def check_job_execution_cancel(job_id, **kwargs):
ctx = context.current()
je = conductor.job_execution_get(ctx, job_id)
if je.tenant_id != ctx.tenant_id:
raise ex.CancelingFailed(
_("Job execution with id '%s' cannot be canceled "
"because it wasn't created in this tenant") % job_id)
if je.is_protected:
raise ex.CancelingFailed(
_("Job Execution with id '%s' cannot be canceled "
"because it's marked as protected") % job_id)
def cancel_job(job_execution_id):
ctx = context.ctx()
job_execution = conductor.job_execution_get(ctx, job_execution_id)
if job_execution.info['status'] in edp.JOB_STATUSES_TERMINATED:
return job_execution
cluster = conductor.cluster_get(ctx, job_execution.cluster_id)
if cluster is None:
return job_execution
engine = _get_job_engine(cluster, job_execution)
if engine is not None:
job_execution = conductor.job_execution_update(
ctx, job_execution_id,
{'info': {
'status': edp.JOB_STATUS_TOBEKILLED
}})
timeout = CONF.job_canceling_timeout
s_time = timeutils.utcnow()
while timeutils.delta_seconds(s_time, timeutils.utcnow()) < timeout:
if job_execution.info['status'] not in edp.JOB_STATUSES_TERMINATED:
try:
job_info = engine.cancel_job(job_execution)
except Exception as ex:
job_info = None
LOG.exception(
_LE("Error during cancel of job execution %(job)s: "
"%(error)s"), {
'job': job_execution.id,
'error': ex
})
if job_info is not None:
job_execution = _write_job_status(job_execution, job_info)
LOG.info(_LI("Job execution %s was canceled successfully"),
job_execution.id)
return job_execution
context.sleep(3)
job_execution = conductor.job_execution_get(
ctx, job_execution_id)
if not job_execution:
LOG.info(
_LI("Job execution %(job_exec_id)s was deleted. "
"Canceling current operation."),
{'job_exec_id': job_execution_id})
return job_execution
else:
LOG.info(
_LI("Job execution status %(job)s: %(status)s"), {
'job': job_execution.id,
'status': job_execution.info['status']
})
return job_execution
else:
raise e.CancelingFailed(
_('Job execution %s was not canceled') % job_execution.id)
def cancel_job(job_execution_id):
ctx = context.ctx()
job_execution = conductor.job_execution_get(ctx, job_execution_id)
if job_execution.info['status'] in edp.JOB_STATUSES_TERMINATED:
LOG.info(
_LI("Job execution is already finished and shouldn't be"
" canceled"))
return job_execution
cluster = conductor.cluster_get(ctx, job_execution.cluster_id)
if cluster is None:
LOG.info(_LI("Can not cancel this job on a non-existant cluster."))
return job_execution
engine = get_job_engine(cluster, job_execution)
if engine is not None:
job_execution = conductor.job_execution_update(
ctx, job_execution_id,
{'info': {
'status': edp.JOB_STATUS_TOBEKILLED
}})
timeout = CONF.job_canceling_timeout
s_time = timeutils.utcnow()
while timeutils.delta_seconds(s_time, timeutils.utcnow()) < timeout:
if job_execution.info['status'] not in edp.JOB_STATUSES_TERMINATED:
try:
job_info = engine.cancel_job(job_execution)
except Exception as ex:
job_info = None
LOG.warning(
_LW("Error during cancel of job execution: "
"{error}").format(error=ex))
if job_info is not None:
job_execution = _write_job_status(job_execution, job_info)
LOG.info(_LI("Job execution was canceled successfully"))
return job_execution
context.sleep(3)
job_execution = conductor.job_execution_get(
ctx, job_execution_id)
if not job_execution:
LOG.info(
_LI("Job execution was deleted. "
"Canceling current operation."))
return job_execution
else:
LOG.info(
_LI("Job execution status: {status}").format(
status=job_execution.info['status']))
return job_execution
else:
raise e.CancelingFailed(
_('Job execution %s was not canceled') % job_execution.id)