def monitor_command(self, command, sql_stmt):
_command = command
time.sleep(10)
try:
_command = Command.find(_command.id)
except:
time.sleep(30)
_command = Command.find(_command.id)
total_sleep_time = 0
retries = 1000
command_id = _command.id
for i in range(retries):
if _command.status == 'error':
raise AirflowException(
'Statement failed: https://api.qubole.com/v2/analyze?command_id=%s\n %s'
% (command_id, sql_stmt))
elif Command.is_done(_command.status):
return
else:
total_sleep_time += 10
if total_sleep_time > self.expected_runtime * 1.5:
raise AirflowException(
"RS Total estimated runtime was exceeded, please adjust estimation in DAG if the process requires more time to complete query %s"
% sql_stmt)
time.sleep(10)
_command = Command.find(command_id)
raise AirflowException(
'RS_monitor_command call for %s failed. https://api.qubole.com/v2/analyze?command_id=%s'
% (sql_stmt, command.id))
def qubole_by_id_raw(api_token,hcid,filename):
Qubole.configure(api_token=api_token)
cmd = Command.find(hcid)
out_file = filename + '.csv'
with open(out_file, 'wb') as writer:
cmd.get_results(writer)
return out_file
def handle_failure_retry(context):
ti = context['ti']
cmd_id = ti.xcom_pull(key='qbol_cmd_id', task_ids=ti.task_id)
if cmd_id is not None:
cmd = Command.find(cmd_id)
if cmd is not None:
if cmd.status == 'running':
log.info('Cancelling the Qubole Command Id: %s', cmd_id)
cmd.cancel()
def qubole_by_id(api_token,hcid,filename):
Qubole.configure(api_token=api_token)
cmd = Command.find(hcid)
out_file = filename + '.csv'
with open(out_file, 'wb') as writer:
cmd.get_results(writer)
df = pd.read_csv(out_file, delimiter='\t')
return df
def handle_failure_retry(context):
ti = context['ti']
cmd_id = ti.xcom_pull(key='qbol_cmd_id', task_ids=ti.task_id)
if cmd_id is not None:
cmd = Command.find(cmd_id)
if cmd is not None:
if cmd.status == 'running':
log = LoggingMixin().log
log.info('Cancelling the Qubole Command Id: %s', cmd_id)
cmd.cancel()
def run_query(self, query, user):
qbol.configure(api_token=self.configuration['token'],
api_url='%s/api' % self.configuration['endpoint'])
try:
cls = PrestoCommand if (self.configuration['query_type']
== 'presto') else HiveCommand
cmd = cls.create(query=query, label=self.configuration['cluster'])
logging.info("Qubole command created with Id: %s and Status: %s",
cmd.id, cmd.status)
while not Command.is_done(cmd.status):
time.sleep(qbol.poll_interval)
cmd = Command.find(cmd.id)
logging.info("Qubole command Id: %s and Status: %s", cmd.id,
cmd.status)
rows = []
columns = []
error = None
if cmd.status == 'done':
fp = StringIO()
cmd.get_results(fp=fp,
inline=True,
delim='\t',
fetch=False,
qlog=None,
arguments=['true'])
results = fp.getvalue()
fp.close()
data = results.split('\r\n')
columns = self.fetch_columns([
(i, TYPE_STRING) for i in data.pop(0).split('\t')
])
rows = [
dict(zip((c['name'] for c in columns), row.split('\t')))
for row in data
]
json_data = json_dumps({'columns': columns, 'rows': rows})
except KeyboardInterrupt:
logging.info('Sending KILL signal to Qubole Command Id: %s',
cmd.id)
cmd.cancel()
error = "Query cancelled by user."
