def test_lifetime_truncate_expiration(root_account, rse_factory, mock_scope, did_factory, rucio_client):
"""
Test the duration of a lifetime exception is truncated if max_extension is defined
"""
nb_datasets = 2
today = datetime.now()
yesterday = today - timedelta(days=1)
tomorrow = today + timedelta(days=1)
next_year = today + timedelta(days=365)
rse, rse_id = rse_factory.make_posix_rse()
datasets = [did_factory.make_dataset() for _ in range(nb_datasets)]
metadata = str(uuid())
for dataset in datasets:
set_metadata(dataset['scope'], dataset['name'], 'datatype', metadata)
set_metadata(dataset['scope'], dataset['name'], 'eol_at', yesterday)
client_datasets = list()
for dataset in datasets:
client_datasets.append({'scope': dataset['scope'].external, 'name': dataset['name'], 'did_type': 'DATASET'})
tomorrow = tomorrow.strftime('%Y-%m-%d')
config_core.set(section='lifetime_model', option='cutoff_date', value=tomorrow)
config_core.get(section='lifetime_model', option='cutoff_date', default=None, use_cache=False)
config_core.set(section='lifetime_model', option='max_extension', value=30)
config_core.get(section='lifetime_model', option='max_extension', default=None, use_cache=False)
result = rucio_client.add_exception(client_datasets, account='root', pattern='wekhewfk', comments='This is a comment', expires_at=next_year)
exception_id = list(result['exceptions'].keys())[0]
exception = [exception for exception in rucio_client.list_exceptions() if exception['id'] == exception_id][0]
assert exception['expires_at'] - exception['created_at'] < timedelta(31)
assert exception['expires_at'] - exception['created_at'] > timedelta(29)
def reset_config_table():
""" Clear the config table and install any default entires needed for the tests.
"""
db_session = session.get_session()
db_session.query(models.Config).delete()
db_session.commit()
config_db.set("vo-map", "testvo1", "tst")
config_db.set("vo-map", "testvo2", "ts2")
def test_get_config_sections(self):
""" CONFIG (CORE): Retreive configuration section only """
expected_sections = [str(generate_uuid()), str(generate_uuid())]
for section in expected_sections:
config_core.set(section, str(generate_uuid()), str(generate_uuid()))
sections = config_core.sections(use_cache=False)
for section in expected_sections:
assert section in sections
def test_throttler_fifo_release_nothing(self):
""" THROTTLER (CLIENTS): throttler release nothing (fifo). """
# two waiting requests and one active requests but threshold is 1
# more than 80% of the transfer limit are already used -> release nothing
set('throttler', '%s,%s' % (self.user_activity, self.dest_rse), 1, session=self.db_session)
request = models.Request(dest_rse_id=self.dest_rse_id, bytes=2, activity=self.user_activity, state=constants.RequestState.SUBMITTED)
request.save(session=self.db_session)
name1 = generate_uuid()
name2 = generate_uuid()
add_replica(self.source_rse_id, self.scope, name1, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name2, 1, self.account, session=self.db_session)
requests = [{
'dest_rse_id': self.dest_rse_id,
'source_rse_id': self.source_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name1,
'account': self.account,
'scope': self.scope,
'rule_id': generate_uuid(),
'retry_count': 1,
'requested_at': datetime.now().replace(year=2018),
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}, {
'dest_rse_id': self.dest_rse_id,
'source_rse_id': self.source_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'requested_at': datetime.now().replace(year=2020),
'name': name2,
'account': self.account,
'scope': self.scope,
'rule_id': generate_uuid(),
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}]
queue_requests(requests, session=self.db_session)
