def run(total_workers=1, once=False, inputfile=None, sleep_time=-1):
"""
Starts up the automatix threads.
"""
setup_logging()
if rucio.db.sqla.util.is_old_db():
raise exception.DatabaseException('Database was not updated, daemon won\'t start')
try:
sites = [s.strip() for s in config_get('automatix', 'sites').split(',')]
except (NoOptionError, NoSectionError, RuntimeError):
raise Exception('Could not load sites from configuration')
if not inputfile:
inputfile = '/opt/rucio/etc/automatix.json'
if sleep_time == -1:
try:
sleep_time = config_get('automatix', 'sleep_time')
except (NoOptionError, NoSectionError, RuntimeError):
sleep_time = 30
try:
account = config_get('automatix', 'account')
except (NoOptionError, NoSectionError, RuntimeError):
account = 'root'
try:
dataset_lifetime = config_get('automatix', 'dataset_lifetime')
except (NoOptionError, NoSectionError, RuntimeError):
dataset_lifetime = None
try:
set_metadata = config_get('automatix', 'set_metadata')
except (NoOptionError, NoSectionError, RuntimeError):
set_metadata = False
try:
scope = config_get('automatix', 'scope')
client = Client()
filters = {'scope': InternalScope('*', vo=client.vo)}
if InternalScope(scope, vo=client.vo) not in list_scopes(filter_=filters):
logging.log(logging.ERROR, 'Scope %s does not exist. Exiting', scope)
GRACEFUL_STOP.set()
except Exception:
scope = False
threads = list()
for worker_number in range(0, total_workers):
kwargs = {'worker_number': worker_number,
'total_workers': total_workers,
'once': once,
'sites': sites,
'sleep_time': sleep_time,
'account': account,
'inputfile': inputfile,
'set_metadata': set_metadata,
'scope': scope,
'dataset_lifetime': dataset_lifetime}
threads.append(threading.Thread(target=automatix, kwargs=kwargs))
[thread.start() for thread in threads]
while threads[0].is_alive():
logging.log(logging.DEBUG, 'Still %i active threads', len(threads))
[thread.join(timeout=3.14) for thread in threads]
def list_scopes():
"""
Lists all scopes.
:returns: A list containing all scopes.
"""
return core_scope.list_scopes()
def list_scopes():
"""
Lists all scopes.
:returns: A list containing all scopes.
"""
return [scope.external for scope in core_scope.list_scopes()]
def list_scopes():
"""
Lists all scopes.
:returns: A list containing all scopes.
"""
return core_scope.list_scopes()
def run(total_workers=1, once=False, inputfile=None):
"""
Starts up the automatix threads.
"""
try:
sites = [s.strip() for s in get('automatix', 'sites').split(',')]
except Exception:
raise Exception('Could not load sites from configuration')
if not inputfile:
inputfile = '/opt/rucio/etc/automatix.json'
try:
sleep_time = get('automatix', 'sleep_time')
except Exception:
sleep_time = 30
try:
account = get('automatix', 'account')
except Exception:
account = 'root'
try:
dataset_lifetime = get('automatix', 'dataset_lifetime')
except Exception:
dataset_lifetime = None
try:
set_metadata = get('automatix', 'set_metadata')
except Exception:
set_metadata = False
try:
scope = get('automatix', 'scope')
if InternalScope(scope) not in list_scopes():
logging.error('Scope %s does not exist. Exiting', scope)
GRACEFUL_STOP.set()
except Exception:
scope = False
threads = list()
for worker_number in range(0, total_workers):
kwargs = {
'worker_number': worker_number,
'total_workers': total_workers,
'once': once,
'sites': sites,
'sleep_time': sleep_time,
'account': account,
'inputfile': inputfile,
'set_metadata': set_metadata,
'scope': scope,
'dataset_lifetime': dataset_lifetime
}
threads.append(threading.Thread(target=automatix, kwargs=kwargs))
[thread.start() for thread in threads]
while threads[0].is_alive():
logging.debug('Still %i active threads', len(threads))
[thread.join(timeout=3.14) for thread in threads]
def add_files(lfns, issuer, ignore_availability):
"""
Bulk add files :
- Create the file and replica.
- If doesn't exist create the dataset containing the file as well as a rule on the dataset on ANY sites.
