def _downloader(self, pfn, protocol, human, input_queue, output_queue, user_agent, threadnb, total_threads, trace_endpoint, trace_pattern, transfer_timeout=None):
rse_dict = {}
thread_prefix = 'Thread %s/%s' % (threadnb, total_threads)
while True:
try:
file = input_queue.get_nowait()
except Empty:
return
dest_dir = file['dest_dir']
file_scope = file['scope']
file_name = file['name']
file_didstr = '%s:%s' % (file_scope, file_name)
# arguments for rsemgr.download already known
dlfile = {}
dlfile['name'] = file_name
dlfile['scope'] = file_scope
dlfile['adler32'] = file['adler32']
ignore_checksum = True if pfn else False
if pfn:
dlfile['pfn'] = pfn
logger.info('%s : Starting the download of %s' % (thread_prefix, file_didstr))
trace = deepcopy(trace_pattern)
trace.update({'scope': file_scope,
'filename': file_name,
'datasetScope': file['dataset_scope'],
'dataset': file['dataset_name'],
'filesize': file['bytes']})
rses = list(file['rses'].keys())
if rses == []:
logger.warning('%s : File %s has no available replicas. Cannot be downloaded.' % (thread_prefix, file_didstr))
trace['clientState'] = 'FILE_NOT_FOUND'
self.send_trace(trace, trace_endpoint, user_agent)
input_queue.task_done()
continue
random.shuffle(rses)
logger.debug('%s : Potential sources : %s' % (thread_prefix, str(rses)))
success = False
while not success and len(rses):
rse_name = rses.pop()
if rse_name not in rse_dict:
try:
rse_dict[rse_name] = rsemgr.get_rse_info(rse_name)
except RSENotFound:
logger.warning('%s : Could not get info of RSE %s' % (thread_prefix, rse_name))
continue
rse = rse_dict[rse_name]
if not rse['availability_read']:
logger.info('%s : %s is blacklisted for reading' % (thread_prefix, rse_name))
continue
try:
if pfn:
protocols = [rsemgr.select_protocol(rse, operation='read', scheme=pfn.split(':')[0])]
else:
protocols = rsemgr.get_protocols_ordered(rse, operation='read', scheme=protocol)
protocols.reverse()
except RSEProtocolNotSupported as error:
logger.info('%s : The protocol specfied (%s) is not supported by %s' % (thread_prefix, protocol, rse_name))
logger.debug(error)
continue
logger.debug('%s : %d possible protocol(s) for read' % (thread_prefix, len(protocols)))
trace['remoteSite'] = rse_name
trace['clientState'] = 'DOWNLOAD_ATTEMPT'
while not success and len(protocols):
protocol_retry = protocols.pop()
logger.debug('%s : Trying protocol %s at %s' % (thread_prefix, protocol_retry['scheme'], rse_name))
trace['protocol'] = protocol_retry['scheme']
out = {}
out['dataset_scope'] = file['dataset_scope']
out['dataset_name'] = file['dataset_name']
out['scope'] = file_scope
out['name'] = file_name
attempt = 0
retries = 2
while not success and attempt < retries:
attempt += 1
out['attemptnr'] = attempt
logger.info('%s : File %s trying from %s' % (thread_prefix, file_didstr, rse_name))
try:
trace['transferStart'] = time.time()
rsemgr.download(rse,
files=[dlfile],
dest_dir=dest_dir,
force_scheme=protocol_retry['scheme'],
ignore_checksum=ignore_checksum,
transfer_timeout=transfer_timeout)
trace['transferEnd'] = time.time()
trace['clientState'] = 'DONE'
out['clientState'] = 'DONE'
success = True
output_queue.put(out)
logger.info('%s : File %s successfully downloaded from %s' % (thread_prefix, file_didstr, rse_name))
except KeyboardInterrupt:
logger.warning('You pressed Ctrl+C! Exiting gracefully')
os.kill(os.getpgid(), signal.SIGINT)
return
except FileConsistencyMismatch as error:
logger.warning(str(error))
try:
pfns_dict = rsemgr.lfns2pfns(rse,
lfns=[{'name': file_name, 'scope': file_scope}],
operation='read',
scheme=protocol)
pfn = pfns_dict[file_didstr]
out['clientState'] = 'CORRUPTED'
out['pfn'] = pfn
output_queue.put(out)
except Exception as error:
logger.debug('%s : %s' % (thread_prefix, str(error)))
trace['clientState'] = 'FAIL_VALIDATE'
logger.debug('%s : Failed attempt %s/%s' % (thread_prefix, attempt, retries))
except Exception as error:
logger.warning(str(error))
trace['clientState'] = str(type(error).__name__)
logger.debug('%s : Failed attempt %s/%s' % (thread_prefix, attempt, retries))
self.send_trace(trace, trace_endpoint, user_agent, threadnb=threadnb, total_threads=total_threads)
if success:
duration = round(trace['transferEnd'] - trace['transferStart'], 2)
if pfn:
logger.info('%s : File %s successfully downloaded in %s seconds' % (thread_prefix, file_didstr, duration))
else:
logger.info('%s : File %s successfully downloaded. %s in %s seconds = %s MBps' % (thread_prefix,
file_didstr,
sizefmt(file['bytes'], human),
duration,
round((file['bytes'] / duration) * 1e-6, 2)))
else:
logger.error('%s : Cannot download file %s' % (thread_prefix, file_didstr))
input_queue.task_done()
def upload(self, items, summary_file_path=None, traces_copy_out=None):
"""
:param items: List of dictionaries. Each dictionary describing a file to upload. Keys:
path - path of the file that will be uploaded
rse - rse expression/name (e.g. 'CERN-PROD_DATADISK') where to upload the file
did_scope - Optional: custom did scope (Default: user.<account>)
did_name - Optional: custom did name (Default: name of the file)
dataset_scope - Optional: custom dataset scope
dataset_name - Optional: custom dataset name
force_scheme - Optional: force a specific scheme (if PFN upload this will be overwritten) (Default: None)