self.db_session.commit()
throttler.run(once=True, sleep_time=1)
request = get_request_by_did(self.scope, name1, self.dest_rse_id)
assert_equal(request['state'], constants.RequestState.WAITING)
request2 = get_request_by_did(self.scope, name2, self.dest_rse_id)
assert_equal(request2['state'], constants.RequestState.WAITING)
def setUpClass(cls):
cls.db_session = session.get_session()
cls.dest_rse = 'MOCK'
cls.source_rse = 'MOCK4'
cls.dest_rse_id = get_rse_id(cls.dest_rse)
cls.source_rse_id = get_rse_id(cls.source_rse)
cls.scope = InternalScope('mock')
cls.account = InternalAccount('root')
cls.user_activity = 'User Subscription'
cls.all_activities = 'all_activities'
set_rse_transfer_limits(cls.dest_rse_id, cls.user_activity, max_transfers=1, session=cls.db_session)
set('throttler_release_strategy', 'dest_%s' % cls.dest_rse_id, 'fifo', session=cls.db_session)
def setUpClass(cls):
cls.db_session = session.get_session()
cls.dialect = cls.db_session.bind.dialect.name
cls.dest_rse = 'MOCK'
cls.source_rse = 'MOCK4'
cls.dest_rse_id = get_rse_id(cls.dest_rse)
cls.source_rse_id = get_rse_id(cls.source_rse)
cls.scope = InternalScope('mock')
cls.account = InternalAccount('root')
cls.user_activity = 'User Subscription'
cls.all_activities = 'all_activities'
set('throttler_release_strategy', 'dest_%s' % cls.dest_rse_id, 'grouped_fifo')
def test_throttler_fifo_release_subset(self):
""" THROTTLER (CLIENTS): throttler release subset of waiting requests (fifo). """
# two waiting requests and no active requests but threshold is 1 -> release only 1 request
set('throttler', '%s,%s' % (self.user_activity, self.dest_rse), 1, session=self.db_session)
name1 = generate_uuid()
name2 = generate_uuid()
add_replica(self.source_rse_id, self.scope, name1, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name2, 1, self.account, session=self.db_session)
requests = [{
'dest_rse_id': self.dest_rse_id,
'source_rse_id': self.source_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name1,
'account': self.account,
'scope': self.scope,
'rule_id': generate_uuid(),
'retry_count': 1,
'requested_at': datetime.now().replace(year=2018),
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}, {
'dest_rse_id': self.dest_rse_id,
'source_rse_id': self.source_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'requested_at': datetime.now().replace(year=2020),
'name': name2,
'account': self.account,
'scope': self.scope,
'rule_id': generate_uuid(),
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}]
queue_requests(requests, session=self.db_session)
self.db_session.commit()
throttler.run(once=True, sleep_time=1)
request = get_request_by_did(self.scope, name1, self.dest_rse_id)
assert_equal(request['state'], constants.RequestState.QUEUED)
request2 = get_request_by_did(self.scope, name2, self.dest_rse_id)
assert_equal(request2['state'], constants.RequestState.WAITING)
def dest_throttler(db_session, mock_request):
config.set('throttler', 'mode', 'DEST_PER_ACT', session=db_session)
set_rse_transfer_limits(mock_request.dest_rse_id,
activity=mock_request.activity,
max_transfers=1,
strategy='fifo',
session=db_session)
db_session.commit()
yield
db_session.query(models.RSETransferLimit).filter_by(
rse_id=mock_request.dest_rse_id).delete()
config.remove_option('throttler', 'mode', session=db_session)
db_session.commit()
def set(section, option, value, issuer=None, vo='def', session=None):
"""
Set the given option to the specified value.
:param section: The name of the section.
:param option: The name of the option.
:param value: The content of the value.
:param issuer: The issuer account.
:param vo: The VO to act on.
:param session: The database session in use.