- Create all the ascendants of the dataset if they do not exist
:param lfns: List of lfn (dictionary {'lfn': <lfn>, 'rse': <rse>, 'bytes': <bytes>, 'adler32': <adler32>, 'guid': <guid>, 'pfn': <pfn>}
:param issuer: The issuer account.
:param ignore_availability: A boolean to ignore blacklisted sites.
"""
scopes = list_scopes()
dids = []
rses = {}
for lfn in lfns:
scope, name = extract_scope(lfn['lfn'], scopes)
dids.append({'scope': scope, 'name': name})
rse = lfn['rse']
if rse not in rses:
rse_id = get_rse_id(rse=rse)
rses[rse] = rse_id
lfn['rse_id'] = rses[rse]
# Check if the issuer can add dids and use skip_availabitlity
for rse in rses:
rse_id = rses[rse]
kwargs = {'rse': rse, 'rse_id': rse_id}
if not has_permission(
issuer=issuer, action='add_replicas', kwargs=kwargs):
raise AccessDenied('Account %s can not add file replicas on %s' %
(issuer, rse))
if not has_permission(issuer=issuer,
action='skip_availability_check',
kwargs=kwargs):
ignore_availability = False
# Check if the issuer can add the files
kwargs = {'issuer': issuer, 'dids': dids}
if not has_permission(issuer=issuer, action='add_dids', kwargs=kwargs):
raise AccessDenied('Account %s can not bulk add data identifier' %
(issuer))
issuer = InternalAccount(issuer)
dirac.add_files(lfns=lfns,
account=issuer,
ignore_availability=ignore_availability,
session=None)
def list_scopes(filter={}, vo='def'):
"""
Lists all scopes.
:param filter: Dictionary of attributes by which the input data should be filtered
:param vo: The VO to act on.
:returns: A list containing all scopes.
"""
# If filter is empty, create a new dict to avoid overwriting the function's default
if not filter:
filter = {}
if 'scope' in filter:
filter['scope'] = InternalScope(scope=filter['scope'], vo=vo)
else:
filter['scope'] = InternalScope(scope='*', vo=vo)
return [scope.external for scope in core_scope.list_scopes(filter=filter)]
def add_files(lfns, account, ignore_availability, session=None):
"""
Bulk add files :
- Create the file and replica.
- If doesn't exist create the dataset containing the file as well as a rule on the dataset on ANY sites.
- Create all the ascendants of the dataset if they do not exist
:param lfns: List of lfn (dictionary {'lfn': <lfn>, 'rse': <rse>, 'bytes': <bytes>, 'adler32': <adler32>, 'guid': <guid>, 'pfn': <pfn>}
:param issuer: The issuer account.
:param ignore_availability: A boolean to ignore blocklisted sites.
:session: The session used
"""
attachments = []
# The list of scopes is necessary for the extract_scope
scopes = list_scopes(session=session)
scopes = [scope.external for scope in scopes]
exist_lfn = []
for lfn in lfns:
# First check if the file exists
filename = lfn['lfn']
lfn_scope, _ = extract_scope(filename, scopes)
lfn_scope = InternalScope(lfn_scope)
exists, did_type = _exists(lfn_scope, filename)
if exists:
continue
# Get all the ascendants of the file
lfn_split = filename.split('/')
lpns = ["/".join(lfn_split[:idx]) for idx in range(2, len(lfn_split))]
lpns.reverse()
print(lpns)
# The parent must be a dataset. Register it as well as the rule
dsn_name = lpns[0]
dsn_scope, _ = extract_scope(dsn_name, scopes)
dsn_scope = InternalScope(dsn_scope)
exists, did_type = _exists(dsn_scope, dsn_name)
if exists and did_type == DIDType.CONTAINER:
raise UnsupportedOperation('Cannot create %s as dataset' %
dsn_name)
if (dsn_name not in exist_lfn) and not exists:
print('Will create %s' % dsn_name)
add_did(scope=dsn_scope,
name=dsn_name,
type=DIDType.DATASET,
account=InternalAccount(account),
statuses=None,
meta=None,
rules=[{
'copies': 1,
'rse_expression': 'ANY=true',
'weight': None,
'account': InternalAccount(account),
'lifetime': None,
'grouping': 'NONE'
}],
lifetime=None,
dids=None,
rse_id=None,
