class Checker(object):
def __init__(self,
namespace,
concurrency=50,
error_file=None,
rebuild_file=None,
full=True):
self.pool = GreenPool(concurrency)
self.error_file = error_file
self.full = bool(full)
if self.error_file:
f = open(self.error_file, 'a')
self.error_writer = csv.writer(f, delimiter=' ')
self.rebuild_file = rebuild_file
if self.rebuild_file:
fd = open(self.rebuild_file, 'a')
self.rebuild_writer = csv.writer(fd, delimiter='|')
conf = {'namespace': namespace}
self.account_client = AccountClient(conf)
self.container_client = ContainerClient(conf)
self.blob_client = BlobClient()
self.accounts_checked = 0
self.containers_checked = 0
self.objects_checked = 0
self.chunks_checked = 0
self.account_not_found = 0
self.container_not_found = 0
self.object_not_found = 0
self.chunk_not_found = 0
self.account_exceptions = 0
self.container_exceptions = 0
self.object_exceptions = 0
self.chunk_exceptions = 0
self.list_cache = {}
self.running = {}
def write_error(self, target):
error = [target.account]
if target.container:
error.append(target.container)
if target.obj:
error.append(target.obj)
if target.chunk:
error.append(target.chunk)
self.error_writer.writerow(error)
def write_rebuilder_input(self, target, obj_meta, ct_meta):
try:
cid = ct_meta['system']['sys.name'].split('.', 1)[0]
except KeyError:
cid = ct_meta['properties']['sys.name'].split('.', 1)[0]
self.rebuild_writer.writerow((cid, obj_meta['id'], target.chunk))
def write_chunk_error(self, target, obj_meta, chunk=None):
if chunk is not None:
target = target.copy()
target.chunk = chunk
if self.error_file:
self.write_error(target)
if self.rebuild_file:
self.write_rebuilder_input(
target, obj_meta,
self.list_cache[(target.account, target.container)][1])
def _check_chunk_xattr(self, target, obj_meta, xattr_meta):
error = False
# Composed position -> erasure coding
attr_prefix = 'meta' if '.' in obj_meta['pos'] else ''
attr_key = attr_prefix + 'chunk_size'
if str(obj_meta['size']) != xattr_meta.get(attr_key):
print(
" Chunk %s '%s' xattr (%s) "
"differs from size in meta2 (%s)" %
(target, attr_key, xattr_meta.get(attr_key), obj_meta['size']))
error = True
attr_key = attr_prefix + 'chunk_hash'
if obj_meta['hash'] != xattr_meta.get(attr_key):
print(
" Chunk %s '%s' xattr (%s) "
"differs from hash in meta2 (%s)" %
(target, attr_key, xattr_meta.get(attr_key), obj_meta['hash']))
error = True
return error
def check_chunk(self, target):
chunk = target.chunk
obj_listing, obj_meta = self.check_obj(target)
error = False
if chunk not in obj_listing:
print(' Chunk %s missing from object listing' % target)
error = True
db_meta = dict()
else:
db_meta = obj_listing[chunk]
try:
xattr_meta = self.blob_client.chunk_head(chunk, xattr=self.full)
except exc.NotFound as e:
self.chunk_not_found += 1
error = True
print(' Not found chunk "%s": %s' % (target, str(e)))
except Exception as e:
self.chunk_exceptions += 1
error = True
print(' Exception chunk "%s": %s' % (target, str(e)))
else:
if db_meta and self.full:
error = self._check_chunk_xattr(target, db_meta, xattr_meta)
if error:
self.write_chunk_error(target, obj_meta)
self.chunks_checked += 1
def check_obj_policy(self, target, obj_meta, chunks):
"""
Check that the list of chunks of an object matches
the object's storage policy.