"""
kwargs = {
'issuer': issuer,
'section': section,
'option': option,
'value': value
}
if not permission.has_permission(issuer=issuer,
vo=vo,
action='config_set',
kwargs=kwargs,
session=session):
raise exception.AccessDenied(
'%s cannot set option %s to %s in section %s' %
(issuer, option, value, section))
return config.set(section, option, value, session=session)
def setUpClass(cls):
if config_get_bool('common', 'multi_vo', raise_exception=False, default=False):
cls.vo = {'vo': 'tst'}
else:
cls.vo = {}
cls.db_session = session.get_session()
cls.dest_rse = 'MOCK'
cls.source_rse = 'MOCK4'
cls.dest_rse_id = get_rse_id(cls.dest_rse, **cls.vo)
cls.source_rse_id = get_rse_id(cls.source_rse, **cls.vo)
cls.scope = InternalScope('mock', **cls.vo)
cls.account = InternalAccount('root', **cls.vo)
cls.user_activity = 'User Subscription'
cls.all_activities = 'all_activities'
set_rse_transfer_limits(cls.dest_rse_id, cls.user_activity, max_transfers=1, session=cls.db_session)
set('throttler_release_strategy', 'dest_%s' % cls.dest_rse_id, 'fifo', session=cls.db_session)
def test_get_and_set_section_option(self):
""" CONFIG (CORE): Retreive configuration option only """
# get and set
section = str(generate_uuid())
option = str(generate_uuid())
expected_value = str(generate_uuid())
config_core.set(section=section, option=option, value=expected_value)
value = config_core.get(section, option, use_cache=False)
assert_equal(value, expected_value)
# default value
section = str(generate_uuid())
config_core.set(section=section,
option=str(generate_uuid()),
value=str(generate_uuid()))
default_value = 'default'
value = config_core.get(section,
'new_option',
default=default_value,
use_cache=False)
assert_equal(value, default_value)
def set(section, option, value, issuer=None):
"""
Set the given option to the specified value.
:param section: The name of the section.
:param option: The name of the option.
:param value: The content of the value.
:param issuer: The issuer account.
"""
kwargs = {'issuer': issuer, 'section': section, 'option': option, 'value': value}
if not permission.has_permission(issuer=issuer, action='config_set', kwargs=kwargs):
raise exception.AccessDenied('%s cannot set option %s to %s in section %s' % (issuer, option, value, section))
return config.set(section, option, value)
def test_throttler_grouped_fifo_subset(self):
""" THROTTLER (CLIENTS): throttler release subset of waiting requests (grouped fifo). """
set_rse_transfer_limits(self.dest_rse_id, self.all_activities, volume=10, max_transfers=1, session=self.db_session)
set('throttler', '%s,%s' % (self.all_activities, self.dest_rse), 1, session=self.db_session) # threshold used by throttler
name1 = generate_uuid()
name2 = generate_uuid()
name3 = generate_uuid()
name4 = generate_uuid()
dataset_1_name = generate_uuid()
add_did(self.scope, dataset_1_name, constants.DIDType.DATASET, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name1, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name2, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name3, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name4, 1, self.account, session=self.db_session)
attach_dids(self.scope, dataset_1_name, [{'name': name1, 'scope': self.scope}], self.account, session=self.db_session)
attach_dids(self.scope, dataset_1_name, [{'name': name2, 'scope': self.scope}], self.account, session=self.db_session)
requests = [{
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name1,
'bytes': 1,
'scope': self.scope,
'retry_count': 1,
'rule_id': generate_uuid(),
'requested_at': datetime.now().replace(year=2000),
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}, {
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name2,
'bytes': 2,
'requested_at': datetime.now().replace(year=2020),
'rule_id': generate_uuid(),
'scope': self.scope,
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 2,
'md5': '',
'adler32': ''
}
}, {
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name3,
'bytes': 3,
'requested_at': datetime.now().replace(year=2021), # requested after the request below but small enough for max_volume check
'rule_id': generate_uuid(),
'scope': self.scope,
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 3,
'md5': '',
'adler32': ''
}
}, {
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name4,
'bytes': 3000,
'requested_at': datetime.now().replace(year=2020),
'rule_id': generate_uuid(),
'scope': self.scope,
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 3000,
'md5': '',
'adler32': ''
}
}]
queue_requests(requests, session=self.db_session)
self.db_session.commit()
throttler.run(once=True, sleep_time=1)
# released because it got requested first
request_1 = get_request_by_did(self.scope, name1, self.dest_rse_id)
