##// END OF EJS Templates
chore: improved logging for request processing
chore: improved logging for request processing

File last commit:

r1277:8ff72d35 default
r1279:9241b309 default
Show More
objectstore_cache.py
173 lines | 6.2 KiB | text/x-python | PythonLexer
feat(archive-cache): synced with CE codebase
r1251 # Copyright (C) 2015-2024 RhodeCode GmbH
#
# This program is free software: you can redistribute it and/or modify
# it under the terms of the GNU Affero General Public License, version 3
# (only), as published by the Free Software Foundation.
#
# This program is distributed in the hope that it will be useful,
# but WITHOUT ANY WARRANTY; without even the implied warranty of
# MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
# GNU General Public License for more details.
#
# You should have received a copy of the GNU Affero General Public License
# along with this program. If not, see <http://www.gnu.org/licenses/>.
#
# This program is dual-licensed. If you wish to learn more about the
# RhodeCode Enterprise Edition, including its added features, Support services,
# and proprietary license terms, please see https://rhodecode.com/licenses/
import codecs
import hashlib
import logging
import os
archive-cache: synced with CE codebase
r1259 import typing
feat(archive-cache): synced with CE codebase
r1251
import fsspec
from .base import BaseCache, BaseShard
from ..utils import ShardFileReader, NOT_GIVEN
from ...type_utils import str2bool
log = logging.getLogger(__name__)
class S3Shard(BaseShard):
archive-cache: synced with CE codebase
r1257 def __init__(self, index, bucket, bucket_folder, fs, **settings):
archive-cache: synced with CE codebase
r1259 self._index: int = index
self._bucket_folder: str = bucket_folder
self.storage_type: str = 'bucket'
self._bucket_main: str = bucket
feat(archive-cache): synced with CE codebase
r1251
archive-cache: synced with CE codebase
r1257 self.fs = fs
feat(archive-cache): synced with CE codebase
r1251
@property
archive-cache: synced with CE codebase
r1259 def bucket(self) -> str:
archive-cache: synced with CE codebase
r1257 """Cache bucket final path."""
return os.path.join(self._bucket_main, self._bucket_folder)
feat(archive-cache): synced with CE codebase
r1251
def _get_keyfile(self, archive_key) -> tuple[str, str]:
archive-cache: synced with CE codebase
r1259 key_file: str = f'{archive_key}-{self.key_suffix}'
feat(archive-cache): synced with CE codebase
r1251 return key_file, os.path.join(self.bucket, key_file)
def _get_writer(self, path, mode):
return self.fs.open(path, 'wb')
def _write_file(self, full_path, iterator, mode):
archive-cache: synced with CE codebase
r1256
archive-cache: synced with CE codebase
r1257 # ensure folder in bucket exists
feat(archive-cache): synced with CE codebase
r1251 destination = self.bucket
if not self.fs.exists(destination):
chore: cleanup, remove unused param in objectstore_cache
r1277 self.fs.mkdir(destination)
feat(archive-cache): synced with CE codebase
r1251
writer = self._get_writer(full_path, mode)
digest = hashlib.sha256()
with writer:
size = 0
for chunk in iterator:
size += len(chunk)
digest.update(chunk)
writer.write(chunk)
sha256 = digest.hexdigest()
log.debug('written new archive cache under %s, sha256: %s', full_path, sha256)
return size, sha256
def store(self, key, value_reader, metadata: dict | None = None):
return self._store(key, value_reader, metadata, mode='wb')
archive-cache: synced with CE codebase
r1259 def fetch(self, key, retry=NOT_GIVEN,
retry_attempts=NOT_GIVEN, retry_backoff=1,
presigned_url_expires: int = 0) -> tuple[ShardFileReader, dict]:
return self._fetch(key, retry, retry_attempts, retry_backoff, presigned_url_expires=presigned_url_expires)
feat(archive-cache): synced with CE codebase
r1251
def remove(self, key):
return self._remove(key)
def random_filename(self):
"""Return filename and full-path tuple for file storage.