"""
stg_met = STORAGE_METHODS.load(obj_meta['chunk_method'])
chunks_by_pos = _sort_chunks(chunks, stg_met.ec)
if stg_met.ec:
required = stg_met.ec_nb_data + stg_met.ec_nb_parity
else:
required = stg_met.nb_copy
for pos, clist in chunks_by_pos.iteritems():
if len(clist) < required:
print(' Missing %d chunks at position %s of %s' %
(required - len(clist), pos, target))
if stg_met.ec:
subs = {x['num'] for x in clist}
for sub in range(required):
if sub not in subs:
self.write_chunk_error(target, obj_meta,
'%d.%d' % (pos, sub))
else:
self.write_chunk_error(target, obj_meta, str(pos))
def check_obj(self, target, recurse=False):
account = target.account
container = target.container
obj = target.obj
if (account, container, obj) in self.running:
self.running[(account, container, obj)].wait()
if (account, container, obj) in self.list_cache:
return self.list_cache[(account, container, obj)]
self.running[(account, container, obj)] = Event()
print('Checking object "%s"' % target)
container_listing, ct_meta = self.check_container(target)
error = False
if obj not in container_listing:
print(' Object %s missing from container listing' % target)
error = True
# checksum = None
else:
# TODO check checksum match
# checksum = container_listing[obj]['hash']
pass
results = []
meta = dict()
try:
meta, results = self.container_client.content_locate(
account=account, reference=container, path=obj)
except exc.NotFound as e:
self.object_not_found += 1
error = True
print(' Not found object "%s": %s' % (target, str(e)))
except Exception as e:
self.object_exceptions += 1
error = True
print(' Exception object "%s": %s' % (target, str(e)))
chunk_listing = dict()
for chunk in results:
chunk_listing[chunk['url']] = chunk
self.check_obj_policy(target.copy(), meta, results)
self.objects_checked += 1
self.list_cache[(account, container, obj)] = (chunk_listing, meta)
self.running[(account, container, obj)].send(True)
del self.running[(account, container, obj)]
if recurse:
for chunk in chunk_listing:
t = target.copy()
t.chunk = chunk
self.pool.spawn_n(self.check_chunk, t)
if error and self.error_file:
self.write_error(target)
return chunk_listing, meta
def check_container(self, target, recurse=False):
account = target.account
container = target.container
if (account, container) in self.running:
self.running[(account, container)].wait()
if (account, container) in self.list_cache:
return self.list_cache[(account, container)]
self.running[(account, container)] = Event()
print('Checking container "%s"' % target)
account_listing = self.check_account(target)
error = False
if container not in account_listing:
error = True
print(' Container %s missing from account listing' % target)
marker = None
results = []
ct_meta = dict()
while True:
try:
_, resp = self.container_client.content_list(
account=account, reference=container, marker=marker)
except exc.NotFound as e:
self.container_not_found += 1
error = True
print(' Not found container "%s": %s' % (target, str(e)))
break
except Exception as e:
self.container_exceptions += 1
error = True
print(' Exception container "%s": %s' % (target, str(e)))
break
if resp['objects']:
marker = resp['objects'][-1]['name']
results.extend(resp['objects'])
else:
ct_meta = resp
ct_meta.pop('objects')
break
container_listing = dict()
for obj in results:
container_listing[obj['name']] = obj
self.containers_checked += 1
self.list_cache[(account, container)] = container_listing, ct_meta
self.running[(account, container)].send(True)
del self.running[(account, container)]
if recurse:
for obj in container_listing:
t = target.copy()
t.obj = obj
self.pool.spawn_n(self.check_obj, t, True)
if error and self.error_file:
self.write_error(target)
return container_listing, ct_meta
def check_account(self, target, recurse=False):
account = target.account
if account in self.running:
self.running[account].wait()
if account in self.list_cache:
return self.list_cache[account]
self.running[account] = Event()
print('Checking account "%s"' % target)
error = False
marker = None
results = []
while True:
try:
resp = self.account_client.container_list(account,
marker=marker)
except Exception as e:
self.account_exceptions += 1
error = True
print(' Exception account "%s": %s' % (target, str(e)))
break
if resp['listing']:
marker = resp['listing'][-1][0]
else:
break
results.extend(resp['listing'])
containers = dict()
for e in results:
containers[e[0]] = (e[1], e[2])
self.list_cache[account] = containers
self.running[account].send(True)
del self.running[account]
self.accounts_checked += 1
if recurse:
for container in containers:
t = target.copy()
t.container = container
self.pool.spawn_n(self.check_container, t, True)
if error and self.error_file:
self.write_error(target)
return containers
def check(self, target):
if target.chunk and target.obj and target.container:
self.pool.spawn_n(self.check_chunk, target)
elif target.obj and target.container:
self.pool.spawn_n(self.check_obj, target, True)
elif target.container:
self.pool.spawn_n(self.check_container, target, True)
else:
self.pool.spawn_n(self.check_account, target, True)
def wait(self):
self.pool.waitall()
def report(self):
def _report_stat(name, stat):
print("{0:18}: {1}".format(name, stat))
print()
print('Report')
_report_stat("Accounts checked", self.accounts_checked)
if self.account_not_found:
_report_stat("Missing accounts", self.account_not_found)
if self.account_exceptions:
_report_stat("Exceptions", self.account_not_found)
print()
_report_stat("Containers checked", self.containers_checked)
if self.container_not_found:
_report_stat("Missing containers", self.container_not_found)
if self.container_exceptions:
_report_stat("Exceptions", self.container_exceptions)
print()
_report_stat("Objects checked", self.objects_checked)
if self.object_not_found:
_report_stat("Missing objects", self.object_not_found)
if self.object_exceptions:
_report_stat("Exceptions", self.object_exceptions)
print()
_report_stat("Chunks checked", self.chunks_checked)
if self.chunk_not_found:
_report_stat("Missing chunks", self.chunk_not_found)
if self.chunk_exceptions:
_report_stat("Exceptions", self.chunk_exceptions)
class Checker(object):
def __init__(self, namespace, concurrency=50, error_file=None):
self.pool = GreenPool(concurrency)
self.error_file = error_file
if self.error_file:
f = open(self.error_file, 'a')
self.error_writer = csv.writer(f, delimiter=' ')
conf = {'namespace': namespace}
self.account_client = AccountClient(conf)
self.container_client = ContainerClient(conf)
self.blob_client = BlobClient()
self.accounts_checked = 0
self.containers_checked = 0
self.objects_checked = 0
self.chunks_checked = 0
self.account_not_found = 0
self.container_not_found = 0
self.object_not_found = 0
self.chunk_not_found = 0
self.account_exceptions = 0
self.container_exceptions = 0
self.object_exceptions = 0
self.chunk_exceptions = 0
self.list_cache = {}
self.running = {}
def write_error(self, target):
error = [target.account]
if target.container:
error.append(target.container)
if target.obj:
error.append(target.obj)
if target.chunk:
error.append(target.chunk)
self.error_writer.writerow(error)
def check_chunk(self, target):
chunk = target.chunk
obj_listing = self.check_obj(target)
error = False
if chunk not in obj_listing:
print(' Chunk %s missing in object listing' % target)
error = True
# checksum = None
else:
# TODO check checksum match
# checksum = obj_listing[chunk]['hash']
pass
try:
self.blob_client.chunk_head(chunk)
except exc.NotFound as e:
self.chunk_not_found += 1
error = True
print(' Not found chunk "%s": %s' % (target, str(e)))
except Exception as e:
self.chunk_exceptions += 1
error = True
print(' Exception chunk "%s": %s' % (target, str(e)))
if error and self.error_file:
self.write_error(target)
self.chunks_checked += 1
def check_obj(self, target, recurse=False):
account = target.account
container = target.container
obj = target.obj
if (account, container, obj) in self.running:
self.running[(account, container, obj)].wait()
if (account, container, obj) in self.list_cache:
return self.list_cache[(account, container, obj)]
self.running[(account, container, obj)] = Event()
print('Checking object "%s"' % target)
container_listing = self.check_container(target)
error = False
if obj not in container_listing:
print(' Object %s missing in container listing' % target)
error = True
# checksum = None
else:
# TODO check checksum match
# checksum = container_listing[obj]['hash']
pass
results = []
try:
_, resp = self.container_client.content_show(acct=account,
ref=container,
path=obj)
except exc.NotFound as e:
self.object_not_found += 1
error = True
print(' Not found object "%s": %s' % (target, str(e)))
except Exception as e:
self.object_exceptions += 1
error = True
print(' Exception object "%s": %s' % (target, str(e)))
else:
results = resp
chunk_listing = dict()
for chunk in results:
chunk_listing[chunk['url']] = chunk
self.objects_checked += 1
self.list_cache[(account, container, obj)] = chunk_listing
self.running[(account, container, obj)].send(True)
del self.running[(account, container, obj)]
if recurse:
for chunk in chunk_listing:
t = target.copy()
t.chunk = chunk
self.pool.spawn_n(self.check_chunk, t)
if error and self.error_file:
self.write_error(target)
return chunk_listing
def check_container(self, target, recurse=False):
account = target.account
container = target.container
if (account, container) in self.running:
self.running[(account, container)].wait()
if (account, container) in self.list_cache:
return self.list_cache[(account, container)]
self.running[(account, container)] = Event()
print('Checking container "%s"' % target)
account_listing = self.check_account(target)
error = False
if container not in account_listing:
error = True
print(' Container %s missing in account listing' % target)
marker = None