assert_equal(request_1['state'], constants.RequestState.QUEUED)
# released because the DID is attached to the same dataset
request_2 = get_request_by_did(self.scope, name2, self.dest_rse_id)
assert_equal(request_2['state'], constants.RequestState.QUEUED)
# released because of available volume
request_3 = get_request_by_did(self.scope, name3, self.dest_rse_id)
assert_equal(request_3['state'], constants.RequestState.QUEUED)
# still waiting because there is no free volume
request_4 = get_request_by_did(self.scope, name4, self.dest_rse_id)
assert_equal(request_4['state'], constants.RequestState.WAITING)
def test_throttler_grouped_fifo_nothing(self):
""" THROTTLER (CLIENTS): throttler release nothing (grouped fifo). """
# four waiting requests and one active requests but threshold is 1
# more than 80% of the transfer limit are already used -> release nothing
set('throttler', '%s,%s' % (self.all_activities, self.dest_rse), 1, session=self.db_session)
request = models.Request(dest_rse_id=self.dest_rse_id, bytes=2, activity=self.user_activity, state=constants.RequestState.SUBMITTED)
request.save(session=self.db_session)
name1 = generate_uuid()
name2 = generate_uuid()
name3 = generate_uuid()
name4 = generate_uuid()
dataset_1_name = generate_uuid()
add_did(self.scope, dataset_1_name, constants.DIDType.DATASET, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name1, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name2, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name3, 1, self.account, session=self.db_session)
add_replica(self.source_rse_id, self.scope, name4, 1, self.account, session=self.db_session)
attach_dids(self.scope, dataset_1_name, [{'name': name1, 'scope': self.scope}], self.account, session=self.db_session)
attach_dids(self.scope, dataset_1_name, [{'name': name2, 'scope': self.scope}], self.account, session=self.db_session)
requests = [{
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name1,
'bytes': 1,
'scope': self.scope,
'retry_count': 1,
'rule_id': generate_uuid(),
'requested_at': datetime.now().replace(year=2000),
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}, {
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name2,
'bytes': 2,
'requested_at': datetime.now().replace(year=2020),
'rule_id': generate_uuid(),
'scope': self.scope,
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 2,
'md5': '',
'adler32': ''
}
}, {
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name3,
'bytes': 3,
'requested_at': datetime.now().replace(year=2021), # requested after the request below but small enough for max_volume check
'rule_id': generate_uuid(),
'scope': self.scope,
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 3,
'md5': '',
'adler32': ''
}
}, {
'source_rse_id': self.source_rse_id,
'dest_rse_id': self.dest_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name4,
'bytes': 3000,
'requested_at': datetime.now().replace(year=2020),
'rule_id': generate_uuid(),
'scope': self.scope,
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 3000,
'md5': '',
'adler32': ''
}
}]
queue_requests(requests, session=self.db_session)
self.db_session.commit()
throttler.run(once=True, sleep_time=1)
request_1 = get_request_by_did(self.scope, name1, self.dest_rse_id)
assert_equal(request_1['state'], constants.RequestState.WAITING)
request_2 = get_request_by_did(self.scope, name2, self.dest_rse_id)
assert_equal(request_2['state'], constants.RequestState.WAITING)
request_3 = get_request_by_did(self.scope, name3, self.dest_rse_id)
assert_equal(request_3['state'], constants.RequestState.WAITING)
request_4 = get_request_by_did(self.scope, name4, self.dest_rse_id)
assert_equal(request_4['state'], constants.RequestState.WAITING)
def test_multihop_sources_created(rse_factory, did_factory, root_account, core_config_mock, caches_mock):
"""
Ensure that multihop transfers are handled and intermediate request correctly created
"""
src_rse_name, src_rse_id = rse_factory.make_posix_rse()
_, jump_rse1_id = rse_factory.make_posix_rse()
_, jump_rse2_id = rse_factory.make_posix_rse()
_, jump_rse3_id = rse_factory.make_posix_rse()
dst_rse_name, dst_rse_id = rse_factory.make_posix_rse()
jump_rses = [jump_rse1_id, jump_rse2_id, jump_rse3_id]
all_rses = jump_rses + [src_rse_id, dst_rse_id]
for rse_id in jump_rses:
rse_core.add_rse_attribute(rse_id, 'available_for_multihop', True)
rse_tombstone_delay = 3600
rse_multihop_tombstone_delay = 12 * 3600
default_multihop_tombstone_delay = 24 * 3600
# if both attributes are set, the multihop one will take precedence
rse_core.add_rse_attribute(jump_rse1_id, 'tombstone_delay', rse_tombstone_delay)
rse_core.add_rse_attribute(jump_rse1_id, 'multihop_tombstone_delay', rse_multihop_tombstone_delay)