pfn - Optional: use a given PFN (this sets no_register to True, and no_register becomes mandatory)
no_register - Optional: if True, the file will not be registered in the rucio catalogue
register_after_upload - Optional: if True, the file will be registered after successful upload
lifetime - Optional: the lifetime of the file after it was uploaded
transfer_timeout - Optional: time after the upload will be aborted
guid - Optional: guid of the file
recursive - Optional: if set, parses the folder structure recursively into collections
:param summary_file_path: Optional: a path where a summary in form of a json file will be stored
:param traces_copy_out: reference to an external list, where the traces should be uploaded
:returns: 0 on success
:raises InputValidationError: if any input arguments are in a wrong format
:raises RSEWriteBlocked: if a given RSE is not available for writing
:raises NoFilesUploaded: if no files were successfully uploaded
:raises NotAllFilesUploaded: if not all files were successfully uploaded
"""
# helper to get rse from rse_expression:
def _pick_random_rse(rse_expression):
rses = [r['rse'] for r in self.client.list_rses(rse_expression)
] # can raise InvalidRSEExpression
random.shuffle(rses)
return rses[0]
logger = self.logger
self.trace['uuid'] = generate_uuid()
# check given sources, resolve dirs into files, and collect meta infos
files = self._collect_and_validate_file_info(items)
logger(
logging.DEBUG,
'Num. of files that upload client is processing: {}'.format(
len(files)))
# check if RSE of every file is available for writing
# and cache rse settings
registered_dataset_dids = set()
registered_file_dids = set()
rse_expression = None
for file in files:
rse_expression = file['rse']
rse = self.rse_expressions.setdefault(
rse_expression, _pick_random_rse(rse_expression))
if not self.rses.get(rse):
rse_settings = self.rses.setdefault(
rse, rsemgr.get_rse_info(rse, vo=self.client.vo))
if rse_settings['availability_write'] != 1:
raise RSEWriteBlocked(
'%s is not available for writing. No actions have been taken'
% rse)
dataset_scope = file.get('dataset_scope')
dataset_name = file.get('dataset_name')
file['rse'] = rse
if dataset_scope and dataset_name:
dataset_did_str = ('%s:%s' % (dataset_scope, dataset_name))
file['dataset_did_str'] = dataset_did_str
registered_dataset_dids.add(dataset_did_str)
registered_file_dids.add('%s:%s' %
(file['did_scope'], file['did_name']))
wrong_dids = registered_file_dids.intersection(registered_dataset_dids)
if len(wrong_dids):
raise InputValidationError(
'DIDs used to address both files and datasets: %s' %
str(wrong_dids))
logger(logging.DEBUG, 'Input validation done.')
# clear this set again to ensure that we only try to register datasets once
registered_dataset_dids = set()
num_succeeded = 0
summary = []
for file in files:
basename = file['basename']
logger(logging.INFO, 'Preparing upload for file %s' % basename)
no_register = file.get('no_register')
register_after_upload = file.get(
'register_after_upload') and not no_register
pfn = file.get('pfn')
force_scheme = file.get('force_scheme')
delete_existing = False
trace = copy.deepcopy(self.trace)
# appending trace to list reference, if the reference exists
if traces_copy_out is not None:
traces_copy_out.append(trace)
rse = file['rse']
trace['scope'] = file['did_scope']
trace['datasetScope'] = file.get('dataset_scope', '')
trace['dataset'] = file.get('dataset_name', '')
trace['remoteSite'] = rse
trace['filesize'] = file['bytes']
file_did = {'scope': file['did_scope'], 'name': file['did_name']}
dataset_did_str = file.get('dataset_did_str')
rse_settings = self.rses[rse]
rse_sign_service = rse_settings.get('sign_url', None)
is_deterministic = rse_settings.get('deterministic', True)
if not is_deterministic and not pfn:
logger(logging.ERROR,
'PFN has to be defined for NON-DETERMINISTIC RSE.')
continue
if pfn and is_deterministic:
logger(
logging.WARNING,
'Upload with given pfn implies that no_register is True, except non-deterministic RSEs'
)
no_register = True
# resolving local area networks
domain = 'wan'
rse_attributes = {}
try:
rse_attributes = self.client.list_rse_attributes(rse)
except:
logger(logging.WARNING,
'Attributes of the RSE: %s not available.' % rse)
if (self.client_location and 'lan' in rse_settings['domain']
and 'site' in rse_attributes):
if self.client_location['site'] == rse_attributes['site']:
domain = 'lan'
logger(logging.DEBUG,
'{} domain is used for the upload'.format(domain))
if not no_register and not register_after_upload:
self._register_file(file, registered_dataset_dids)
# if register_after_upload, file should be overwritten if it is not registered
# otherwise if file already exists on RSE we're done
if register_after_upload:
if rsemgr.exists(rse_settings,
pfn if pfn else file_did,
domain=domain,
auth_token=self.auth_token,
logger=logger):
try:
self.client.get_did(file['did_scope'],
file['did_name'])
logger(logging.INFO,
'File already registered. Skipping upload.')
trace['stateReason'] = 'File already exists'
continue
except DataIdentifierNotFound:
logger(
logging.INFO,
'File already exists on RSE. Previous left overs will be overwritten.'