results = []
while True:
try:
resp = self.container_client.container_list(acct=account,
ref=container,
marker=marker)
except exc.NotFound as e:
self.container_not_found += 1
error = True
print(' Not found container "%s": %s' % (target, str(e)))
break
except Exception as e:
self.container_exceptions += 1
error = True
print(' Exception container "%s": %s' % (target, str(e)))
break
if resp['objects']:
marker = resp['objects'][-1]['name']
else:
break
results.extend(resp['objects'])
container_listing = dict()
for obj in results:
container_listing[obj['name']] = obj
self.containers_checked += 1
self.list_cache[(account, container)] = container_listing
self.running[(account, container)].send(True)
del self.running[(account, container)]
if recurse:
for obj in container_listing:
t = target.copy()
t.obj = obj
self.pool.spawn_n(self.check_obj, t, True)
if error and self.error_file:
self.write_error(target)
return container_listing
def check_account(self, target, recurse=False):
account = target.account
if account in self.running:
self.running[account].wait()
if account in self.list_cache:
return self.list_cache[account]
self.running[account] = Event()
print('Checking account "%s"' % target)
error = False
marker = None
results = []
while True:
try:
resp = self.account_client.containers_list(account,
marker=marker)
except Exception as e:
self.account_exceptions += 1
error = True
print(' Exception account "%s": %s' % (target, str(e)))
break
if resp['listing']:
marker = resp['listing'][-1][0]
else:
break
results.extend(resp['listing'])
containers = dict()
for e in results:
containers[e[0]] = (e[1], e[2])
self.list_cache[account] = containers
self.running[account].send(True)
del self.running[account]
self.accounts_checked += 1
if recurse:
for container in containers:
t = target.copy()
t.container = container
self.pool.spawn_n(self.check_container, t, True)
if error and self.error_file:
self.write_error(target)
return containers
def check(self, target):
if target.chunk and target.obj and target.container:
self.pool.spawn_n(self.check_chunk, target)
elif target.obj and target.container:
self.pool.spawn_n(self.check_obj, target, True)
elif target.container:
self.pool.spawn_n(self.check_container, target, True)
else:
self.pool.spawn_n(self.check_account, target, True)
def wait(self):
self.pool.waitall()
def report(self):
def _report_stat(name, stat):
print("{0:18}: {1}".format(name, stat))
print()
print('Report')
_report_stat("Accounts checked", self.accounts_checked)
if self.account_not_found:
_report_stat("Missing accounts", self.account_not_found)
if self.account_exceptions:
_report_stat("Exceptions", self.account_not_found)
print()
_report_stat("Containers checked", self.containers_checked)
if self.container_not_found:
_report_stat("Missing containers", self.container_not_found)
if self.container_exceptions:
_report_stat("Exceptions", self.container_exceptions)
print()
_report_stat("Objects checked", self.objects_checked)
if self.object_not_found:
_report_stat("Missing objects", self.object_not_found)
if self.object_exceptions:
_report_stat("Exceptions", self.object_exceptions)
print()
_report_stat("Chunks checked", self.chunks_checked)
if self.chunk_not_found:
_report_stat("Missing chunks", self.chunk_not_found)
if self.chunk_exceptions:
_report_stat("Exceptions", self.chunk_exceptions)
class RawxDecommissionTask(XcuteTask):
def __init__(self, conf, job_params, logger=None):
super(RawxDecommissionTask, self).__init__(conf,
job_params,
logger=logger)
self.service_id = job_params['service_id']
self.rawx_timeout = job_params['rawx_timeout']
self.min_chunk_size = job_params['min_chunk_size']
self.max_chunk_size = job_params['max_chunk_size']
self.excluded_rawx = job_params['excluded_rawx']
self.blob_client = BlobClient(self.conf, logger=self.logger)
self.content_factory = ContentFactory(self.conf)
self.conscience_client = ConscienceClient(self.conf,
logger=self.logger)
self.fake_excluded_chunks = self._generate_fake_excluded_chunks(
self.excluded_rawx)
def _generate_fake_excluded_chunks(self, excluded_rawx):
fake_excluded_chunks = list()
fake_chunk_id = '0' * 64
for service_id in excluded_rawx:
service_addr = self.conscience_client.resolve_service_id(
'rawx', service_id)
chunk = dict()
chunk['hash'] = '0000000000000000000000000000000000'
chunk['pos'] = '0'
chunk['size'] = 1
chunk['score'] = 1
chunk['url'] = 'http://{}/{}'.format(service_id, fake_chunk_id)
chunk['real_url'] = 'http://{}/{}'.format(service_addr,
fake_chunk_id)
fake_excluded_chunks.append(chunk)
return fake_excluded_chunks
def process(self, task_id, task_payload, reqid=None):
container_id = task_payload['container_id']
content_id = task_payload['content_id']
chunk_id = task_payload['chunk_id']
chunk_url = 'http://{}/{}'.format(self.service_id, chunk_id)
try:
meta = self.blob_client.chunk_head(chunk_url,
timeout=self.rawx_timeout,
reqid=reqid)
except NotFound:
# The chunk is still present in the rdir,
# but the chunk no longer exists in the rawx.
# We ignore it because there is nothing to move.
return {'skipped_chunks_no_longer_exist': 1}
if container_id != meta['container_id']:
raise ValueError('Mismatch container ID: %s != %s', container_id,
meta['container_id'])
if content_id != meta['content_id']:
raise ValueError('Mismatch content ID: %s != %s', content_id,
meta['content_id'])
chunk_size = int(meta['chunk_size'])
# Maybe skip the chunk because it doesn't match the size constaint
if chunk_size < self.min_chunk_size:
self.logger.debug('[reqid=%s] SKIP %s too small', reqid, chunk_url)
return {'skipped_chunks_too_small': 1}
if self.max_chunk_size > 0 and chunk_size > self.max_chunk_size:
self.logger.debug('[reqid=%s] SKIP %s too big', reqid, chunk_url)
return {'skipped_chunks_too_big': 1}
# Start moving the chunk
try:
content = self.content_factory.get(container_id,
content_id,
reqid=reqid)
content.move_chunk(chunk_id,
fake_excluded_chunks=self.fake_excluded_chunks,
reqid=reqid)
except (ContentNotFound, OrphanChunk):
return {'orphan_chunks': 1}
return {'moved_chunks': 1, 'moved_bytes': chunk_size}
class ContentRepairerWorker(ToolWorker):
def __init__(self, tool, queue_workers, queue_reply):
super(ContentRepairerWorker, self).__init__(tool, queue_workers,
queue_reply)
self.chunk_operator = ChunkOperator(self.conf, logger=self.logger)
self.blob_client = BlobClient(self.conf)
self.container_client = ContainerClient(self.conf, logger=self.logger)
def _safe_chunk_rebuild(self, item, content_id, chunk_id_or_pos, **kwargs):
_, account, container, _, _ = item
try:
container_id = cid_from_name(account, container)
self.chunk_operator.rebuild(container_id, content_id,
chunk_id_or_pos, **kwargs)
except Exception as exc: # pylint: disable=broad-except
self.logger.error('ERROR when rebuilding chunk %s (%s): %s',
self.tool.string_from_item(item),
chunk_id_or_pos, exc)
return exc
def _repair_metachunk(self, item, content_id, stg_met, pos, chunks):
"""
Check that a metachunk has the right number of chunks.
