return True
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 lookup_manifest_map_row(tag_manifest):
try:
TagManifestToManifest.get(tag_manifest=tag_manifest)
return True
except TagManifestToManifest.DoesNotExist:
return False
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
def create_manifest_label(manifest_id,
key,
value,
source_type_name,
media_type_name=None,
adjust_old_model=True):
""" Creates a new manifest label on a specific tag manifest. """
if not key:
raise InvalidLabelKeyException()
# Note that we don't prevent invalid label names coming from the manifest to be stored, as Docker
# does not currently prevent them from being put into said manifests.
if not validate_label_key(key) and source_type_name != "manifest":
raise InvalidLabelKeyException("Key `%s` is invalid" % key)
# Find the matching media type. If none specified, we infer.
if media_type_name is None:
media_type_name = "text/plain"
if is_json(value):
media_type_name = "application/json"
try:
media_type_id = Label.media_type.get_id(media_type_name)
except MediaType.DoesNotExist:
raise InvalidMediaTypeException()
source_type_id = Label.source_type.get_id(source_type_name)
# Ensure the manifest exists.
try:
manifest = (Manifest.select(Manifest,
Repository).join(Repository).where(
Manifest.id == manifest_id).get())
except Manifest.DoesNotExist:
return None
repository = manifest.repository
# TODO: Remove this code once the TagManifest table is gone.
tag_manifest = None
if adjust_old_model:
try:
mapping_row = (TagManifestToManifest.select(
TagManifestToManifest, TagManifest).join(TagManifest).where(
TagManifestToManifest.manifest == manifest).get())
tag_manifest = mapping_row.tag_manifest
except TagManifestToManifest.DoesNotExist:
tag_manifest = None
with db_transaction():
label = Label.create(key=key,
value=value,
source_type=source_type_id,
media_type=media_type_id)
manifest_label = ManifestLabel.create(manifest=manifest_id,
label=label,
repository=repository)
# If there exists a mapping to a TagManifest, add the old-style label.
# TODO: Remove this code once the TagManifest table is gone.
if tag_manifest:
tag_manifest_label = TagManifestLabel.create(
annotated=tag_manifest, label=label, repository=repository)
TagManifestLabelMap.create(
manifest_label=manifest_label,
tag_manifest_label=tag_manifest_label,
label=label,
manifest=manifest,
tag_manifest=tag_manifest,
)
return label