# if multihop delay not set, it's the default multihop takes precedence. Not normal tombstone delay.
rse_core.add_rse_attribute(jump_rse2_id, 'tombstone_delay', rse_tombstone_delay)
core_config.set(section='transfers', option='multihop_tombstone_delay', value=default_multihop_tombstone_delay)
# if multihop delay is set to 0, the replica will have no tombstone
rse_core.add_rse_attribute(jump_rse3_id, 'multihop_tombstone_delay', 0)
distance_core.add_distance(src_rse_id, jump_rse1_id, ranking=10)
distance_core.add_distance(jump_rse1_id, jump_rse2_id, ranking=10)
distance_core.add_distance(jump_rse2_id, jump_rse3_id, ranking=10)
distance_core.add_distance(jump_rse3_id, dst_rse_id, ranking=10)
did = did_factory.upload_test_file(src_rse_name)
rule_core.add_rule(dids=[did], account=root_account, copies=1, rse_expression=dst_rse_name, grouping='ALL', weight=None, lifetime=None, locked=False, subscription_id=None)
submitter(once=True, rses=[{'id': rse_id} for rse_id in all_rses], partition_wait_time=None, transfertool='mock', transfertype='single', filter_transfertool=None)
# Ensure that each intermediate request was correctly created
for rse_id in jump_rses:
assert request_core.get_request_by_did(rse_id=rse_id, **did)
@read_session
def __ensure_source_exists(rse_id, scope, name, session=None):
return session.query(Source). \
filter(Source.rse_id == rse_id). \
filter(Source.scope == scope). \
filter(Source.name == name). \
one()
# Ensure that sources where created for transfers
for rse_id in jump_rses + [src_rse_id]:
__ensure_source_exists(rse_id, **did)
# Ensure the tombstone is correctly set on intermediate replicas
expected_tombstone = datetime.utcnow() + timedelta(seconds=rse_multihop_tombstone_delay)
replica = replica_core.get_replica(jump_rse1_id, **did)
assert expected_tombstone - timedelta(minutes=5) < replica['tombstone'] < expected_tombstone + timedelta(minutes=5)
expected_tombstone = datetime.utcnow() + timedelta(seconds=default_multihop_tombstone_delay)
replica = replica_core.get_replica(jump_rse2_id, **did)
assert expected_tombstone - timedelta(minutes=5) < replica['tombstone'] < expected_tombstone + timedelta(minutes=5)
replica = replica_core.get_replica(jump_rse3_id, **did)
assert replica['tombstone'] is None
def test_lifetime_creation_core(root_account, rse_factory, mock_scope,
did_factory):
"""
Test the creation of a lifetime exception on the core side
"""
nb_datatype = 3
nb_datasets = 2 * nb_datatype
yesterday = datetime.now() - timedelta(days=1)
tomorrow = datetime.now() + timedelta(days=1)
rse, rse_id = rse_factory.make_posix_rse()
datasets = [did_factory.make_dataset() for _ in range(nb_datasets)]
metadata = [str(uuid()) for _ in range(nb_datatype)]
list_dids = []
for cnt, meta in enumerate(metadata):
dids = []
for dataset in datasets[2 * cnt:2 * (cnt + 1)]:
set_metadata(dataset['scope'], dataset['name'], 'datatype', meta)
if cnt < nb_datatype - 1:
set_metadata(dataset['scope'], dataset['name'], 'eol_at',
yesterday)
dids.append((dataset['scope'], dataset['name']))
dids.sort()
list_dids.append(dids)
datasets.extend([{
'scope': mock_scope,
'name': 'dataset_%s' % str(uuid()),
'did_type': DIDType.DATASET
} for _ in range(2)])
# Test with cutoff_date not defined
try:
config_core.remove_option('lifetime_model', 'cutoff_date')
except (ConfigNotFound, NoSectionError):
pass
with pytest.raises(UnsupportedOperation):
add_exception(datasets,
root_account,
pattern='wekhewfk',
comments='This is a comment',
expires_at=datetime.now())
# Test with cutoff_date wrongly defined
config_core.set(section='lifetime_model',
option='cutoff_date',
value='wrong_value')
config_core.get(section='lifetime_model',
option='cutoff_date',