:returns: the list (generator) of missing chunks
"""
exceptions = list()
required = stg_met.expected_chunks
if len(chunks) < required:
if stg_met.ec:
subs = {x['num'] for x in chunks}
for sub in range(required):
if sub not in subs:
exc = self._safe_chunk_rebuild(item, content_id,
"%d.%d" % (pos, sub))
if exc:
exceptions.append(exc)
else:
missing_chunks = required - len(chunks)
for _ in range(missing_chunks):
exc = self._safe_chunk_rebuild(item, content_id, pos)
if exc:
exceptions.append(exc)
for chunk in chunks:
try:
self.blob_client.chunk_head(chunk['url'],
xattr=True,
check_hash=True)
except (NotFound, ClientPreconditionFailed) as e:
kwargs = {
'try_chunk_delete': isinstance(e, ClientPreconditionFailed)
}
exc = self._safe_chunk_rebuild(item, content_id, chunk['url'],
**kwargs)
if exc:
exceptions.append(exc)
except Exception as exc: # pylint: disable=broad-except
self.logger.error('ERROR when checking chunk %s (%s): %s',
self.tool.string_from_item(item),
chunk['url'], exc)
exceptions.append(exc)
return exceptions
def _process_item(self, item):
namespace, account, container, obj_name, version = item
if namespace != self.tool.namespace:
raise ValueError('Invalid namespace (actual=%s, expected=%s)' %
(namespace, self.tool.namespace))
obj_meta, chunks = self.container_client.content_locate(
account=account,
reference=container,
path=obj_name,
version=version,
properties=False)
content_id = obj_meta['id']
exceptions = list()
stg_met = STORAGE_METHODS.load(obj_meta['chunk_method'])
chunks_by_pos = _sort_chunks(chunks, stg_met.ec)
for pos, chunks in iteritems(chunks_by_pos):
try:
exceptions += self._repair_metachunk(item, content_id, stg_met,
pos, chunks)
except Exception as exc: # pylint: disable=broad-except
self.logger.error('ERROR when repair metachunk %s (%d): %s',
self.tool.string_from_item(item), pos, exc)
exceptions.append(exc)
if exceptions:
raise Exception(exceptions)
self.container_client.content_touch(account=account,
reference=container,
path=obj_name,
version=version)
class Checker(object):
def __init__(self,
namespace,
concurrency=50,
error_file=None,
rebuild_file=None,
full=True,
limit_listings=0,
request_attempts=1):
self.pool = GreenPool(concurrency)
self.error_file = error_file
self.full = bool(full)
# Optimisation for when we are only checking one object
# or one container.
# 0 -> do not limit
# 1 -> limit account listings (list of containers)
# 2 -> limit container listings (list of objects)
self.limit_listings = limit_listings
if self.error_file:
f = open(self.error_file, 'a')
self.error_writer = csv.writer(f, delimiter=' ')
self.rebuild_file = rebuild_file
if self.rebuild_file:
fd = open(self.rebuild_file, 'a')
self.rebuild_writer = csv.writer(fd, delimiter='|')
conf = {'namespace': namespace}
self.account_client = AccountClient(conf,
max_retries=request_attempts - 1)
self.container_client = ContainerClient(
conf,
max_retries=request_attempts - 1,
request_attempts=request_attempts)
self.blob_client = BlobClient(conf=conf)
self.accounts_checked = 0
self.containers_checked = 0
self.objects_checked = 0
self.chunks_checked = 0
self.account_not_found = 0
self.container_not_found = 0
self.object_not_found = 0
self.chunk_not_found = 0
self.account_exceptions = 0
self.container_exceptions = 0
self.object_exceptions = 0
self.chunk_exceptions = 0
self.list_cache = {}
self.running = {}
def write_error(self, target, irreparable=False):
error = list()
if irreparable:
error.append('#IRREPARABLE')
error.append(target.account)
if target.container:
error.append(target.container)
if target.obj:
error.append(target.obj)
if target.chunk:
error.append(target.chunk)
self.error_writer.writerow(error)
def write_rebuilder_input(self, target, obj_meta, irreparable=False):
ct_meta = self.list_cache[(target.account, target.container)][1]
try:
cid = ct_meta['system']['sys.name'].split('.', 1)[0]
except KeyError:
cid = ct_meta['properties']['sys.name'].split('.', 1)[0]
error = list()
if irreparable:
error.append('#IRREPARABLE')
error.append(cid)
error.append(obj_meta['id'])
error.append(target.chunk)
self.rebuild_writer.writerow(error)
def write_chunk_error(self,
target,
obj_meta,
chunk=None,
irreparable=False):
if chunk is not None:
target = target.copy()
target.chunk = chunk
if self.error_file:
self.write_error(target, irreparable=irreparable)
if self.rebuild_file:
self.write_rebuilder_input(target,
obj_meta,
irreparable=irreparable)
def _check_chunk_xattr(self, target, obj_meta, xattr_meta):
error = False
# Composed position -> erasure coding
attr_prefix = 'meta' if '.' in obj_meta['pos'] else ''
attr_key = attr_prefix + 'chunk_size'
if str(obj_meta['size']) != xattr_meta.get(attr_key):
print(