)
delete_existing = True
elif not is_deterministic and not no_register:
if rsemgr.exists(rse_settings,
pfn,
domain=domain,
auth_token=self.auth_token,
logger=logger):
logger(
logging.INFO,
'File already exists on RSE with given pfn. Skipping upload. Existing replica has to be removed first.'
)
trace['stateReason'] = 'File already exists'
continue
elif rsemgr.exists(rse_settings,
file_did,
domain=domain,
auth_token=self.auth_token,
logger=logger):
logger(
logging.INFO,
'File already exists on RSE with different pfn. Skipping upload.'
)
trace['stateReason'] = 'File already exists'
continue
else:
if rsemgr.exists(rse_settings,
pfn if pfn else file_did,
domain=domain,
auth_token=self.auth_token,
logger=logger):
logger(logging.INFO,
'File already exists on RSE. Skipping upload')
trace['stateReason'] = 'File already exists'
continue
# protocol handling and upload
protocols = rsemgr.get_protocols_ordered(rse_settings=rse_settings,
operation='write',
scheme=force_scheme,
domain=domain)
protocols.reverse()
success = False
state_reason = ''
logger(logging.DEBUG, str(protocols))
while not success and len(protocols):
protocol = protocols.pop()
cur_scheme = protocol['scheme']
logger(logging.INFO,
'Trying upload with %s to %s' % (cur_scheme, rse))
lfn = {}
lfn['filename'] = basename
lfn['scope'] = file['did_scope']
lfn['name'] = file['did_name']
for checksum_name in GLOBALLY_SUPPORTED_CHECKSUMS:
if checksum_name in file:
lfn[checksum_name] = file[checksum_name]
lfn['filesize'] = file['bytes']
sign_service = None
if cur_scheme == 'https':
sign_service = rse_sign_service
trace['protocol'] = cur_scheme
trace['transferStart'] = time.time()
logger(logging.DEBUG,
'Processing upload with the domain: {}'.format(domain))
try:
pfn = self._upload_item(
rse_settings=rse_settings,
rse_attributes=rse_attributes,
lfn=lfn,
source_dir=file['dirname'],
domain=domain,
force_scheme=cur_scheme,
force_pfn=pfn,
transfer_timeout=file.get('transfer_timeout'),
delete_existing=delete_existing,
sign_service=sign_service)
logger(logging.DEBUG, 'Upload done.')
success = True
file['upload_result'] = {
0: True,
1: None,
'success': True,
'pfn': pfn
} # needs to be removed
except (ServiceUnavailable, ResourceTemporaryUnavailable,
RSEOperationNotSupported, RucioException) as error:
logger(logging.WARNING, 'Upload attempt failed')
logger(logging.INFO,
'Exception: %s' % str(error),
exc_info=True)
state_reason = str(error)
if success:
num_succeeded += 1
trace['transferEnd'] = time.time()
trace['clientState'] = 'DONE'
file['state'] = 'A'
logger(logging.INFO,
'Successfully uploaded file %s' % basename)
self._send_trace(trace)
if summary_file_path:
summary.append(copy.deepcopy(file))
if not no_register:
if register_after_upload:
self._register_file(file, registered_dataset_dids)
replica_for_api = self._convert_file_for_api(file)
if not self.client.update_replicas_states(
rse, files=[replica_for_api]):
logger(logging.WARNING,
'Failed to update replica state')
# add file to dataset if needed
if dataset_did_str and not no_register:
try:
self.client.attach_dids(file['dataset_scope'],
file['dataset_name'],
[file_did])
except Exception as error:
logger(logging.WARNING,
'Failed to attach file to the dataset')
logger(logging.DEBUG,
'Attaching to dataset {}'.format(str(error)))
else:
trace['clientState'] = 'FAILED'
trace['stateReason'] = state_reason
self._send_trace(trace)
logger(logging.ERROR, 'Failed to upload file %s' % basename)
if summary_file_path:
logger(logging.DEBUG,
'Summary will be available at {}'.format(summary_file_path))
final_summary = {}
for file in summary:
file_scope = file['did_scope']
file_name = file['did_name']
file_did_str = '%s:%s' % (file_scope, file_name)
final_summary[file_did_str] = {
'scope': file_scope,
'name': file_name,
'bytes': file['bytes'],
'rse': file['rse'],
'pfn': file['upload_result'].get('pfn', ''),
'guid': file['meta']['guid']
}
for checksum_name in GLOBALLY_SUPPORTED_CHECKSUMS:
if checksum_name in file:
final_summary[file_did_str][checksum_name] = file[
checksum_name]
with open(summary_file_path, 'w') as summary_file:
json.dump(final_summary,
summary_file,
sort_keys=True,
indent=1)
if num_succeeded == 0:
raise NoFilesUploaded()
elif num_succeeded != len(files):
raise NotAllFilesUploaded()
return 0
def upload(self, items, summary_file_path=None):
"""
:param items: List of dictionaries. Each dictionary describing a file to upload. Keys:
path - path of the file that will be uploaded
rse - rse name (e.g. 'CERN-PROD_DATADISK') where to upload the file
did_scope - Optional: custom did scope (Default: user.<account>)
did_name - Optional: custom did name (Default: name of the file)
dataset_scope - Optional: custom dataset scope
dataset_name - Optional: custom dataset name
force_scheme - Optional: force a specific scheme (if PFN upload this will be overwritten) (Default: None)
pfn - Optional: use a given PFN (this sets no_register to True)
no_register - Optional: if True, the file will not be registered in the rucio catalogue
lifetime - Optional: the lifetime of the file after it was uploaded
