def test_read_timeout(self):
policy = RetryPolicy()
# if this is the second or greater attempt, rethrow
retry, consistency = policy.on_read_timeout(
query=None, consistency=ONE, required_responses=1, received_responses=2,
data_retrieved=True, retry_num=1)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# if we didn't get enough responses, rethrow
retry, consistency = policy.on_read_timeout(
query=None, consistency=ONE, required_responses=2, received_responses=1,
data_retrieved=True, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# if we got enough responses, but also got a data response, rethrow
retry, consistency = policy.on_read_timeout(
query=None, consistency=ONE, required_responses=2, received_responses=2,
data_retrieved=True, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# we got enough responses but no data response, so retry
retry, consistency = policy.on_read_timeout(
query=None, consistency=ONE, required_responses=2, received_responses=2,
data_retrieved=False, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETRY)
self.assertEqual(consistency, ONE)
def test_write_timeout(self):
policy = RetryPolicy()
# if this is the second or greater attempt, rethrow
retry, consistency = policy.on_write_timeout(
query=None,
consistency="ONE",
write_type=WriteType.SIMPLE,
required_responses=1,
received_responses=2,
retry_num=1)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# if it's not a BATCH_LOG write, don't retry it
retry, consistency = policy.on_write_timeout(
query=None,
consistency="ONE",
write_type=WriteType.SIMPLE,
required_responses=1,
received_responses=2,
retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# retry BATCH_LOG writes regardless of received responses
retry, consistency = policy.on_write_timeout(
query=None,
consistency="ONE",
write_type=WriteType.BATCH_LOG,
required_responses=10000,
received_responses=1,
retry_num=0)
self.assertEqual(retry, RetryPolicy.RETRY)
self.assertEqual(consistency, 'ONE')
def test_unavailable(self):
"""
Use the same tests for test_write_timeout, but ensure they only RETHROW
"""
policy = RetryPolicy()
retry, consistency = policy.on_unavailable(query=None,
consistency="ONE",
required_replicas=1,
alive_replicas=2,
retry_num=1)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
retry, consistency = policy.on_unavailable(query=None,
consistency="ONE",
required_replicas=1,
alive_replicas=2,
retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
retry, consistency = policy.on_unavailable(query=None,
consistency="ONE",
required_replicas=10000,
alive_replicas=1,
retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
def _local_connect(self):
"""
assumes single node Cassandra cluster
"""
connection.setup(hosts=self.hosts,
default_keyspace=self.keyspace,
consistency=CL.ONE,
port=self.port,
cql_version=self.cql_version,
lazy_connect=False,
retry_connect=True,
compression=True,
auth_provider=None,
load_balancing_policy=RoundRobinPolicy(),
protocol_version=4,
executor_threads=2,
reconnection_policy=ConstantReconnectionPolicy(3.0, 5),
default_retry_policy=RetryPolicy(),
conviction_policy_factory=None,
metrics_enabled=False,
connection_class=None,
ssl_options=None,
sockopts=None,
max_schema_agreement_wait=10,
control_connection_timeout=2.0,
idle_heartbeat_interval=30,
schema_event_refresh_window=2,
topology_event_refresh_window=10,
connect_timeout=self.connect_timeout)
return
def test_retry_policy(self):
self.assertEqual(
insights_registry.serialize(RetryPolicy()), {
'type': 'RetryPolicy',
'namespace': 'cassandra.policies',
'options': {}
})
def test_exec_profile_clone(self):
cluster = Cluster(execution_profiles={EXEC_PROFILE_DEFAULT: ExecutionProfile(), 'one': ExecutionProfile()})
session = Session(cluster, hosts=[Host("127.0.0.1", SimpleConvictionPolicy)])
profile_attrs = {'request_timeout': 1,
'consistency_level': ConsistencyLevel.ANY,
'serial_consistency_level': ConsistencyLevel.SERIAL,
'row_factory': tuple_factory,
'retry_policy': RetryPolicy(),
'load_balancing_policy': default_lbp_factory()}
reference_attributes = ('retry_policy', 'load_balancing_policy')
# default and one named
for profile in (EXEC_PROFILE_DEFAULT, 'one'):
active = session.get_execution_profile(profile)
clone = session.execution_profile_clone_update(profile)
self.assertIsNot(clone, active)
all_updated = session.execution_profile_clone_update(clone, **profile_attrs)
self.assertIsNot(all_updated, clone)
for attr, value in profile_attrs.items():
self.assertEqual(getattr(clone, attr), getattr(active, attr))
if attr in reference_attributes:
self.assertIs(getattr(clone, attr), getattr(active, attr))
self.assertNotEqual(getattr(all_updated, attr), getattr(active, attr))
# cannot clone nonexistent profile
self.assertRaises(ValueError, session.execution_profile_clone_update, 'DOES NOT EXIST', **profile_attrs)
def test_retry_policy_on_request_error(self):
"""
Test to verify that on_request_error is called properly.
