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

This commit is contained in:
Joseph Schorr 2017-03-22 23:45:46 -04:00
parent 3d0e63d8e5
commit bdda74d6df
2 changed files with 60 additions and 7 deletions

View file

@ -8,7 +8,7 @@ from data.model import (config, db_transaction, InvalidImageException, TorrentIn
DataModelException, _basequery)
from data.database import (ImageStorage, Image, ImageStoragePlacement, ImageStorageLocation,
ImageStorageTransformation, ImageStorageSignature,
ImageStorageSignatureKind, Repository, Namespace, TorrentInfo,
ImageStorageSignatureKind, Repository, Namespace, TorrentInfo, Blob,
ensure_under_transaction)
@ -95,15 +95,24 @@ def garbage_collect_storage(storage_id_whitelist):
unreferenced_checksums = set()
if content_checksums:
# Check the current image storage.
query = (ImageStorage
.select(ImageStorage.content_checksum)
.where(ImageStorage.content_checksum << list(content_checksums)))
referenced_checksums = set([image_storage.content_checksum for image_storage in query])
if referenced_checksums:
logger.warning('GC attempted to remove CAS checksums %s, which are still referenced',
referenced_checksums)
is_referenced_checksums = set([image_storage.content_checksum for image_storage in query])
if is_referenced_checksums:
logger.warning('GC attempted to remove CAS checksums %s, which are still IS referenced',
is_referenced_checksums)
unreferenced_checksums = content_checksums - referenced_checksums
# Check the new Blob table as well.
query = Blob.select(Blob.digest).where(Blob.digest << list(content_checksums))
blob_referenced_checksums = set([blob.digest for blob in query])
if blob_referenced_checksums:
logger.warning('GC attempted to remove CAS checksums %s, which are still Blob referenced',
blob_referenced_checksums)
unreferenced_checksums = (content_checksums - blob_referenced_checksums -
is_referenced_checksums)
# Return all placements for all image storages found not at a CAS path or with a content
# checksum that is referenced.