transfer_timeout - Optional: time after the upload will be aborted
guid - Optional: guid of the file
:param summary_file_path: Optional: a path where a summary in form of a json file will be stored
:returns: 0 on success
:raises InputValidationError: if any input arguments are in a wrong format
:raises RSEBlacklisted: if a given RSE is not available for writing
:raises NoFilesUploaded: if no files were successfully uploaded
:raises NotAllFilesUploaded: if not all files were successfully uploaded
"""
logger = self.logger
self.trace['uuid'] = generate_uuid()
# check given sources, resolve dirs into files, and collect meta infos
files = self._collect_and_validate_file_info(items)
# check if RSE of every file is available for writing
# and cache rse settings
registered_dataset_dids = set()
registered_file_dids = set()
for file in files:
rse = file['rse']
if not self.rses.get(rse):
rse_settings = self.rses.setdefault(rse,
rsemgr.get_rse_info(rse))
if rse_settings['availability_write'] != 1:
raise RSEBlacklisted(
'%s is blacklisted for writing. No actions have been taken'
% rse)
dataset_scope = file.get('dataset_scope')
dataset_name = file.get('dataset_name')
if dataset_scope and dataset_name:
dataset_did_str = ('%s:%s' % (dataset_scope, dataset_name))
file['dataset_did_str'] = dataset_did_str
registered_dataset_dids.add(dataset_did_str)
registered_file_dids.add('%s:%s' %
(file['did_scope'], file['did_name']))
wrong_dids = registered_file_dids.intersection(registered_dataset_dids)
if len(wrong_dids):
raise InputValidationError(
'DIDs used to address both files and datasets: %s' %
str(wrong_dids))
# clear this set again to ensure that we only try to register datasets once
registered_dataset_dids = set()
num_succeeded = 0
for file in files:
basename = file['basename']
logger.info('Preparing upload for file %s' % basename)
no_register = file.get('no_register')
pfn = file.get('pfn')
force_scheme = file.get('force_scheme')
self.trace['scope'] = file['did_scope']
self.trace['datasetScope'] = file.get('dataset_scope', '')
self.trace['dataset'] = file.get('dataset_name', '')
self.trace['remoteSite'] = rse
self.trace['filesize'] = file['bytes']
file_did = {'scope': file['did_scope'], 'name': file['did_name']}
dataset_did_str = file.get('dataset_did_str')
if not no_register:
self._register_file(file, registered_dataset_dids)
rse = file['rse']
rse_settings = self.rses[rse]
# if file already exists on RSE we're done
if rsemgr.exists(rse_settings, file_did):
logger.info('File already exists on RSE. Skipping upload')
continue
protocols = rsemgr.get_protocols_ordered(rse_settings=rse_settings,
operation='write',
scheme=force_scheme)
protocols.reverse()
success = False
summary = []
while not success and len(protocols):
protocol = protocols.pop()
cur_scheme = protocol['scheme']
logger.info('Trying upload with %s to %s' % (cur_scheme, rse))
lfn = {}
lfn['filename'] = basename
lfn['scope'] = file['did_scope']
lfn['name'] = file['did_name']
lfn['adler32'] = file['adler32']
lfn['filesize'] = file['bytes']
self.trace['protocol'] = cur_scheme
self.trace['transferStart'] = time.time()
try:
state = rsemgr.upload(
rse_settings=rse_settings,
lfns=lfn,
source_dir=file['dirname'],
force_scheme=cur_scheme,
force_pfn=pfn,
transfer_timeout=file.get('transfer_timeout'))
success = True
file['upload_result'] = state
except (ServiceUnavailable,
ResourceTemporaryUnavailable) as error:
logger.warning('Upload attempt failed')
logger.debug('Exception: %s' % str(error))
if success:
num_succeeded += 1
self.trace['transferEnd'] = time.time()
self.trace['clientState'] = 'DONE'
file['state'] = 'A'
logger.info('Successfully uploaded file %s' % basename)
send_trace(self.trace, self.client.host,
self.client.user_agent)
if summary_file_path:
summary.append(copy.deepcopy(file))
# add file to dataset if needed
if dataset_did_str and not no_register:
try:
self.client.attach_dids(file['dataset_scope'],
file['dataset_name'],
[file_did])
except Exception as error:
logger.warning('Failed to attach file to the dataset')
logger.debug(error)
if not no_register:
replica_for_api = self._convert_file_for_api(file)
if not self.client.update_replicas_states(
rse, files=[replica_for_api]):
logger.warning('Failed to update replica state')
else:
logger.error('Failed to upload file %s' % basename)
if summary_file_path:
final_summary = {}
for file in summary:
file_scope = file['did_scope']
file_name = file['did_name']
file_did_str = '%s:%s' % (file_scope, file_name)
final_summary[file_did_str] = {
'scope': file['scope'],
'name': file['name'],
'bytes': file['bytes'],
'rse': file['rse'],
'pfn': file['upload_result']['pfn'],
'guid': file['meta']['guid'],
'adler32': file['adler32'],
'md5': file['md5']
}
with open(summary_file_path, 'wb') as summary_file:
json.dump(final_summary,
summary_file,
sort_keys=True,
indent=1)
if num_succeeded == 0:
raise NoFilesUploaded()
elif num_succeeded != len(files):
raise NotAllFilesUploaded()
return 0
def _download_item(self, item, trace, log_prefix=''):
"""
Downloads the given item and sends traces for success/failure.