json_data = None
return json_data, error
def handle_failure_retry(context) -> None:
"""Handle retries in case of failures"""
ti = context['ti']
cmd_id = ti.xcom_pull(key='qbol_cmd_id', task_ids=ti.task_id)
if cmd_id is not None:
cmd = Command.find(cmd_id)
if cmd is not None:
if cmd.status == 'done':
log.info('Command ID: %s has been succeeded, hence marking this TI as Success.', cmd_id)
ti.state = State.SUCCESS
elif cmd.status == 'running':
log.info('Cancelling the Qubole Command Id: %s', cmd_id)
cmd.cancel()
def handle_failure_retry(context):
ti = context['ti']
cmd_id = ti.xcom_pull(key='qbol_cmd_id', task_ids=ti.task_id)
if cmd_id is not None:
cmd = Command.find(cmd_id)
if cmd is not None:
log = LoggingMixin().log
if cmd.status == 'done':
log.info('Command ID: %s has been succeeded, hence marking this '
'TI as Success.', cmd_id)
ti.state = State.SUCCESS
elif cmd.status == 'running':
log.info('Cancelling the Qubole Command Id: %s', cmd_id)
cmd.cancel()
def handle_failure_retry(context):
ti = context['ti']
cmd_id = ti.xcom_pull(key='qbol_cmd_id', task_ids=ti.task_id)
if cmd_id is not None:
logger = logging.getLogger('airflow').getChild("QuboleHook")
cmd = Command.find(cmd_id)
if cmd is not None:
if cmd.status == 'done':
logger.info(
'Command ID: %s has been succeeded, hence marking this '
'TI as Success.', cmd_id)
ti.state = State.SUCCESS
elif cmd.status == 'running':
logger.info('Cancelling the Qubole Command Id: %s', cmd_id)
cmd.cancel()
def qubole(api_token,sql,replacements,filename):
Qubole.configure(api_token=api_token)
with open(sql,'r') as f:
query = f.read()
label='Trading-spark'
query = find_replace_multi(query,replacements)
hc = HiveCommand.run(query=query, label=label)
cmd = Command.find(hc.id)
out_file = filename + '.csv'
with open(out_file, 'wb') as writer:
cmd.get_results(writer)
df = pd.read_csv(out_file, delimiter='\t')
return df
def run_query(self, query, user):
qbol.configure(api_token=self.configuration['token'],
api_url='%s/api' % self.configuration['endpoint'])
try:
cls = PrestoCommand if(self.configuration['query_type'] == 'presto') else HiveCommand
cmd = cls.create(query=query, label=self.configuration['cluster'])
logging.info("Qubole command created with Id: %s and Status: %s", cmd.id, cmd.status)
while not Command.is_done(cmd.status):
time.sleep(qbol.poll_interval)
cmd = Command.find(cmd.id)
logging.info("Qubole command Id: %s and Status: %s", cmd.id, cmd.status)
rows = []
columns = []
error = None
if cmd.status == 'done':
fp = StringIO()
cmd.get_results(fp=fp, inline=True, delim='\t', fetch=False,
qlog=None, arguments=['true'])
results = fp.getvalue()
fp.close()
data = results.split('\r\n')
columns = self.fetch_columns([(i, TYPE_STRING) for i in data.pop(0).split('\t')])
rows = [dict(zip((c['name'] for c in columns), row.split('\t'))) for row in data]
json_data = json_dumps({'columns': columns, 'rows': rows})
except KeyboardInterrupt:
logging.info('Sending KILL signal to Qubole Command Id: %s', cmd.id)
cmd.cancel()
error = "Query cancelled by user."
json_data = None
return json_data, error
def run_query(self, query, user):
qbol.configure(
api_token=self.configuration.get("token"),
api_url="%s/api" % self.configuration.get("endpoint"),
)
try:
query_type = self.configuration.get("query_type", "hive")
if query_type == "quantum":
cmd = SqlCommand.create(query=query)
elif query_type == "hive":
cmd = HiveCommand.create(
query=query, label=self.configuration.get("cluster"))
elif query_type == "presto":
cmd = PrestoCommand.create(
query=query, label=self.configuration.get("cluster"))
else:
raise Exception("Invalid Query Type:%s.\
It must be : hive / presto / quantum." %
self.configuration.get("query_type"))
logging.info("Qubole command created with Id: %s and Status: %s",
cmd.id, cmd.status)
while not Command.is_done(cmd.status):
time.sleep(qbol.poll_interval)
cmd = Command.find(cmd.id)
logging.info("Qubole command Id: %s and Status: %s", cmd.id,
cmd.status)
rows = []
columns = []
error = None
if cmd.status == "done":
fp = StringIO()
cmd.get_results(
fp=fp,
inline=True,
delim="\t",
fetch=False,
qlog=None,
arguments=["true"],
)
results = fp.getvalue()
fp.close()
data = results.split("\r\n")
columns = self.fetch_columns([
(i, TYPE_STRING) for i in data.pop(0).split("\t")
])
rows = [
dict(
zip((column["name"] for column in columns),
row.split("\t"))) for row in data
]
json_data = json_dumps({"columns": columns, "rows": rows})
except KeyboardInterrupt:
logging.info("Sending KILL signal to Qubole Command Id: %s",
cmd.id)
cmd.cancel()
error = "Query cancelled by user."
json_data = None
return json_data, error
