def execute(self, context) -> None:
"""Execute call"""
args = self.cls.parse(self.create_cmd_args(context))
self.cmd = self.cls.create(**args)
self.task_instance = context['task_instance']
context['task_instance'].xcom_push(key='qbol_cmd_id', value=self.cmd.id) # type: ignore[attr-defined]
self.log.info(
"Qubole command created with Id: %s and Status: %s",
self.cmd.id, # type: ignore[attr-defined]
self.cmd.status, # type: ignore[attr-defined]
)
while not Command.is_done(self.cmd.status): # type: ignore[attr-defined]
time.sleep(Qubole.poll_interval)
self.cmd = self.cls.find(self.cmd.id) # type: ignore[attr-defined]
self.log.info(
"Command Id: %s and Status: %s", self.cmd.id, self.cmd.status # type: ignore[attr-defined]
)
if 'fetch_logs' in self.kwargs and self.kwargs['fetch_logs'] is True:
self.log.info(
"Logs for Command Id: %s \n%s", self.cmd.id, self.cmd.get_log() # type: ignore[attr-defined]
)
if self.cmd.status != 'done': # type: ignore[attr-defined]
raise AirflowException(
'Command Id: {} failed with Status: {}'.format(
self.cmd.id, self.cmd.status # type: ignore[attr-defined]
)
)
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 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 execute(self, context):
args = self.cls.parse(self.args)
self.cmd = self.cls.create(**args)
context['task_instance'].xcom_push(key='qbol_cmd_id', value=self.cmd.id)
logging.info("Qubole command created with Id: {0} and Status: {1}".format(str(self.cmd.id), self.cmd.status))
while not Command.is_done(self.cmd.status):
time.sleep(Qubole.poll_interval)
self.cmd = self.cls.find(self.cmd.id)
logging.info("Command Id: {0} and Status: {1}".format(str(self.cmd.id), self.cmd.status))
if self.kwargs.has_key('fetch_logs') and self.kwargs['fetch_logs'] == True:
logging.info("Logs for Command Id: {0} \n{1}".format(str(self.cmd.id), self.cmd.get_log()))
if self.cmd.status != 'done':
raise AirflowException('Command Id: {0} failed with Status: {1}'.format(self.cmd.id, self.cmd.status))
def execute(self, context):
args = self.cls.parse(self.create_cmd_args(context))
self.cmd = self.cls.create(**args)
context['task_instance'].xcom_push(key='qbol_cmd_id', value=self.cmd.id)
logging.info("Qubole command created with Id: {0} and Status: {1}".format(str(self.cmd.id), self.cmd.status))
while not Command.is_done(self.cmd.status):
time.sleep(Qubole.poll_interval)
self.cmd = self.cls.find(self.cmd.id)
logging.info("Command Id: {0} and Status: {1}".format(str(self.cmd.id), self.cmd.status))
if 'fetch_logs' in self.kwargs and self.kwargs['fetch_logs'] is True:
logging.info("Logs for Command Id: {0} \n{1}".format(str(self.cmd.id), self.cmd.get_log()))
if self.cmd.status != 'done':
raise AirflowException('Command Id: {0} failed with Status: {1}'.format(self.cmd.id, self.cmd.status))
def execute(self, context):
args = self.cls.parse(self.create_cmd_args(context))
self.cmd = self.cls.create(**args)
context['task_instance'].xcom_push(key='qbol_cmd_id', value=self.cmd.id)
_log.info("Qubole command created with Id: %s and Status: %s",
self.cmd.id, self.cmd.status)
while not Command.is_done(self.cmd.status):
time.sleep(Qubole.poll_interval)
self.cmd = self.cls.find(self.cmd.id)
_log.info("Command Id: %s and Status: %s", self.cmd.id, self.cmd.status)
if 'fetch_logs' in self.kwargs and self.kwargs['fetch_logs'] is True:
_log.info("Logs for Command Id: %s \n%s", self.cmd.id, self.cmd.get_log())
if self.cmd.status != 'done':
raise AirflowException('Command Id: {0} failed with Status: {1}'.format(
self.cmd.id, self.cmd.status))
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