(This function is meant to be used as class internal only)
:param item: dictionary that describes the item to download
:param trace: dictionary representing a pattern of trace that will be send
:param log_prefix: string that will be put at the beginning of every log message
:returns: dictionary with all attributes from the input item and a clientState attribute
"""
logger = self.logger
did_scope = item['scope']
did_name = item['name']
did_str = '%s:%s' % (did_scope, did_name)
logger.info('%sPreparing download of %s' % (log_prefix, did_str))
trace['scope'] = did_scope
trace['filename'] = did_name
trace.setdefault('dataset_scope', item.get('dataset_scope', ''))
trace.setdefault('dataset', item.get('dataset_name', ''))
trace.setdefault('filesize', item.get('bytes'))
# if file already exists, set state, send trace, and return
dest_dir_path = item['dest_dir_path']
dest_file_path = os.path.join(dest_dir_path, did_name)
if os.path.isfile(dest_file_path):
logger.info('%sFile exists already locally: %s' %
(log_prefix, did_str))
item['clientState'] = 'ALREADY_DONE'
trace['transferStart'] = time.time()
trace['transferEnd'] = time.time()
trace['clientState'] = 'ALREADY_DONE'
send_trace(trace, self.client.host, self.user_agent)
return item
# check if file has replicas
rse_names = list(item['rses'].keys())
if not len(rse_names):
logger.warning(
'%sFile %s has no available replicas. Cannot be downloaded' %
(log_prefix, did_str))
item['clientState'] = 'FILE_NOT_FOUND'
trace['clientState'] = 'FILE_NOT_FOUND'
send_trace(trace, self.client.host, self.user_agent)
return item
# list_replicas order is: best rse at [0]
rse_names.reverse()
logger.debug('%sPotential sources: %s' % (log_prefix, str(rse_names)))
success = False
# retry with different rses if one is not available or fails
while not success and len(rse_names):
rse_name = rse_names.pop()
try:
rse = rsemgr.get_rse_info(rse_name)
except RSENotFound:
logger.warning('%sCould not get info of RSE %s' %
(log_prefix, rse_name))
continue
if not rse['availability_read']:
logger.info('%s%s is blacklisted for reading' %
(log_prefix, rse_name))
continue
force_scheme = item.get('force_scheme')
try:
protocols = rsemgr.get_protocols_ordered(rse,
operation='read',
scheme=force_scheme)
protocols.reverse()
except RSEProtocolNotSupported as error:
logger.info(
'%sThe protocol specfied (%s) is not supported by %s' %
(log_prefix, force_scheme, rse_name))
logger.debug(error)
continue
logger.debug('%sPotential protocol(s) read: %s' %
(log_prefix, protocols))
trace['remoteSite'] = rse_name
trace['clientState'] = 'DOWNLOAD_ATTEMPT'
# retry with different protocols on the given rse
while not success and len(protocols):
protocol = protocols.pop()
cur_scheme = protocol['scheme']
trace['protocol'] = cur_scheme
logger.info('%sTrying to download with %s from %s: %s ' %
(log_prefix, cur_scheme, rse_name, did_str))
attempt = 0
retries = 2
# do some retries with the same rse and protocol if the download fails
while not success and attempt < retries:
attempt += 1
item['attemptnr'] = attempt
try:
start_time = time.time()
rsemgr.download(
rse,
files=item,
dest_dir=dest_dir_path,
force_scheme=cur_scheme,
ignore_checksum=item.get('ignore_checksum', False),
transfer_timeout=item.get('transfer_timeout'))
end_time = time.time()
trace['transferStart'] = start_time
trace['transferEnd'] = end_time
trace['clientState'] = 'DONE'
item['clientState'] = 'DONE'
success = True
except FileConsistencyMismatch as error:
logger.warning(str(error))
try:
pfn = item.get('pfn')
if not pfn:
pfns_dict = rsemgr.lfns2pfns(rse,
lfns={
'name':
did_name,
'scope':
did_scope
},
operation='read',
scheme=cur_scheme)
pfn = pfns_dict[did_str]
corrupted_item = copy.deepcopy(item)
corrupted_item['clientState'] = 'FAIL_VALIDATE'
corrupted_item['pfn'] = pfn
# self.corrupted_files.append(corrupted_item)
except Exception as error:
logger.debug('%s%s' % (log_prefix, str(error)))
trace['clientState'] = 'FAIL_VALIDATE'
except Exception as error:
logger.warning(str(error))
trace['clientState'] = str(type(error).__name__)
if not success:
logger.debug('%sFailed attempt %s/%s' %
(log_prefix, attempt, retries))
send_trace(trace, self.client.host, self.user_agent)
if not success:
logger.error('%sFailed to download file %s' %
(log_prefix, did_str))
item['clientState'] = 'FAILED'
return item
duration = round(end_time - start_time, 2)
size = item.get('bytes')
size_str = sizefmt(size, self.is_human_readable)
if size and duration:
rate = round((size / duration) * 1e-6, 2)
logger.info(
'%sFile %s successfully downloaded. %s in %s seconds = %s MBps'
% (log_prefix, did_str, size_str, duration, rate))
else:
logger.info('%sFile %s successfully downloaded in %s seconds' %
(log_prefix, did_str, duration))
return item
def upload(self, sources_with_settings, summary_file_path=None):
"""
List of dictionaries of file descriptions. None means optional
[{'path': 'file1',
'rse': 'rse_name1',
'did_scope': None,
'did_name': None,
'dataset_name': None,
'dataset_scope': None,
'scheme': None,
'pfn': None,
'no_register': None,
'lifetime': None },
{'path': 'file2',
'rse': 'rse_name2',
'did_scope': None,
'did_name': None,
