def _complete(self):
ret = self._pipe.returncode
exec_time_s = time.time() - self._start_time
job_filename = self._job.name
illegal_chars = ("/", "\\", ":")
for char in illegal_chars:
job_filename = job_filename.replace(char, "-")
status = "finished"
output_dir = os.path.join(self._output, JOBS_OUTPUT_DIR,
self._job.name)
bytes_consumed = get_directory_size_bytes(output_dir)
event = StructuredLogEvent(
source=self._job.name,
category=EVENT_CATEGORY_RESOURCE_UTIL,
name=EVENT_NAME_BYTES_CONSUMED,
message="job output directory size",
bytes_consumed=bytes_consumed,
)
log_event(event)
result = Result(self._job.name, ret, status, exec_time_s)
ResultsAggregator.append(self._results_filename, result)
logger.info("Job %s completed return_code=%s exec_time_s=%s",
self._job.name, ret, exec_time_s)
def cancel(self):
self._return_code = 1
self._is_complete = True
result = Result(self._job.name, self._return_code,
JobCompletionStatus.CANCELED, 0.0)
ResultsAggregator.append(self._output, result, batch_id=self._batch_id)
logger.info("Canceled job %s", self._job.name)
def cancel(self):
self._return_code = 1
self._is_complete = True
if self._is_manager_node:
result = Result(
self._job.name,
self._return_code,
JobCompletionStatus.CANCELED,
0.0,
hpc_job_id=self._hpc_job_id,
)
ResultsAggregator.append(self._output, result, batch_id=self._batch_id)
logger.info("Canceled job %s", self._job.name)
else:
logger.info("Canceled job %s on non-manager node", self._job.name)
def _complete(self):
self._return_code = self._pipe.returncode
exec_time_s = time.time() - self._start_time
if not self._is_manager_node:
# This will happen on a multi-node job. Don't complete it multiple times.
logger.info(
"Job %s completed on non-manager node return_code=%s exec_time_s=%s",
self._job.name,
self._return_code,
exec_time_s,
)
return
status = JobCompletionStatus.FINISHED
output_dir = self._output / JOBS_OUTPUT_DIR / self._job.name
bytes_consumed = get_directory_size_bytes(output_dir)
event = StructuredLogEvent(
source=self._job.name,
category=EVENT_CATEGORY_RESOURCE_UTIL,
name=EVENT_NAME_BYTES_CONSUMED,
message="job output directory size",
bytes_consumed=bytes_consumed,
)
log_event(event)
result = Result(
self._job.name, self._return_code, status, exec_time_s, hpc_job_id=self._hpc_job_id
)
ResultsAggregator.append(self._output, result, batch_id=self._batch_id)
logger.info(
"Job %s completed return_code=%s exec_time_s=%s hpc_job_id=%s",
self._job.name,
self._return_code,
exec_time_s,
self._hpc_job_id,
)