Filename will be a randomly generated 28 character hexadecimal string
with ".archive_cache" suffixed. Two levels of sub-directories will be used to
reduce the size of directories. On older filesystems, lookups in
directories with many files may be slow.
"""
hex_name = codecs.encode(os.urandom(16), 'hex').decode('utf-8')
archive_name = hex_name[4:] + '.archive_cache'
filename = f"{hex_name[:2]}-{hex_name[2:4]}-{archive_name}"
full_path = os.path.join(self.bucket, filename)
return archive_name, full_path
def __repr__(self):
return f'{self.__class__.__name__}(index={self._index}, bucket={self.bucket})'
class ObjectStoreCache(BaseCache):
archive-cache: synced with CE codebase
r1259 shard_name: str = 'shard-{:03d}'
archive-cache: synced with CE codebase
r1257 shard_cls = S3Shard
feat(archive-cache): synced with CE codebase
r1251
def __init__(self, locking_url, **settings):
"""
Initialize objectstore cache instance.
:param str locking_url: redis url for a lock
:param settings: settings dict
"""
self._locking_url = locking_url
self._config = settings
objectstore_url = self.get_conf('archive_cache.objectstore.url')
archive-cache: synced with CE codebase
r1257 self._storage_path = objectstore_url # common path for all from BaseCache
feat(archive-cache): synced with CE codebase
r1251
archive-cache: synced with CE codebase
r1257 self._shard_count = int(self.get_conf('archive_cache.objectstore.bucket_shards', pop=True))
if self._shard_count < 1:
raise ValueError('cache_shards must be 1 or more')
self._bucket = settings.pop('archive_cache.objectstore.bucket')
if not self._bucket:
raise ValueError('archive_cache.objectstore.bucket needs to have a value')
feat(archive-cache): synced with CE codebase
r1251
self._eviction_policy = self.get_conf('archive_cache.objectstore.eviction_policy', pop=True)
self._cache_size_limit = self.gb_to_bytes(int(self.get_conf('archive_cache.objectstore.cache_size_gb')))
self.retry = str2bool(self.get_conf('archive_cache.objectstore.retry', pop=True))
self.retry_attempts = int(self.get_conf('archive_cache.objectstore.retry_attempts', pop=True))
self.retry_backoff = int(self.get_conf('archive_cache.objectstore.retry_backoff', pop=True))
archive-cache: synced with CE codebase
r1257 endpoint_url = settings.pop('archive_cache.objectstore.url')
key = settings.pop('archive_cache.objectstore.key')
secret = settings.pop('archive_cache.objectstore.secret')
feat(s3 region): added configurable s3 region.
r1264 region = settings.pop('archive_cache.objectstore.region')
archive-cache: synced with CE codebase
r1257
fix: fixed logging in archive-cache
r1258 log.debug('Initializing %s archival cache instance', self)
archive-cache: synced with CE codebase
r1257
feat(s3 region): added configurable s3 region.
r1264 fs = fsspec.filesystem(
's3', anon=False, endpoint_url=endpoint_url, key=key, secret=secret, client_kwargs={'region_name': region}
)
archive-cache: synced with CE codebase
r1257
# init main bucket
if not fs.exists(self._bucket):
fs.mkdir(self._bucket)
feat(archive-cache): synced with CE codebase
r1251 self._shards = tuple(
archive-cache: synced with CE codebase
r1257 self.shard_cls(
feat(archive-cache): synced with CE codebase
r1251 index=num,
archive-cache: synced with CE codebase
r1257 bucket=self._bucket,
archive-cache: synced with CE codebase
r1259 bucket_folder=self.shard_name.format(num),
archive-cache: synced with CE codebase
r1257 fs=fs,
feat(archive-cache): synced with CE codebase
r1251 **settings,
)
archive-cache: synced with CE codebase
r1257 for num in range(self._shard_count)
feat(archive-cache): synced with CE codebase
r1251 )
self._hash = self._shards[0].hash
def _get_size(self, shard, archive_path):
return shard.fs.info(archive_path)['size']
archive-cache: synced with CE codebase
r1259
def set_presigned_url_expiry(self, val: int) -> None:
self.presigned_url_expires = val