'dataset_name': None,
'dataset_scope': None,
'scheme': None,
'pfn': None,
'no_register': None,
'lifetime': None }]
raises InputValidationError
raises RSEBlacklisted
"""
logger = self.logger
self.trace['uuid'] = generate_uuid()
# check given sources, resolve dirs into files, and collect meta infos
files = self.collect_and_validate_file_info(sources_with_settings)
# check if RSE of every file is available for writing
# and cache rse settings
registered_dataset_dids = set()
registered_file_dids = set()
for file in files:
rse = file['rse']
if not self.rses.get(rse):
rse_settings = self.rses.setdefault(rse, rsemgr.get_rse_info(rse))
if rse_settings['availability_write'] != 1:
raise RSEBlacklisted('%s is blacklisted for writing. No actions have been taken' % rse)
dataset_scope = file.get('dataset_scope')
dataset_name = file.get('dataset_name')
if dataset_scope and dataset_name:
dataset_did_str = ('%s:%s' % (dataset_scope, dataset_name))
file['dataset_did_str'] = dataset_did_str
registered_dataset_dids.add(dataset_did_str)
registered_file_dids.add('%s:%s' % (file['did_scope'], file['did_name']))
wrong_dids = registered_file_dids.intersection(registered_dataset_dids)
if len(wrong_dids):
raise InputValidationError('DIDs used to address both files and datasets: %s' % str(wrong_dids))
# clear this set again to ensure that we only try to register datasets once
registered_dataset_dids = set()
for file in files:
basename = file['basename']
logger.info('Preparing upload for file %s' % basename)
no_register = file.get('no_register')
pfn = file.get('pfn')
scheme = file.get('scheme')
self.trace['scope'] = file['did_scope']
self.trace['datasetScope'] = file.get('dataset_scope', '')
self.trace['dataset'] = file.get('dataset_name', '')
self.trace['remoteSite'] = rse
self.trace['filesize'] = file['bytes']
file_scope = file['did_scope']
file_name = file['did_name']
file_did = {'scope': file_scope, 'name': file_name}
file_did_str = '%s:%s' % (file_scope, file_name)
dataset_did_str = file.get('dataset_did_str')
rse = file['rse']
rse_settings = self.rses[rse]
# register a dataset if we need to
if dataset_did_str and dataset_did_str not in registered_dataset_dids and not no_register:
registered_dataset_dids.add(dataset_did_str)
try:
self.client.add_dataset(scope=file['dataset_scope'],
name=file['dataset_name'],
rules=[{'account': self.account,
'copies': 1,
'rse_expression': rse,
'grouping': 'DATASET',
'lifetime': file['lifetime']}])
logger.info('Dataset %s successfully created' % dataset_did_str)
except DataIdentifierAlreadyExists:
# TODO: Need to check the rules thing!!
logger.info("Dataset %s already exists" % dataset_did_str)
replica_for_api = self.convert_file_for_api(file)
try:
# if the remote checksum is different this did must not be used
meta = self.client.get_metadata(file_scope, file_name)
logger.info('Comparing checksums of %s and %s' % (basename, file_did_str))
if meta['adler32'] != file['adler32']:
logger.error('Local checksum %s does not match remote checksum %s' % (file['adler32'], meta['adler32']))
raise DataIdentifierAlreadyExists
# add file to rse if it is not registered yet
replicastate = list(self.client.list_replicas([file_did], all_states=True))
if rse not in replicastate[0]['rses'] and not no_register:
logger.info('Adding replica at %s in Rucio catalog' % rse)
self.client.add_replicas(rse=file['rse'], files=[replica_for_api])
except DataIdentifierNotFound:
if not no_register:
logger.info('Adding replica at %s in Rucio catalog' % rse)
self.client.add_replicas(rse=file['rse'], files=[replica_for_api])
if not dataset_did_str:
# only need to add rules for files if no dataset is given
logger.info('Adding replication rule at %s' % rse)
self.client.add_replication_rule([file_did], copies=1, rse_expression=rse, lifetime=file['lifetime'])
# if file already exists on RSE we're done
if not rsemgr.exists(rse_settings, file_did):
protocols = rsemgr.get_protocols_ordered(rse_settings=rse_settings, operation='write', scheme=scheme)
protocols.reverse()
success = False
summary = []
while not success and len(protocols):
protocol = protocols.pop()
logger.info('Trying upload to %s with protocol %s' % (rse, protocol['scheme']))
lfn = {}
lfn['filename'] = file['basename']
lfn['scope'] = file['did_scope']
lfn['name'] = file['did_name']
lfn['adler32'] = file['adler32']
lfn['filesize'] = file['bytes']
self.trace['protocol'] = protocol['scheme']
self.trace['transferStart'] = time.time()
try:
state = rsemgr.upload(rse_settings=rse_settings,
lfns=lfn,
source_dir=file['dirname'],
force_scheme=protocol['scheme'],
force_pfn=pfn)
success = True
file['upload_result'] = state
except (ServiceUnavailable, ResourceTemporaryUnavailable) as error:
logger.warning('Upload attempt failed')
logger.debug('Exception: %s' % str(error))
if success:
self.trace['transferEnd'] = time.time()
self.trace['clientState'] = 'DONE'
file['state'] = 'A'
logger.info('File %s successfully uploaded' % basename)
send_trace(self.trace, self.client.host, self.user_agent, logger=logger)
if summary_file_path:
summary.append(copy.deepcopy(file))
else:
logger.error('Failed to upload file %s' % basename)