@since 3.18
@jira_ticket PYTHON-1064
@expected_result the appropriate retry policy is called
@test_category connection
"""
overloaded_error = {"result": "overloaded", "message": "overloaded"}
bootstrapping_error = {
"result": "is_bootstrapping",
"message": "isbootstrapping"
}
truncate_error = {
"result": "truncate_error",
"message": "truncate_error"
}
server_error = {"result": "server_error", "message": "server_error"}
# Test the on_request_error call
retry_policy = CounterRetryPolicy()
self.set_cluster(retry_policy)
for e in [
overloaded_error, bootstrapping_error, truncate_error,
server_error
]:
query_to_prime = "SELECT * from simulacron_keyspace.simulacron_table;"
prime_query(query_to_prime, then=e, rows=None, column_types=None)
rf = self.session.execute_async(query_to_prime)
try:
rf.result()
except:
pass
self.assertEqual(len(rf.attempted_hosts), 1) # no retry
self.assertEqual(next(retry_policy.request_error), 4)
# Test that by default, retry on next host
retry_policy = RetryPolicy()
self.set_cluster(retry_policy)
for e in [
overloaded_error, bootstrapping_error, truncate_error,
server_error
]:
query_to_prime = "SELECT * from simulacron_keyspace.simulacron_table;"
prime_query(query_to_prime, then=e, rows=None, column_types=None)
rf = self.session.execute_async(query_to_prime)
try:
rf.result()
except:
pass
self.assertEqual(len(rf.attempted_hosts), 3) # all 3 nodes failed
def connect(self, nodes):
cluster = Cluster(contact_points=nodes,
default_retry_policy=RetryPolicy())
metadata = cluster.metadata
self.session = cluster.connect('DATA')
print '\nConnected to cluster: %s' % metadata.cluster_name
for host in metadata.all_hosts():
print 'Datacenter: %s; Host: %s; Rack: %s' % (
host.datacenter, host.address, host.rack)
def test_default_connections(self):
"""
Ensure errors are not thrown when using non-default policies
"""
Cluster(load_balancing_policy=RoundRobinPolicy(),
reconnection_policy=ExponentialReconnectionPolicy(1.0, 600.0),
default_retry_policy=RetryPolicy(),
conviction_policy_factory=SimpleConvictionPolicy)
def setup_tables(self):
key_space = self.env.config.get(ConfigKeys.KEY_SPACE,
domain=ConfigKeys.STORAGE)
hosts = self.env.config.get(ConfigKeys.HOST, domain=ConfigKeys.STORAGE)
hosts = hosts.split(",")
# required to specify execution profiles in future versions
profiles = {
# override the default so we can set consistency level later
EXEC_PROFILE_DEFAULT:
ExecutionProfile(
load_balancing_policy=TokenAwarePolicy(
DCAwareRoundRobinPolicy()),
retry_policy=RetryPolicy(),
request_timeout=10.0,
row_factory=Session._row_factory, # noqa
# should probably be changed to QUORUM when having more than 3 nodes in the cluster
consistency_level=ConsistencyLevel.LOCAL_ONE,
),
# TODO: there doesn't seem to be a way to specify execution profile when
# using the library's object mapping approach, only when writing pure
# cql queries:
# https://docs.datastax.com/en/developer/python-driver/3.24/execution_profiles/
# batch profile has longer timeout since they are run async anyway
"batch":
ExecutionProfile(
load_balancing_policy=TokenAwarePolicy(
DCAwareRoundRobinPolicy()),
request_timeout=120.0,
consistency_level=ConsistencyLevel.LOCAL_ONE,
)
}
kwargs = {
"default_keyspace": key_space,
"protocol_version": 3,
"retry_connect": True,
"execution_profiles": profiles,
}
username = self._get_from_conf(ConfigKeys.USER, ConfigKeys.STORAGE)
password = self._get_from_conf(ConfigKeys.PASSWORD, ConfigKeys.STORAGE)