session=session)
exist_lfn.append(dsn_name)
parent_name = lpns[1]
parent_scope, _ = extract_scope(parent_name, scopes)
parent_scope = InternalScope(parent_scope)
attachments.append({
'scope': parent_scope,
'name': parent_name,
'dids': [{
'scope': dsn_scope,
'name': dsn_name
}]
})
# Register the file
rse_id = lfn.get('rse_id', None)
if not rse_id:
raise InvalidType('Missing rse_id')
bytes = lfn.get('bytes', None)
guid = lfn.get('guid', None)
adler32 = lfn.get('adler32', None)
pfn = lfn.get('pfn', None)
files = {
'scope': lfn_scope,
'name': filename,
'bytes': bytes,
'adler32': adler32
}
if pfn:
files['pfn'] = str(pfn)
if guid:
files['meta'] = {'guid': guid}
add_replicas(rse_id=rse_id,
files=[files],
dataset_meta=None,
account=InternalAccount(account),
ignore_availability=ignore_availability,
session=session)
add_rule(dids=[{
'scope': lfn_scope,
'name': filename
}],
account=InternalAccount(account),
copies=1,
rse_expression=lfn['rse'],
grouping=None,
weight=None,
lifetime=86400,
locked=None,
subscription_id=None,
session=session)
attachments.append({
'scope': dsn_scope,
'name': dsn_name,
'dids': [{
'scope': lfn_scope,
'name': filename
}]
})
# Now loop over the ascendants of the dataset and created them
for lpn in lpns[1:]:
child_scope, _ = extract_scope(lpn, scopes)
child_scope = InternalScope(child_scope)
exists, did_type = _exists(child_scope, lpn)
if exists and did_type == DIDType.DATASET:
raise UnsupportedOperation('Cannot create %s as container' %
lpn)
if (lpn not in exist_lfn) and not exists:
print('Will create %s' % lpn)
add_did(scope=child_scope,
name=lpn,
type=DIDType.CONTAINER,
account=InternalAccount(account),
statuses=None,
meta=None,
rules=None,
lifetime=None,
dids=None,
rse_id=None,
session=session)
exist_lfn.append(lpn)
parent_name = lpns[lpns.index(lpn) + 1]
parent_scope, _ = extract_scope(parent_name, scopes)
parent_scope = InternalScope(parent_scope)
attachments.append({
'scope':
parent_scope,
'name':
parent_name,
'dids': [{
'scope': child_scope,
'name': lpn
}]
})
# Finally attach everything
attach_dids_to_dids(attachments,
account=InternalAccount(account),
ignore_duplicate=True,
session=session)
def add_files(lfns, account, ignore_availability, vo='def', session=None):
"""
Bulk add files :
- Create the file and replica.
- If doesn't exist create the dataset containing the file as well as a rule on the dataset on ANY sites.
- Create all the ascendants of the dataset if they do not exist
:param lfns: List of lfn (dictionary {'lfn': <lfn>, 'rse': <rse>, 'bytes': <bytes>, 'adler32': <adler32>, 'guid': <guid>, 'pfn': <pfn>}
:param issuer: The issuer account.
:param ignore_availability: A boolean to ignore blocklisted sites.
:param vo: The VO to act on
:param session: The session used
"""
rule_extension_list = []
attachments = []
# The list of scopes is necessary for the extract_scope
filter_ = {'scope': InternalScope(scope='*', vo=vo)}
scopes = list_scopes(filter_=filter_, session=session)
scopes = [scope.external for scope in scopes]
exist_lfn = []
try:
lifetime_dict = config_get(section='dirac', option='lifetime', session=session)
lifetime_dict = loads(lifetime_dict)
except ConfigNotFound:
lifetime_dict = {}
except JSONDecodeError as err:
raise InvalidType('Problem parsing lifetime option in dirac section : %s' % str(err))
except Exception as err:
raise RucioException(str(err))
for lfn in lfns:
# First check if the file exists
filename = lfn['lfn']
lfn_scope, _ = extract_scope(filename, scopes)
lfn_scope = InternalScope(lfn_scope, vo=vo)
exists, did_type = _exists(lfn_scope, filename)
if exists:
continue
# Get all the ascendants of the file
lfn_split = filename.split('/')
lpns = ["/".join(lfn_split[:idx]) for idx in range(2, len(lfn_split))]