default=None,
use_cache=False)
with pytest.raises(UnsupportedOperation):
add_exception(datasets,
root_account,
pattern='wekhewfk',
comments='This is a comment',
expires_at=datetime.now())
# Test with cutoff_date properly defined
tomorrow = tomorrow.strftime('%Y-%m-%d')
config_core.set(section='lifetime_model',
option='cutoff_date',
value=tomorrow)
config_core.get(section='lifetime_model',
option='cutoff_date',
default=None,
use_cache=False)
result = add_exception(datasets,
root_account,
pattern='wekhewfk',
comments='This is a comment',
expires_at=datetime.now())
# Check if the Not Existing DIDs are identified
result_unknown = [(entry['scope'], entry['name'])
for entry in result['unknown']]
result_unknown.sort()
unknown = [(entry['scope'], entry['name'])
for entry in datasets[nb_datasets:nb_datasets + 2]]
unknown.sort()
assert result_unknown == unknown
# Check if the DIDs not affected by the Lifetime Model are identified
result_not_affected = [(entry['scope'], entry['name'])
for entry in result['not_affected']]
result_not_affected.sort()
not_affected = list_dids[-1]
assert result_not_affected == not_affected
# Check if an exception was done for each datatype
list_exceptions = list()
for exception_id in result['exceptions']:
dids = [(entry['scope'], entry['name'])
for entry in result['exceptions'][exception_id]]
dids.sort()
list_exceptions.append(dids)
for did in list_dids[:nb_datatype - 1]:
assert did in list_exceptions
def test_bb8_full_workflow(vo, root_account, jdoe_account, rse_factory,
mock_scope, did_factory):
"""BB8: Test the rebalance rule method"""
config_core.set(section='bb8', option='allowed_accounts', value='jdoe')
tot_rses = 4
rses = [rse_factory.make_posix_rse() for _ in range(tot_rses)]
rse1, rse1_id = rses[0]
rse2, rse2_id = rses[1]
rse3, rse3_id = rses[2]
rse4, rse4_id = rses[3]
# Add Tags
# RSE 1 and 2 nmatch expression T1=true
# RSE 3 and 4 nmatch expression T2=true
T1 = tag_generator()
T2 = tag_generator()
add_rse_attribute(rse1_id, T1, True)
add_rse_attribute(rse2_id, T1, True)
add_rse_attribute(rse3_id, T2, True)
add_rse_attribute(rse4_id, T2, True)
# Add fake weights
add_rse_attribute(rse1_id, "fakeweight", 10)
add_rse_attribute(rse2_id, "fakeweight", 0)
add_rse_attribute(rse3_id, "fakeweight", 0)
add_rse_attribute(rse4_id, "fakeweight", 0)
add_rse_attribute(rse1_id, "freespace", 1)
add_rse_attribute(rse2_id, "freespace", 1)
add_rse_attribute(rse3_id, "freespace", 1)
add_rse_attribute(rse4_id, "freespace", 1)
# Add quota
set_local_account_limit(jdoe_account, rse1_id, -1)
set_local_account_limit(jdoe_account, rse2_id, -1)
set_local_account_limit(jdoe_account, rse3_id, -1)
set_local_account_limit(jdoe_account, rse4_id, -1)
set_local_account_limit(root_account, rse1_id, -1)
set_local_account_limit(root_account, rse2_id, -1)
set_local_account_limit(root_account, rse3_id, -1)
set_local_account_limit(root_account, rse4_id, -1)
# Invalid the cache because the result of parse_expression is cached
REGION.invalidate()
tot_datasets = 4
# Create a list of datasets
datasets = [did_factory.make_dataset() for _ in range(tot_datasets)]
dsn = [dataset['name'] for dataset in datasets]
rules = list()
base_unit = 100000000000
nb_files1 = 7
nb_files2 = 5
nb_files3 = 3
nb_files4 = 2
file_size = 1 * base_unit
rule_to_rebalance = None
# Add one secondary file
files = create_files(1, mock_scope, rse1_id, bytes_=1)
add_rule(dids=[{
'scope': mock_scope,
'name': files[0]['name']
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
for cnt in range(3, tot_rses):
add_replicas(rses[cnt][1], files, jdoe_account)
add_rule(dids=[{
'scope': mock_scope,
'name': files[0]['name']
}],
account=jdoe_account,
copies=1,
rse_expression=rses[cnt][0],