" Chunk %s '%s' xattr (%s) "
"differs from size in meta2 (%s)" %
(target, attr_key, xattr_meta.get(attr_key), obj_meta['size']))
error = True
attr_key = attr_prefix + 'chunk_hash'
if obj_meta['hash'] != xattr_meta.get(attr_key):
print(
" Chunk %s '%s' xattr (%s) "
"differs from hash in meta2 (%s)" %
(target, attr_key, xattr_meta.get(attr_key), obj_meta['hash']))
error = True
return error
def _check_chunk(self, target):
chunk = target.chunk
obj_listing, obj_meta = self.check_obj(target)
error = False
if chunk not in obj_listing:
print(' Chunk %s missing from object listing' % target)
error = True
db_meta = dict()
else:
db_meta = obj_listing[chunk]
try:
xattr_meta = self.blob_client.chunk_head(chunk, xattr=self.full)
except exc.NotFound as e:
self.chunk_not_found += 1
error = True
print(' Not found chunk "%s": %s' % (target, str(e)))
except Exception as e:
self.chunk_exceptions += 1
error = True
print(' Exception chunk "%s": %s' % (target, str(e)))
else:
if db_meta and self.full:
error = self._check_chunk_xattr(target, db_meta, xattr_meta)
self.chunks_checked += 1
return error, obj_meta
def check_chunk(self, target):
error, obj_meta = self._check_chunk(target)
if error:
self.write_chunk_error(target, obj_meta)
def _check_metachunk(self,
target,
obj_meta,
stg_met,
pos,
chunks,
recurse=False):
required = stg_met.expected_chunks
chunk_errors = list()
if len(chunks) < required:
missing_chunks = required - len(chunks)
print(' Missing %d chunks at position %s of %s' %
(missing_chunks, pos, target))
if stg_met.ec:
subs = {x['num'] for x in chunks}
for sub in range(required):
if sub not in subs:
chunk_errors.append(
(target, obj_meta, '%d.%d' % (pos, sub)))
else:
for _ in range(missing_chunks):
chunk_errors.append((target, obj_meta, str(pos)))
if recurse:
for chunk in chunks:
t = target.copy()
t.chunk = chunk['url']
error, obj_meta = self._check_chunk(t)
if error:
chunk_errors.append((t, obj_meta))
irreparable = required - len(chunk_errors) < stg_met.min_chunks_to_read
for chunk_error in chunk_errors:
self.write_chunk_error(*chunk_error, irreparable=irreparable)
def _check_obj_policy(self, target, obj_meta, chunks, recurse=False):
"""
Check that the list of chunks of an object matches
the object's storage policy.
"""
stg_met = STORAGE_METHODS.load(obj_meta['chunk_method'])
chunks_by_pos = _sort_chunks(chunks, stg_met.ec)
for pos, chunks in chunks_by_pos.iteritems():
self.pool.spawn_n(self._check_metachunk,
target.copy(),
obj_meta,
stg_met,
pos,
chunks,
recurse=recurse)
def check_obj(self, target, recurse=False):
account = target.account
container = target.container
obj = target.obj
if (account, container, obj) in self.running:
self.running[(account, container, obj)].wait()
if (account, container, obj) in self.list_cache:
return self.list_cache[(account, container, obj)]
self.running[(account, container, obj)] = Event()
print('Checking object "%s"' % target)
container_listing, ct_meta = self.check_container(target)
error = False
if obj not in container_listing:
print(' Object %s missing from container listing' % target)
error = True
# checksum = None
else:
# TODO check checksum match
# checksum = container_listing[obj]['hash']
pass
results = []
meta = dict()
try:
meta, results = self.container_client.content_locate(
account=account,
reference=container,
path=obj,
properties=False)
except exc.NotFound as e:
self.object_not_found += 1
error = True
print(' Not found object "%s": %s' % (target, str(e)))
except Exception as e:
self.object_exceptions += 1
error = True
print(' Exception object "%s": %s' % (target, str(e)))
chunk_listing = dict()
for chunk in results:
chunk_listing[chunk['url']] = chunk
if meta:
self.list_cache[(account, container, obj)] = (chunk_listing, meta)
self.objects_checked += 1
self.running[(account, container, obj)].send(True)
del self.running[(account, container, obj)]
# Skip the check if we could not locate the object
if meta:
self._check_obj_policy(target, meta, results, recurse=recurse)
if error and self.error_file:
self.write_error(target)
return chunk_listing, meta
def check_container(self, target, recurse=False):
account = target.account
container = target.container
if (account, container) in self.running:
self.running[(account, container)].wait()
if (account, container) in self.list_cache:
return self.list_cache[(account, container)]
self.running[(account, container)] = Event()
print('Checking container "%s"' % target)
account_listing = self.check_account(target)
error = False
if container not in account_listing:
error = True
print(' Container %s missing from account listing' % target)
marker = None
results = []
ct_meta = dict()
extra_args = dict()
if self.limit_listings > 1 and target.obj:
# When we are explicitly checking one object, start the listing
# where this object is supposed to be, and list only one object.
extra_args['prefix'] = target.obj
extra_args['limit'] = 1
while True:
try:
_, resp = self.container_client.content_list(
account=account,
reference=container,
marker=marker,
**extra_args)
except exc.NotFound as e:
self.container_not_found += 1
error = True
print(' Not found container "%s": %s' % (target, str(e)))
break
except Exception as e:
self.container_exceptions += 1
error = True
print(' Exception container "%s": %s' % (target, str(e)))
break
if resp['objects']:
marker = resp['objects'][-1]['name']
results.extend(resp['objects'])
if self.limit_listings > 1:
break
else:
ct_meta = resp
ct_meta.pop('objects')
break
container_listing = dict()
for obj in results:
container_listing[obj['name']] = obj
if self.limit_listings <= 1:
# We just listed the whole container, keep the result in a cache
self.containers_checked += 1
self.list_cache[(account, container)] = container_listing, ct_meta
self.running[(account, container)].send(True)
del self.running[(account, container)]
if recurse:
for obj in container_listing:
t = target.copy()
t.obj = obj
self.pool.spawn_n(self.check_obj, t, True)
if error and self.error_file:
self.write_error(target)
return container_listing, ct_meta
def check_account(self, target, recurse=False):
account = target.account
if account in self.running:
self.running[account].wait()
if account in self.list_cache:
return self.list_cache[account]
self.running[account] = Event()
print('Checking account "%s"' % target)
error = False
marker = None
results = []
extra_args = dict()
if self.limit_listings > 0 and target.container:
# When we are explicitly checking one container, start the listing
# where this container is supposed to be, and list only one
# container.
extra_args['prefix'] = target.container
extra_args['limit'] = 1
while True:
try:
resp = self.account_client.container_list(account,
marker=marker,
**extra_args)
except Exception as e:
self.account_exceptions += 1
error = True
print(' Exception account "%s": %s' % (target, str(e)))
break
if resp['listing']:
marker = resp['listing'][-1][0]
results.extend(resp['listing'])
if self.limit_listings > 0:
break
else:
break
containers = dict()
for e in results:
containers[e[0]] = (e[1], e[2])
if self.limit_listings <= 0:
# We just listed the whole account, keep the result in a cache
self.accounts_checked += 1
self.list_cache[account] = containers
self.running[account].send(True)
del self.running[account]
if recurse:
for container in containers:
t = target.copy()
t.container = container
self.pool.spawn_n(self.check_container, t, True)
if error and self.error_file:
self.write_error(target)
return containers
def check(self, target):
if target.chunk and target.obj and target.container:
self.pool.spawn_n(self.check_chunk, target)
elif target.obj and target.container:
self.pool.spawn_n(self.check_obj, target, True)
elif target.container:
self.pool.spawn_n(self.check_container, target, True)
else:
self.pool.spawn_n(self.check_account, target, True)
def wait(self):
self.pool.waitall()
def report(self):
success = True
def _report_stat(name, stat):
print("{0:18}: {1}".format(name, stat))
print()
print('Report')
_report_stat("Accounts checked", self.accounts_checked)
if self.account_not_found:
success = False
_report_stat("Missing accounts", self.account_not_found)
if self.account_exceptions:
success = False
_report_stat("Exceptions", self.account_exceptions)
print()
_report_stat("Containers checked", self.containers_checked)
if self.container_not_found:
success = False
_report_stat("Missing containers", self.container_not_found)
if self.container_exceptions:
success = False
_report_stat("Exceptions", self.container_exceptions)
print()
_report_stat("Objects checked", self.objects_checked)
if self.object_not_found:
success = False
_report_stat("Missing objects", self.object_not_found)
if self.object_exceptions:
success = False
_report_stat("Exceptions", self.object_exceptions)
print()
_report_stat("Chunks checked", self.chunks_checked)
if self.chunk_not_found:
success = False
_report_stat("Missing chunks", self.chunk_not_found)
if self.chunk_exceptions:
success = False
_report_stat("Exceptions", self.chunk_exceptions)
return success