if password is not None:
auth_provider = PlainTextAuthProvider(
username=username,
password=password,
)
kwargs["auth_provider"] = auth_provider
connection.setup(hosts, **kwargs)
sync_table(MessageModel)
sync_table(AttachmentModel)
def cassandra_cluster():
from cassandra.cluster import Cluster
from cassandra.policies import DCAwareRoundRobinPolicy, TokenAwarePolicy, RetryPolicy
cassandra_hosts = ['10.0.0.251', '10.0.0.250', '10.0.0.249']
try:
return Cluster(contact_points=cassandra_hosts,
load_balancing_policy=TokenAwarePolicy(
DCAwareRoundRobinPolicy(local_dc='Cassandra')),
default_retry_policy=RetryPolicy())
except Error as er:
print er
def test_write_timeout(self):
policy = RetryPolicy()
# if this is the second or greater attempt, rethrow
retry, consistency = policy.on_write_timeout(
query=None, consistency="ONE", write_type=WriteType.SIMPLE,
required_responses=1, received_responses=2, retry_num=1)
self.assertEqual(retry, RetryPolicy.RETHROW)
# if it's not a BATCH_LOG write, don't retry it
retry, consistency = policy.on_write_timeout(
query=None, consistency="ONE", write_type=WriteType.SIMPLE,
required_responses=1, received_responses=2, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
# retry BATCH_LOG writes regardless of received responses
retry, consistency = policy.on_write_timeout(
query=None, consistency="ONE", write_type=WriteType.BATCH_LOG,
required_responses=10000, received_responses=1, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETRY)
self.assertEqual(consistency, "ONE")
def cluster(cfg, keyspace=None):
auth = PlainTextAuthProvider(username=cfg['cassandra_user'],
password=cfg['cassandra_pass'])
cluster = Cluster(cfg['cassandra_host'],
load_balancing_policy=RoundRobinPolicy(),
default_retry_policy=RetryPolicy(),
reconnection_policy=ExponentialReconnectionPolicy(
1.0, 600.0),
port=cfg['cassandra_port'],
auth_provider=auth)
return cluster
def test_read_timeout(self):
policy = RetryPolicy()
# if this is the second or greater attempt, rethrow
retry, consistency = policy.on_read_timeout(
query=None, consistency="ONE", required_responses=1, received_responses=2,
data_retrieved=True, retry_num=1)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# if we didn't get enough responses, rethrow
retry, consistency = policy.on_read_timeout(
query=None, consistency="ONE", required_responses=2, received_responses=1,
data_retrieved=True, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# if we got enough responses, but also got a data response, rethrow
retry, consistency = policy.on_read_timeout(
query=None, consistency="ONE", required_responses=2, received_responses=2,
data_retrieved=True, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
# we got enough responses but no data response, so retry
retry, consistency = policy.on_read_timeout(
query=None, consistency="ONE", required_responses=2, received_responses=2,
data_retrieved=False, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETRY)
self.assertEqual(consistency, 'ONE')
def connect(self, *args, **kwargs):
auth = PlainTextAuthProvider(username=self.user,
password=self.password)
cluster = Cluster(contact_points=self.host,
auth_provider=auth,
protocol_version=3,
load_balancing_policy=default_lbp_factory(),
default_retry_policy=RetryPolicy(),
reconnection_policy=ConstantReconnectionPolicy(
delay=1, max_attempts=10),
*args,
**kwargs)
return cluster
def _connect_to_cassandra(keyspace):
"""
Connect to the Cassandra cluster and return the session.