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
rule_cleaner(once=True)
# Create dataset 1 of 800 GB and create a rule on RSE 1 and RSE 3
files = create_files(nb_files1, mock_scope, rse1_id, bytes_=file_size)
attach_dids(mock_scope, dsn[0], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[0]
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
add_replicas(rse3_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[0]
}],
account=jdoe_account,
copies=1,
rse_expression=rse3,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
# Create dataset 2 of 500 GB and create a rule on RSE 1 and RSE 2
files = create_files(nb_files2, mock_scope, rse1_id, bytes_=file_size)
attach_dids(mock_scope, dsn[1], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[1]
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
add_replicas(rse2_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[1]
}],
account=jdoe_account,
copies=1,
rse_expression=rse2,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
# Create dataset 3 of 300 GB and create a rule on RSE 1. The copy on RSE 3 is secondary
files = create_files(nb_files3, mock_scope, rse1_id, bytes_=file_size)
attach_dids(mock_scope, dsn[2], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[2]
}],
account=jdoe_account,
copies=1,
rse_expression=rse1,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rule_to_rebalance = rule_id
rules.append(rule_id)
add_replicas(rse3_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[2]
}],
account=jdoe_account,
copies=1,
rse_expression=rse3,
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
rule_cleaner(once=True)
try:
rule = get_rule(rule_id)
except:
pytest.raises(RuleNotFound, get_rule, rule_id)
# Create dataset 4 of 200 GB and create a rule on RSE 3. The copy on RSE 2 is secondary
files = create_files(nb_files4, mock_scope, rse3_id, bytes_=file_size)
attach_dids(mock_scope, dsn[3], files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[3]
}],
account=jdoe_account,
copies=1,
rse_expression=rse3,
grouping='DATASET',
weight=None,
lifetime=None,
locked=False,
subscription_id=None)[0]
rules.append(rule_id)
add_replicas(rse2_id, files, jdoe_account)
rule_id = add_rule(dids=[{
'scope': mock_scope,
'name': dsn[3]
}],
account=jdoe_account,
copies=1,
rse_expression=rse2,
grouping='DATASET',
weight=None,
lifetime=-86400,
locked=False,
subscription_id=None)[0]
rule_cleaner(once=True)
try:
rule = get_rule(rule_id)
except:
pytest.raises(RuleNotFound, get_rule, rule_id)
for dataset in dsn:
set_status(mock_scope, dataset, open=False)
for rse in rses:
fill_rse_expired(rse[1])
set_rse_usage(rse_id=rse[1],
source='min_free_space',
used=2 * base_unit,
free=2 * base_unit,
session=None)
set_rse_usage(rse_id=rse[1],
source='storage',
used=15 * base_unit,
free=2 * base_unit,
session=None)
set_rse_usage(rse_id=rse2_id,
source='min_free_space',
used=1 * base_unit,
free=1 * base_unit,
session=None)
set_rse_usage(rse_id=rse2_id,
source='storage',
used=6 * base_unit,
free=5 * base_unit,
session=None)
run_abacus(once=True, threads=1, fill_history_table=False, sleep_time=10)
# Summary :
# RSE 1 : 1500 GB primary + 1 B secondary
tot_space = [
src for src in get_rse_usage(rse1_id) if src['source'] == 'rucio'
][0]
expired = [
src for src in get_rse_usage(rse1_id) if src['source'] == 'expired'
][0]
assert tot_space['used'] == (nb_files1 + nb_files2 +
nb_files3) * file_size + 1
assert expired['used'] == 1
# RSE 2 : 500 GB primary + 100 GB secondary
tot_space = [
src for src in get_rse_usage(rse2_id) if src['source'] == 'rucio'
][0]
expired = [
src for src in get_rse_usage(rse2_id) if src['source'] == 'expired'
][0]
assert tot_space['used'] == (nb_files2 + nb_files4) * file_size
