def run_do(self):
try:
running_tasks = job_utils.query_task(status='running',
run_ip=get_lan_ip())
stop_job_ids = set()
# detect_logger.info('start to detect running job..')
for task in running_tasks:
try:
process_exist = job_utils.check_job_process(
int(task.f_run_pid))
if not process_exist:
detect_logger.info(
'job {} component {} on {} {} task {} {} process does not exist'
.format(task.f_job_id, task.f_component_name,
task.f_role, task.f_party_id,
task.f_task_id, task.f_run_pid))
stop_job_ids.add(task.f_job_id)
except Exception as e:
detect_logger.exception(e)
if stop_job_ids:
schedule_logger().info(
'start to stop jobs: {}'.format(stop_job_ids))
for job_id in stop_job_ids:
jobs = job_utils.query_job(job_id=job_id)
if jobs:
initiator_party_id = jobs[0].f_initiator_party_id
job_work_mode = jobs[0].f_work_mode
if len(jobs) > 1:
# i am initiator
my_party_id = initiator_party_id
else:
my_party_id = jobs[0].f_party_id
initiator_party_id = jobs[0].f_initiator_party_id
api_utils.federated_api(
job_id=job_id,
method='POST',
endpoint='/{}/job/stop'.format(API_VERSION),
src_party_id=my_party_id,
dest_party_id=initiator_party_id,
src_role=None,
json_body={
'job_id': job_id,
'operate': 'kill'
},
work_mode=job_work_mode)
TaskScheduler.finish_job(job_id=job_id,
job_runtime_conf=json_loads(
jobs[0].f_runtime_conf),
stop=True)
except Exception as e:
detect_logger.exception(e)
finally:
detect_logger.info('finish detect running job')
def run_do(self):
try:
running_tasks = job_utils.query_task(status='running',
run_ip=get_lan_ip())
stop_job_ids = set()
detect_logger.info('start to detect running job..')
for task in running_tasks:
try:
process_exist = job_utils.check_job_process(
int(task.f_run_pid))
if not process_exist:
detect_logger.info(
'job {} component {} on {} {} task {} {} process does not exist'
.format(task.f_job_id, task.f_component_name,
task.f_role, task.f_party_id,
task.f_task_id, task.f_run_pid))
stop_job_ids.add(task.f_job_id)
except Exception as e:
detect_logger.exception(e)
if stop_job_ids:
schedule_logger.info(
'start to stop jobs: {}'.format(stop_job_ids))
for job_id in stop_job_ids:
jobs = job_utils.query_job(job_id=job_id)
if jobs:
initiator_party_id = jobs[0].f_initiator_party_id
job_work_mode = jobs[0].f_work_mode
if len(jobs) > 1:
# i am initiator
my_party_id = initiator_party_id
else:
my_party_id = jobs[0].f_party_id
initiator_party_id = jobs[0].f_initiator_party_id
api_utils.local_api(
method='POST',
endpoint='/{}/job/stop'.format(API_VERSION),
json_body={'job_id': job_id})
schedule_logger.info(
'send stop job {} command'.format(job_id))
except Exception as e:
detect_logger.exception(e)
finally:
detect_logger.info('finish detect running job')
def detect_running_task(cls):
detect_logger().info('start to detect running task..')
count = 0
try:
running_tasks = JobSaver.query_task(
party_status=TaskStatus.RUNNING,
run_on_this_party=True,
run_ip=RuntimeConfig.JOB_SERVER_HOST,
only_latest=False)
stop_job_ids = set()
for task in running_tasks:
count += 1
try:
process_exist = job_utils.check_job_process(
int(task.f_run_pid))
if not process_exist:
detect_logger(job_id=task.f_job_id).info(
'job {} task {} {} on {} {} process {} does not exist'
.format(task.f_job_id, task.f_task_id,
task.f_task_version, task.f_role,
task.f_party_id, task.f_run_pid))
stop_job_ids.add(task.f_job_id)
except Exception as e:
detect_logger(job_id=task.f_job_id).exception(e)
if stop_job_ids:
detect_logger().info(
'start to stop jobs: {}'.format(stop_job_ids))
stop_jobs = set()
for job_id in stop_job_ids:
jobs = JobSaver.query_job(job_id=job_id)
if jobs:
stop_jobs.add(jobs[0])
cls.request_stop_jobs(jobs=stop_jobs,
stop_msg="task executor process abort",
stop_status=JobStatus.CANCELED)
except Exception as e:
detect_logger().exception(e)
finally:
detect_logger().info(f"finish detect {count} running task")
def is_alive(task):
return job_utils.check_job_process(int(task.f_run_pid))