"""
if 'BACKEND_STORAGE_IP' in os.environ:
host = os.environ['BACKEND_STORAGE_IP']
else:
host = '127.0.0.1'
retry=RetryPolicy()
retry.RETRY=10
cluster = Cluster(
['127.0.0.1',host,],
reconnection_policy=ConstantReconnectionPolicy(5.0, 100),
load_balancing_policy=DCAwareRoundRobinPolicy(local_dc="ASH2"),
default_retry_policy=retry
)
session = cluster.connect(keyspace)
session.default_timeout = 99999
session.default_fetch_size = 1000
return session
def test(self):
try:
status = 0
cluster = Cluster(contact_points = self.hosts, default_retry_policy = RetryPolicy())
if self.db == "job":
session = cluster.connect("mgrast_abundance")
elif self.db == "m5nr":
session = cluster.connect("m5nr_v1")
if session:
status = 1
cluster.shutdown()
return status
except:
return 0
def test_unavailable(self):
"""
Use the same tests for test_write_timeout, but ensure they only RETHROW
"""
policy = RetryPolicy()
retry, consistency = policy.on_unavailable(
query=None, consistency=ONE,
required_replicas=1, alive_replicas=2, retry_num=1)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
retry, consistency = policy.on_unavailable(
query=None, consistency=ONE,
required_replicas=1, alive_replicas=2, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
retry, consistency = policy.on_unavailable(
query=None, consistency=ONE,
required_replicas=10000, alive_replicas=1, retry_num=0)
self.assertEqual(retry, RetryPolicy.RETHROW)
self.assertEqual(consistency, None)
def test_init(self):
# just make sure Statement attributes are accepted
kwargs = {'query_string': object(),
'retry_policy': RetryPolicy(),
'consistency_level': object(),
'fetch_size': object(),
'keyspace': object(),
'custom_payload': object()}
statement = SimpleGraphStatement(**kwargs)
for k, v in kwargs.items():
self.assertIs(getattr(statement, k), v)
# but not a bogus parameter
kwargs['bogus'] = object()
self.assertRaises(TypeError, SimpleGraphStatement, **kwargs)
def test_no_legacy_with_profile(self):
cluster_init = Cluster(execution_profiles={'name': ExecutionProfile()})
cluster_add = Cluster()
cluster_add.add_execution_profile('name', ExecutionProfile())
# for clusters with profiles added either way...
for cluster in (cluster_init, cluster_init):
# don't allow legacy parameters set
for attr, value in (('default_retry_policy', RetryPolicy()),
('load_balancing_policy', default_lbp_factory())):
self.assertRaises(ValueError, setattr, cluster, attr, value)
session = Session(cluster, hosts=[Host("127.0.0.1", SimpleConvictionPolicy)])
for attr, value in (('default_timeout', 1),
('default_consistency_level', ConsistencyLevel.ANY),
('default_serial_consistency_level', ConsistencyLevel.SERIAL),
('row_factory', tuple_factory)):
self.assertRaises(ValueError, setattr, session, attr, value)
def _get_session(self) -> Session:
if self._username and self._password:
auth_provider = PlainTextAuthProvider(username=self._username,
password=self._password)
else:
auth_provider = None
cluster = Cluster(
contact_points=self._contact_points,
port=self._port,
# load_balancing_policy=
reconnection_policy=ConstantReconnectionPolicy(delay=5.0),
default_retry_policy=RetryPolicy(),
auth_provider=auth_provider)
session = cluster.connect()
session.set_keyspace('nexustiles')
return session
def test_setting_retry_policy_to_statement(self):
"""
Test to verify that the retry policy is called as expected
for bound and prepared statements when set to the prepared statement
@since 3.13
@jira_ticket PYTHON-861