# TODO trace?
continue # skip attach_did and update_states for this file
else:
logger.info('File already exists on RSE. Skipped upload')
if not no_register:
# add file to dataset if needed
if dataset_did_str:
try:
logger.info('Attaching file to dataset %s' % dataset_did_str)
self.client.attach_dids(file['dataset_scope'], file['dataset_name'], [file_did])
except Exception as error:
logger.warning('Failed to attach file to the dataset')
logger.warning(error)
logger.info('Setting replica state to available')
replica_for_api = self.convert_file_for_api(file)
self.client.update_replicas_states(rse, files=[replica_for_api])
if summary_file_path:
final_summary = {}
for file in summary:
file_scope = file['did_scope']
file_name = file['did_name']
file_did_str = '%s:%s' % (file_scope, file_name)
final_summary[file_did_str] = {'scope': file['scope'],
'name': file['name'],
'bytes': file['bytes'],
'rse': file['rse'],
'pfn': file['upload_result']['pfn'],
'guid': file['meta']['guid'],
'adler32': file['adler32'],
'md5': file['md5']}
with open(summary_file_path, 'wb') as summary_file:
json.dump(final_summary, summary_file, sort_keys=True, indent=1)
def upload(self, items, summary_file_path=None, traces_copy_out=None):
"""
:param items: List of dictionaries. Each dictionary describing a file to upload. Keys:
path - path of the file that will be uploaded
rse - rse name (e.g. 'CERN-PROD_DATADISK') where to upload the file
did_scope - Optional: custom did scope (Default: user.<account>)
did_name - Optional: custom did name (Default: name of the file)
dataset_scope - Optional: custom dataset scope
dataset_name - Optional: custom dataset name
force_scheme - Optional: force a specific scheme (if PFN upload this will be overwritten) (Default: None)
pfn - Optional: use a given PFN (this sets no_register to True, and no_register becomes mandatory)
no_register - Optional: if True, the file will not be registered in the rucio catalogue
register_after_upload - Optional: if True, the file will be registered after successful upload
lifetime - Optional: the lifetime of the file after it was uploaded
transfer_timeout - Optional: time after the upload will be aborted
guid - Optional: guid of the file
:param summary_file_path: Optional: a path where a summary in form of a json file will be stored
:param traces_copy_out: reference to an external list, where the traces should be uploaded
:returns: 0 on success
:raises InputValidationError: if any input arguments are in a wrong format
:raises RSEBlacklisted: if a given RSE is not available for writing
:raises NoFilesUploaded: if no files were successfully uploaded
:raises NotAllFilesUploaded: if not all files were successfully uploaded
"""
logger = self.logger
from admix.helper import helper
helper.global_dictionary['logger'].Info(' r: start')
self.trace['uuid'] = generate_uuid()
helper.global_dictionary['logger'].Info(' r: collect and validate')
# check given sources, resolve dirs into files, and collect meta infos
files = self._collect_and_validate_file_info(items)
# check if RSE of every file is available for writing
# and cache rse settings
registered_dataset_dids = set()
registered_file_dids = set()
helper.global_dictionary['logger'].Info(' r: check if files are available')
for file in files:
rse = file['rse']
if not self.rses.get(rse):
rse_settings = self.rses.setdefault(rse, rsemgr.get_rse_info(rse))
if rse_settings['availability_write'] != 1:
raise RSEBlacklisted('%s is blacklisted for writing. No actions have been taken' % rse)
dataset_scope = file.get('dataset_scope')
dataset_name = file.get('dataset_name')
if dataset_scope and dataset_name:
dataset_did_str = ('%s:%s' % (dataset_scope, dataset_name))
file['dataset_did_str'] = dataset_did_str
registered_dataset_dids.add(dataset_did_str)
registered_file_dids.add('%s:%s' % (file['did_scope'], file['did_name']))
wrong_dids = registered_file_dids.intersection(registered_dataset_dids)
if len(wrong_dids):
raise InputValidationError('DIDs used to address both files and datasets: %s' % str(wrong_dids))
# clear this set again to ensure that we only try to register datasets once
registered_dataset_dids = set()
num_succeeded = 0
summary = []
helper.global_dictionary['logger'].Info(' r: Starting loop')
for file in files:
helper.global_dictionary['logger'].Info('Start')
basename = file['basename']
logger.info('Preparing upload for file %s' % basename)
no_register = file.get('no_register')
register_after_upload = file.get('register_after_upload') and not no_register
pfn = file.get('pfn')
force_scheme = file.get('force_scheme')
delete_existing = False
helper.global_dictionary['logger'].Info('1')
trace = copy.deepcopy(self.trace)
helper.global_dictionary['logger'].Info('2')
# appending trace to list reference, if the reference exists
if traces_copy_out is not None:
helper.global_dictionary['logger'].Info('3')
traces_copy_out.append(trace)
helper.global_dictionary['logger'].Info('4')
trace['scope'] = file['did_scope']
trace['datasetScope'] = file.get('dataset_scope', '')
trace['dataset'] = file.get('dataset_name', '')
trace['remoteSite'] = rse
trace['filesize'] = file['bytes']
file_did = {'scope': file['did_scope'], 'name': file['did_name']}
dataset_did_str = file.get('dataset_did_str')
rse = file['rse']
rse_settings = self.rses[rse]
rse_sign_service = rse_settings.get('sign_url', None)
is_deterministic = rse_settings.get('deterministic', True)
if not is_deterministic and not pfn:
logger.error('PFN has to be defined for NON-DETERMINISTIC RSE.')
continue
if pfn and is_deterministic:
logger.warning('Upload with given pfn implies that no_register is True, except non-deterministic RSEs')
no_register = True
helper.global_dictionary['logger'].Info('5')
if not no_register and not register_after_upload:
helper.global_dictionary['logger'].Info('6')
self._register_file(file, registered_dataset_dids,helper)
helper.global_dictionary['logger'].Info('7')
