def block_polling(self,
interval_sec=30,
timeout_sec=3600) -> SparkJobStatus:
"""
Blocks until the Dataproc job is completed or failed.
Args:
interval_sec (int): Polling interval.
timeout_sec (int): Timeout limit.
Returns:
SparkJobStatus: Latest job status
Raise:
SparkJobFailure: Raise error if the job neither failed nor completed within the timeout limit.
"""
start = time.time()
while True:
elapsed_time = time.time() - start
if timeout_sec and elapsed_time >= timeout_sec:
raise SparkJobFailure(
f"Job is still not completed after {timeout_sec}.")
status = self.get_status()
if status in [SparkJobStatus.FAILED, SparkJobStatus.COMPLETED]:
break
time.sleep(interval_sec)
return status
def get_output_file_uri(self, timeout_sec=None):
state = _wait_for_job_state(self._emr_client, self._job_ref,
TERMINAL_STEP_STATES, timeout_sec)
if state in SUCCEEDED_STEP_STATES:
return self._output_file_uri
else:
raise SparkJobFailure("Spark job failed")
def get_output_file_uri(self, timeout_sec=None, block=True):
if not block:
return self._output_file_uri
status = self.block_polling(timeout_sec=timeout_sec)
if status == SparkJobStatus.COMPLETED:
return self._output_file_uri
raise SparkJobFailure(self.get_error_message())
def get_output_file_uri(self, timeout_sec=None, block=True):
if not block:
return self._output_file_uri
if self._wait_for_complete(timeout_sec):
return self._output_file_uri
else:
raise SparkJobFailure("Spark job failed")
def get_output_file_uri(self, timeout_sec: int = None):
with self._process as p:
try:
p.wait(timeout_sec)
except Exception:
p.kill()
raise SparkJobFailure(
"Timeout waiting for subprocess to return")
if self._process.returncode != 0:
stderr = "" if self._process.stderr is None else self._process.stderr.read(
)
stdout = "" if self._process.stdout is None else self._process.stdout.read(
)
raise SparkJobFailure(
f"Non zero return code: {self._process.returncode}. stderr: {stderr} stdout: {stdout}"
)
def get_output_file_uri(self, timeout_sec=None, block=True):
if not block:
return self._output_file_uri
try:
self._operation.result(timeout_sec)
except Exception as err:
raise SparkJobFailure(err)
return self._output_file_uri
def get_output_file_uri(self, timeout_sec=None):
status = self._wait_for_job_status(
goal_status=[SparkJobStatus.COMPLETED, SparkJobStatus.FAILED],
timeout_seconds=600,
)
if status == SparkJobStatus.COMPLETED:
return self._output_file_uri
else:
raise SparkJobFailure("Spark job failed")