@expected_result the appropriate retry policy is called
@test_category connection
"""
retry_policy = RetryPolicy()
self.set_cluster(retry_policy)
then = {
"result": "write_timeout",
"delay_in_ms": 0,
"consistency_level": "LOCAL_QUORUM",
"received": 1,
"block_for": 2,
"write_type": "SIMPLE",
"ignore_on_prepare": True
}
query_to_prime_prepared = "SELECT * from simulacron_keyspace.simulacron_table WHERE key = :key"
when = {"params": {"key": "0"}, "param_types": {"key": "ascii"}}
prime_query(query_to_prime_prepared,
when=when,
then=then,
rows=None,
column_types=None)
counter_policy = CounterRetryPolicy()
prepared_stmt = self.session.prepare(query_to_prime_prepared)
prepared_stmt.retry_policy = counter_policy
self.session.execute(prepared_stmt, ("0", ))
self.assertEqual(next(counter_policy.write_timeout), 1)
counter_policy.reset_counters()
bound_stmt = prepared_stmt.bind({"key": "0"})
bound_stmt.retry_policy = counter_policy
self.session.execute(bound_stmt)
self.assertEqual(next(counter_policy.write_timeout), 1)
def execute(self):
# Connect to cassandra server
cluster = Cluster(
load_balancing_policy=TokenAwarePolicy(DCAwareRoundRobinPolicy()),
default_retry_policy=RetryPolicy())
session = cluster.connect('cs4224')
session.default_consistency_level = DEFAULT_CONSISTENCY_LEVEL
session.execute.im_self.default_timeout = 1000000000000000
# Reading transactions line by line, parsing and execute
while True:
# Break if interrupted by user
try:
line = sys.stdin.readline().strip()
except KeyboardInterrupt:
break
# Break on empty line
if not line:
break
# Count transaction
self.stats_collector.transactions.count()
# Parsing transaction
transaction_type = self.parser.get_transaction_type(line)
extra_line_cnt = self.parser.get_transaction_extra_line_count(transaction_type, line)
extra_lines = []
for i in range(extra_line_cnt):
extra_line = sys.stdin.readline().strip()
extra_lines.append(extra_line)
transaction_params = self.parser.parse(line, transaction_type, extra_lines)
# Execute transaction and measure time
self.stats_collector.transaction_timer.start()
self.execute_transaction(session, transaction_type, transaction_params)
self.stats_collector.transaction_timer.finish()
self.output()
def setUp(self):
# create bogus json
self.filename = 'testConfig.json'
# get current dir
self.cur_dir = dirname(abspath(__file__))
# load config
self.config = self.loadConfigFile(self.filename)
# create cassandra structure
cassandra = self.config['cassandra']
clus = Cluster(cassandra['url'], default_retry_policy=RetryPolicy())
sess = clus.connect()
sess.row_factory = dict_factory
sess.default_timeout = None
# save for later
cas = {
'cluster': clus,
'session': sess,
'keyspace': cassandra['keyspace']
}
self.cassandra = cas
self.data_amt = 10000
self.idList = self._generateIDs(self.data_amt)
# connect to elasticsearch and create the data
self.elastic = {'session': Elasticsearch()}
# reset data
self.resetData()
sql = ''' CREATE KEYSPACE %s
WITH REPLICATION = { 'class' : 'SimpleStrategy', 'replication_factor' : 1 }''' \
% cassandra['keyspace']
sess.execute(sql)
def create(hosts):
global CASS_CLUSTER
if not CASS_CLUSTER:
CASS_CLUSTER = Cluster(contact_points = hosts, default_retry_policy = RetryPolicy())
return CASS_CLUSTER
def main(args):
parser = OptionParser()
parser.add_option("",
"--batch",
dest="batch",
type="int",
default=100,
help="query batch size, default: 100")
parser.add_option("",
"--count",
dest="count",
type="int",
default=10,
