vbatts/quay
Archived
1
0
Fork 0
Code Issues Pull requests Releases Wiki Activity

Merge pull request #2404 from coreos-inc/cas-gc-fix

Browse source 
Fix GC handling around CAS paths
This commit is contained in:
josephschorr 2017-03-10 17:34:21 -05:00 • committed by GitHub
commit cbac673d58
3 changed files with 149 additions and 8 deletions
Download patch file Download diff file Expand all files Collapse all files

48
data/model/storage.py
View file

@ -8,7 +8,8 @@ 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,
ensure_under_transaction)
logger = logging.getLogger(__name__)
@ -79,10 +80,38 @@ def garbage_collect_storage(storage_id_whitelist):
if len(storage_id_whitelist) == 0:
return []
def placements_query_to_paths_set(placements_query):
return {(get_image_location_for_id(placement.location_id).name,
get_layer_path(placement.storage))
for placement in placements_query}
def placements_to_filtered_paths_set(placements_list):
""" Returns the list of paths to remove from storage, filtered from the given placements
query by removing any CAS paths that are still referenced by storage(s) in the database.
"""
with ensure_under_transaction():
if not placements_list:
return set()
# Find the content checksums not referenced by other storages. Any that are, we cannot
# remove.
content_checksums = set([placement.storage.content_checksum for placement in placements_list
if placement.storage.cas_path])
unreferenced_checksums = set()
if content_checksums:
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)
unreferenced_checksums = content_checksums - referenced_checksums
# Return all placements for all image storages found not at a CAS path or with a content
# checksum that is referenced.
return {(get_image_location_for_id(placement.location_id).name,
get_layer_path(placement.storage))
for placement in placements_list
if not placement.storage.cas_path or
placement.storage.content_checksum in unreferenced_checksums}
# Note: Both of these deletes must occur in the same transaction (unfortunately) because a
# storage without any placement is invalid, and a placement cannot exist without a storage.
@ -96,12 +125,10 @@ def garbage_collect_storage(storage_id_whitelist):
return []
placements_to_remove = list(ImageStoragePlacement
.select()
.select(ImageStoragePlacement, ImageStorage)
.join(ImageStorage)
.where(ImageStorage.id << orphaned_storage_ids))
paths_to_remove = placements_query_to_paths_set(placements_to_remove)
# Remove the placements for orphaned storages
if len(placements_to_remove) > 0:
placement_ids_to_remove = [placement.id for placement in placements_to_remove]
@ -130,6 +157,11 @@ def garbage_collect_storage(storage_id_whitelist):
.execute())
logger.debug('Removed %s image storage records', storages_removed)
# Determine the paths to remove. We cannot simply remove all paths matching storages, as CAS
# can share the same path. We further filter these paths by checking for any storages still in
# the database with the same content checksum.
paths_to_remove = placements_to_filtered_paths_set(placements_to_remove)
# We are going to make the conscious decision to not delete image storage blobs inside
# transactions.
# This may end up producing garbage in s3, trading off for higher availability in the database.