lpns.reverse()
print(lpns)
# The parent must be a dataset. Register it as well as the rule
dsn_name = lpns[0]
dsn_scope, _ = extract_scope(dsn_name, scopes)
dsn_scope = InternalScope(dsn_scope, vo=vo)
# Compute lifetime
lifetime = None
if dsn_scope in lifetime_dict:
lifetime = lifetime_dict[dsn_scope]
else:
for pattern in lifetime_dict:
if re.match(pattern, dsn_scope):
lifetime = lifetime_dict[pattern]
break
exists, did_type = _exists(dsn_scope, dsn_name)
if exists and did_type == DIDType.CONTAINER:
raise UnsupportedOperation('Cannot create %s as dataset' % dsn_name)
if (dsn_name not in exist_lfn) and not exists:
print('Will create %s' % dsn_name)
# to maintain a compatibility between master and LTS-1.26 branches remove keywords for first 3 arguments
add_did(dsn_scope,
dsn_name,
DIDType.DATASET,
account=InternalAccount(account, vo=vo),
statuses=None,
meta=None,
rules=[{'copies': 1, 'rse_expression': 'ANY=true', 'weight': None, 'account': InternalAccount(account, vo=vo), 'lifetime': None, 'grouping': 'NONE'}],
lifetime=None,
dids=None,
rse_id=None,
session=session)
exist_lfn.append(dsn_name)
parent_name = lpns[1]
parent_scope, _ = extract_scope(parent_name, scopes)
parent_scope = InternalScope(parent_scope, vo=vo)
attachments.append({'scope': parent_scope, 'name': parent_name, 'dids': [{'scope': dsn_scope, 'name': dsn_name}]})
rule_extension_list.append((dsn_scope, dsn_name))
if lifetime and (dsn_scope, dsn_name) not in rule_extension_list:
# Reset the lifetime of the rule to the configured value
rule = [rul for rul in list_rules({'scope': dsn_scope, 'name': dsn_name, 'account': InternalAccount(account, vo=vo)}, session=session) if rul['rse_expression'] == 'ANY=true']
if rule:
update_rule(rule[0]['id'], options={'lifetime': lifetime}, session=session)
rule_extension_list.append((dsn_scope, dsn_name))
# Register the file
rse_id = lfn.get('rse_id', None)
if not rse_id:
raise InvalidType('Missing rse_id')
bytes_ = lfn.get('bytes', None)
guid = lfn.get('guid', None)
adler32 = lfn.get('adler32', None)
pfn = lfn.get('pfn', None)
files = {'scope': lfn_scope, 'name': filename, 'bytes': bytes_, 'adler32': adler32}
if pfn:
files['pfn'] = str(pfn)
if guid:
files['meta'] = {'guid': guid}
add_replicas(rse_id=rse_id,
files=[files],
dataset_meta=None,
account=InternalAccount(account, vo=vo),
ignore_availability=ignore_availability,
session=session)
add_rule(dids=[{'scope': lfn_scope, 'name': filename}],
account=InternalAccount(account, vo=vo),
copies=1,
rse_expression=lfn['rse'],
grouping=None,
weight=None,
lifetime=86400,
locked=None,
subscription_id=None,
session=session)
attachments.append({'scope': dsn_scope, 'name': dsn_name, 'dids': [{'scope': lfn_scope, 'name': filename}]})
# Now loop over the ascendants of the dataset and created them
for lpn in lpns[1:]:
child_scope, _ = extract_scope(lpn, scopes)
child_scope = InternalScope(child_scope, vo=vo)
exists, did_type = _exists(child_scope, lpn)
if exists and did_type == DIDType.DATASET:
raise UnsupportedOperation('Cannot create %s as container' % lpn)
if (lpn not in exist_lfn) and not exists:
print('Will create %s' % lpn)
add_did(child_scope,
lpn,
DIDType.CONTAINER,
account=InternalAccount(account, vo=vo),
statuses=None,
meta=None,
rules=None,
lifetime=None,
dids=None,
rse_id=None,
session=session)
exist_lfn.append(lpn)
parent_name = lpns[lpns.index(lpn) + 1]
parent_scope, _ = extract_scope(parent_name, scopes)
parent_scope = InternalScope(parent_scope, vo=vo)
attachments.append({'scope': parent_scope, 'name': parent_name, 'dids': [{'scope': child_scope, 'name': lpn}]})
# Finally attach everything
attach_dids_to_dids(attachments,
account=InternalAccount(account, vo=vo),
ignore_duplicate=True,
session=session)