help="number of query iterations, default: 10")
parser.add_option("",
"--phost",
dest="phost",
default=None,
help="Postgres host")
parser.add_option("",
"--pname",
dest="pname",
default=None,
help="Postgres name")
parser.add_option("",
"--puser",
dest="puser",
default=None,
help="Postgres user")
parser.add_option("",
"--chost",
dest="chost",
default=None,
help="Cassandra host")
parser.add_option("",
"--cname",
dest="cname",
default=None,
help="Cassandra name")
parser.add_option("",
"--shost",
dest="shost",
default=None,
help="SOLR host")
parser.add_option("",
"--sname",
dest="sname",
default=None,
help="SOLR collection")
(opts, args) = parser.parse_args()
handle = None
mode = ""
# test PSQL
if opts.phost and opts.pname and opts.puser:
mode = "sql"
handle = psycopg2.connect(host=opts.phost,
database=opts.pname,
user=opts.puser)
# test CASS
elif opts.chost and opts.cname:
mode = "cql"
handle = Cluster(contact_points=[opts.chost],
default_retry_policy=RetryPolicy())
# test solr
elif opts.shost and opts.sname:
mode = "solr"
surl = "http://" + opts.shost + "/solr/" + opts.sname + "/select"
shead = {'Content-Type': 'application/x-www-form-urlencoded'}
sfields = "%2C".join(
['md5_id', 'source', 'md5', 'accession', 'function', 'organism'])
else:
parser.error("Invalid usage")
start = time.time()
found = set()
for i in range(opts.count):
ints = random_array(opts.batch)
query = "SELECT * FROM md5_id_annotation WHERE id IN (" + ",".join(
map(str, ints)) + ");"
if mode == "sql":
cursor = handle.cursor()
cursor.execute(query)
data = cursor.fetchone()
while (data):
found.add(data[2])
data = cursor.fetchone()
elif mode == "cql":
session = handle.connect(opts.cname)
rows = session.execute(query)
for r in rows:
found.add(r.md5)
elif mode == "solr":
query = "md5_id:(" + " OR ".join(map(str, ints)) + ")"
sdata = "q=*%3A*&fq=" + query + "&start=0&rows=1000000000&wt=json&fl=" + sfields
req = requests.post(surl,
headers=shead,
data=sdata,
allow_redirects=True)
rj = req.json()
for d in rj['response']['docs']:
found.add(d['md5'])
end = time.time()
print "%d loops of size %d ran in %d seconds" % (opts.count, opts.batch,
int(round(end - start)))
print "%d ids requested, %d md5s found" % (
(opts.count * opts.batch), len(found))
def get_retry():
return RetryPolicy()
def __init__(self, export_process):
RetryPolicy.__init__(self)
self.max_attempts = export_process.csv_options['maxattempts']
self.printmsg = lambda txt: export_process.printmsg(txt)
# theState=gpsStationState[2] # yellow
theState = 2
# No data is available on selected date for this station, and station is
# not within either the 1 day or 30 day window.
elif (dataOnDate == False):
# theState=gpsStationState[3] # light blue
theState = 3
return theState
#########SAVE TO CASSANDRA DB###########
cluster = Cluster(contact_points=['127.0.0.1'], \
port=9042, \
load_balancing_policy= TokenAwarePolicy(DCAwareRoundRobinPolicy(local_dc='datacenter1')), \
default_retry_policy = RetryPolicy()
)
session = cluster.connect('system')
rows = cluster.metadata.keyspaces.keys()
KEYSPACE = 'GPS_' + dataSet
KEYSPACE = KEYSPACE.lower()
# Create database
keyspace_drop_stmt = session.prepare("DROP KEYSPACE " + KEYSPACE)
if KEYSPACE in rows:
print("dropping existing keyspace...")
session.execute(keyspace_drop_stmt)
print("creating keyspace...")
def __init__(self, export_process):
RetryPolicy.__init__(self)
self.max_attempts = export_process.csv_options['maxattempts']
self.printmsg = lambda txt: export_process.printmsg(txt)