assert expired['used'] == nb_files4 * file_size
# Total primary on T1=true : 2000 GB
# Total secondary on T1=true : 200 GB
# Ratio secondary / primary = 10 %
# Ratio on RSE 1 : 0 %
# Ratio on RSE 2 : 40 %
# Now run BB8
re_evaluator(once=True, sleep_time=30, did_limit=100)
bb8_run(once=True,
rse_expression='%s=true' % str(T1),
move_subscriptions=False,
use_dump=False,
sleep_time=300,
threads=1,
dry_run=False)
for rule_id in rules:
rule = get_rule(rule_id)
if rule_id != rule_to_rebalance:
assert (rule['child_rule_id'] is None)
else:
assert (rule['child_rule_id'] is not None)
assert (
rule['expires_at'] <= datetime.utcnow() + timedelta(seconds=1)
) # timedelta needed to prevent failure due to rounding effects
child_rule_id = rule['child_rule_id']
child_rule = get_rule(child_rule_id)
assert (child_rule['rse_expression'] == rse2)
# For teardown, delete child rule
update_rule(child_rule_id, {'lifetime': -86400})
rule_cleaner(once=True)
for dataset in dsn:
set_metadata(mock_scope, dataset, 'lifetime', -86400)
undertaker.run(once=True)
def test_throttler_fifo_release_all(self):
""" THROTTLER (CLIENTS): throttler release all waiting requests (fifo). """
# no threshold -> release all waiting requests
name1 = generate_uuid()
name2 = generate_uuid()
add_replica(self.source_rse,
self.scope,
name1,
1,
self.account,
session=self.db_session)
add_replica(self.source_rse,
self.scope,
name2,
1,
self.account,
session=self.db_session)
requests = [{
'dest_rse_id': self.dest_rse_id,
'source_rse_id': self.source_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name1,
'account': self.account,
'scope': self.scope,
'rule_id': generate_uuid(),
'retry_count': 1,
'requested_at': datetime.now().replace(year=2018),
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}, {
'dest_rse_id': self.dest_rse_id,
'source_rse_id': self.source_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'requested_at': datetime.now().replace(year=2020),
'name': name2,
'account': self.account,
'scope': self.scope,
'rule_id': generate_uuid(),
'retry_count': 1,
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}]
queue_requests(requests, session=self.db_session)
self.db_session.commit()
throttler.run(once=True, sleep_time=1)
request = get_request_by_did(self.scope, name1, self.dest_rse)
assert_equal(request['state'], constants.RequestState.QUEUED)
request2 = get_request_by_did(self.scope, name2, self.dest_rse)
assert_equal(request2['state'], constants.RequestState.QUEUED)
# active transfers + waiting requests are less than the threshold -> release all waiting requests
self.db_session.query(models.Request).delete()
self.db_session.commit()
name1 = generate_uuid()
add_replica(self.source_rse,
self.scope,
name1,
1,
self.account,
session=self.db_session)
set('throttler',
'%s,%s' % (self.user_activity, self.dest_rse),
3,
session=self.db_session)
request = models.Request(dest_rse_id=self.dest_rse_id,
activity=self.user_activity,
state=constants.RequestState.SUBMITTED)
request.save(session=self.db_session)
requests = [{
'dest_rse_id': self.dest_rse_id,
'source_rse_id': self.source_rse_id,
'request_type': constants.RequestType.TRANSFER,
'request_id': generate_uuid(),
'name': name1,
'account': self.account,
'scope': self.scope,
'rule_id': generate_uuid(),
'retry_count': 1,
'requested_at': datetime.now().replace(year=2018),
'attributes': {
'activity': self.user_activity,
'bytes': 1,
'md5': '',
'adler32': ''
}
}]
queue_requests(requests, session=self.db_session)
self.db_session.commit()
throttler.run(once=True, sleep_time=1)
request = get_request_by_did(self.scope, name1, self.dest_rse)
assert_equal(request['state'], constants.RequestState.QUEUED)