# if register_after_upload, file should be overwritten if it is not registered
# otherwise if file already exists on RSE we're done
# if register_after_upload:
# helper.global_dictionary['logger'].Info('8')
# if rsemgr.exists(rse_settings, pfn if pfn else file_did):
# helper.global_dictionary['logger'].Info('9')
# try:
# helper.global_dictionary['logger'].Info('10')
# self.client.get_did(file['did_scope'], file['did_name'])
# logger.info('File already registered. Skipping upload.')
# trace['stateReason'] = 'File already exists'
# continue
# except DataIdentifierNotFound:
# logger.info('File already exists on RSE. Previous left overs will be overwritten.')
# delete_existing = True
# elif not is_deterministic and not no_register:
# if rsemgr.exists(rse_settings, pfn):
# logger.info('File already exists on RSE with given pfn. Skipping upload. Existing replica has to be removed first.')
# trace['stateReason'] = 'File already exists'
# continue
# elif rsemgr.exists(rse_settings, file_did):
# logger.info('File already exists on RSE with different pfn. Skipping upload.')
# trace['stateReason'] = 'File already exists'
# continue
# else:
# if rsemgr.exists(rse_settings, pfn if pfn else file_did):
# logger.info('File already exists on RSE. Skipping upload')
# trace['stateReason'] = 'File already exists'
# continue
helper.global_dictionary['logger'].Info('11')
# resolving local area networks
domain = 'wan'
# rse_attributes = {}
# try:
# rse_attributes = self.client.list_rse_attributes(rse)
# helper.global_dictionary['logger'].Info('12')
# except:
# logger.warning('Attributes of the RSE: %s not available.' % rse)
# if (self.client_location and 'lan' in rse_settings['domain'] and 'site' in rse_attributes):
# if self.client_location['site'] == rse_attributes['site']:
# domain = 'lan'
# protocol handling and upload
protocols = rsemgr.get_protocols_ordered(rse_settings=rse_settings, operation='write', scheme=force_scheme, domain=domain)
helper.global_dictionary['logger'].Info('14')
protocols.reverse()
helper.global_dictionary['logger'].Info('15')
success = False
state_reason = ''
while not success and len(protocols):
helper.global_dictionary['logger'].Info('16')
protocol = protocols.pop()
cur_scheme = protocol['scheme']
logger.info('Trying upload with %s to %s' % (cur_scheme, rse))
lfn = {}
lfn['filename'] = basename
lfn['scope'] = file['did_scope']
lfn['name'] = file['did_name']
for checksum_name in GLOBALLY_SUPPORTED_CHECKSUMS:
if checksum_name in file:
lfn[checksum_name] = file[checksum_name]
lfn['filesize'] = file['bytes']
sign_service = None
if cur_scheme == 'https':
sign_service = rse_sign_service
trace['protocol'] = cur_scheme
trace['transferStart'] = time.time()
try:
helper.global_dictionary['logger'].Info('17')
state = rsemgr.upload(rse_settings=rse_settings,
lfns=lfn,
source_dir=file['dirname'],
force_scheme=cur_scheme,
force_pfn=pfn,
transfer_timeout=file.get('transfer_timeout'),
delete_existing=delete_existing,
sign_service=sign_service)
helper.global_dictionary['logger'].Info('18')
success = state['success']
file['upload_result'] = state
except (ServiceUnavailable, ResourceTemporaryUnavailable) as error:
logger.warning('Upload attempt failed')
logger.debug('Exception: %s' % str(error))
state_reason = str(error)
helper.global_dictionary['logger'].Info('19')
if success:
num_succeeded += 1
trace['transferEnd'] = time.time()
trace['clientState'] = 'DONE'
file['state'] = 'A'
logger.info('Successfully uploaded file %s' % basename)
print('Successfully uploaded file %s' % basename)
self._send_trace(trace)
helper.global_dictionary['logger'].Info('Before if deepcopy')
if summary_file_path:
helper.global_dictionary['logger'].Info('Before deepcopy')
summary.append(copy.deepcopy(file))
helper.global_dictionary['logger'].Info('Before if register')
if not no_register:
helper.global_dictionary['logger'].Info('Before if2 register')
if register_after_upload:
helper.global_dictionary['logger'].Info('Before register')
self._register_file(file, registered_dataset_dids,helper)
replica_for_api = self._convert_file_for_api(file)
helper.global_dictionary['logger'].Info('Before if register2')
if not self.client.update_replicas_states(rse, files=[replica_for_api]):
helper.global_dictionary['logger'].Info('Before if register3')
logger.warning('Failed to update replica state')
# add file to dataset if needed
helper.global_dictionary['logger'].Info('Before if attach')
if dataset_did_str and not no_register:
try:
helper.global_dictionary['logger'].Info('Before attach')
self.client.attach_dids(file['dataset_scope'], file['dataset_name'], [file_did])
helper.global_dictionary['logger'].Info('After attach')
except Exception as error:
helper.global_dictionary['logger'].Info('Failed to attach file to the dataset')
logger.warning('Failed to attach file to the dataset')
logger.debug(error)
helper.global_dictionary['logger'].Info('Really finished')
else:
trace['clientState'] = 'FAILED'
trace['stateReason'] = state_reason
self._send_trace(trace)
logger.error('Failed to upload file %s' % basename)
if summary_file_path:
final_summary = {}
for file in summary:
file_scope = file['did_scope']
file_name = file['did_name']
file_did_str = '%s:%s' % (file_scope, file_name)
final_summary[file_did_str] = {'scope': file_scope,
'name': file_name,
'bytes': file['bytes'],
'rse': file['rse'],
'pfn': file['upload_result'].get('pfn', ''),
'guid': file['meta']['guid']}
for checksum_name in GLOBALLY_SUPPORTED_CHECKSUMS:
if checksum_name in file:
final_summary[file_did_str][checksum_name] = file[checksum_name]
with open(summary_file_path, 'wb') as summary_file:
json.dump(final_summary, summary_file, sort_keys=True, indent=1)
if num_succeeded == 0:
raise NoFilesUploaded()
elif num_succeeded != len(files):
raise NotAllFilesUploaded()
return 0
