def _schedule_tick_state(instance, stream, tick_data):
tick = instance.create_schedule_tick(tick_data)
holder = _ScheduleTickHolder(tick=tick, instance=instance)
try:
yield holder
except Exception: # pylint: disable=broad-except
error_data = serializable_error_info_from_exc_info(sys.exc_info())
holder.update_with_status(ScheduleTickStatus.FAILURE, error=error_data)
stream.send(ScheduledExecutionFailed(run_id=None, errors=[error_data]))
finally:
holder.write()
def _schedule_tick_context(instance, stream, tick_data):
tick = instance.create_job_tick(tick_data)
context = _ScheduleLaunchContext(tick=tick, instance=instance, stream=stream)
try:
yield context
except Exception: # pylint: disable=broad-except
error_data = serializable_error_info_from_exc_info(sys.exc_info())
context.update_state(JobTickStatus.FAILURE, error=error_data)
stream.send(ScheduledExecutionFailed(run_id=None, errors=[error_data]))
finally:
context.write()
def _launch_scheduled_executions(instance, repo_location, external_repo,
external_schedule, tick_context):
pipeline_selector = PipelineSelector(
location_name=repo_location.name,
repository_name=external_repo.name,
pipeline_name=external_schedule.pipeline_name,
solid_selection=external_schedule.solid_selection,
)
subset_pipeline_result = repo_location.get_subset_external_pipeline_result(
pipeline_selector)
external_pipeline = ExternalPipeline(
subset_pipeline_result.external_pipeline_data,
external_repo.handle,
)
schedule_execution_data = repo_location.get_external_schedule_execution_data(
instance=instance,
repository_handle=external_repo.handle,
schedule_name=external_schedule.name,
scheduled_execution_time=
None, # No way to know this in general for this scheduler
)
if isinstance(schedule_execution_data, ExternalScheduleExecutionErrorData):
error = schedule_execution_data.error
tick_context.update_state(JobTickStatus.FAILURE, error=error)
tick_context.stream.send(
ScheduledExecutionFailed(run_id=None, errors=[error]))
return
if not schedule_execution_data.run_requests:
# Update tick to skipped state and return
tick_context.update_state(JobTickStatus.SKIPPED)
tick_context.stream.send(ScheduledExecutionSkipped())
return
for run_request in schedule_execution_data.run_requests:
_launch_run(instance, repo_location, external_schedule,
external_pipeline, tick_context, run_request)
tick_context.update_state(JobTickStatus.SUCCESS)
def _launch_scheduled_execution(instance, schedule_def, pipeline, tick,
stream):
pipeline_def = pipeline.get_definition()
# Run should_execute and halt if it returns False
schedule_context = ScheduleExecutionContext(instance)
with user_code_error_boundary(
ScheduleExecutionError,
lambda:
'Error occurred during the execution of should_execute for schedule '
'{schedule_name}'.format(schedule_name=schedule_def.name),
):
should_execute = schedule_def.should_execute(schedule_context)
if not should_execute:
# Update tick to skipped state and return
tick.update_with_status(ScheduleTickStatus.SKIPPED)
stream.send(ScheduledExecutionSkipped())
return
errors = []
run_config = {}
schedule_tags = {}
try:
with user_code_error_boundary(
ScheduleExecutionError,
lambda:
'Error occurred during the execution of run_config_fn for schedule '
'{schedule_name}'.format(schedule_name=schedule_def.name),
):
run_config = schedule_def.get_run_config(schedule_context)
except DagsterUserCodeExecutionError:
error_data = serializable_error_info_from_exc_info(sys.exc_info())
errors.append(error_data)
try:
with user_code_error_boundary(
ScheduleExecutionError,
lambda:
'Error occurred during the execution of tags_fn for schedule '
'{schedule_name}'.format(schedule_name=schedule_def.name),
):
schedule_tags = schedule_def.get_tags(schedule_context)
except DagsterUserCodeExecutionError:
error_data = serializable_error_info_from_exc_info(sys.exc_info())
errors.append(error_data)
pipeline_tags = pipeline_def.tags or {}
check_tags(pipeline_tags, 'pipeline_tags')
tags = merge_dicts(pipeline_tags, schedule_tags)
mode = schedule_def.mode
execution_plan_snapshot = None
try:
execution_plan = create_execution_plan(
pipeline_def,
run_config=run_config,
mode=mode,
)
execution_plan_snapshot = snapshot_from_execution_plan(
execution_plan, pipeline_def.get_pipeline_snapshot_id())
except DagsterInvalidConfigError:
error_data = serializable_error_info_from_exc_info(sys.exc_info())
errors.append(error_data)
# Enter the run in the DB with the information we have
possibly_invalid_pipeline_run = instance.create_run(
pipeline_name=schedule_def.pipeline_name,
run_id=None,
run_config=run_config,
mode=mode,
solids_to_execute=pipeline.solids_to_execute,
step_keys_to_execute=None,
solid_selection=pipeline.solid_selection,
status=None,
root_run_id=None,
parent_run_id=None,
tags=tags,
pipeline_snapshot=pipeline_def.get_pipeline_snapshot(),
execution_plan_snapshot=execution_plan_snapshot,
parent_pipeline_snapshot=pipeline_def.get_parent_pipeline_snapshot(),
)
tick.update_with_status(ScheduleTickStatus.SUCCESS,
run_id=possibly_invalid_pipeline_run.run_id)
# If there were errors, inject them into the event log and fail the run
if len(errors) > 0:
for error in errors:
instance.report_engine_event(
error.message,
possibly_invalid_pipeline_run,
EngineEventData.engine_error(error),
)
instance.report_run_failed(possibly_invalid_pipeline_run)
stream.send(
ScheduledExecutionFailed(
run_id=possibly_invalid_pipeline_run.run_id, errors=errors))
return
# Otherwise the run should be valid so lets launch it
# Need an ExternalPipeline to launch so make one here
recon_repo = pipeline.get_reconstructable_repository()
repo_location = InProcessRepositoryLocation(recon_repo)
external_pipeline = repo_location.get_repository(
recon_repo.get_definition().name).get_full_external_pipeline(
pipeline_def.name)
try:
launched_run = instance.launch_run(
possibly_invalid_pipeline_run.run_id, external_pipeline)
except DagsterLaunchFailedError:
error = serializable_error_info_from_exc_info(sys.exc_info())
instance.report_engine_event(
error.message,
possibly_invalid_pipeline_run,
EngineEventData.engine_error(error),
)
instance.report_run_failed(possibly_invalid_pipeline_run)
stream.send(
ScheduledExecutionFailed(
run_id=possibly_invalid_pipeline_run.run_id, errors=[error]))
return
stream.send(ScheduledExecutionSuccess(run_id=launched_run.run_id))
return
def _launch_run(instance, repo_location, external_schedule, external_pipeline,
tick_context, run_request):
run_config = run_request.run_config
schedule_tags = run_request.tags
execution_plan_snapshot = None
errors = []
try:
external_execution_plan = repo_location.get_external_execution_plan(
external_pipeline,
run_config,
external_schedule.mode,
step_keys_to_execute=None,
)
execution_plan_snapshot = external_execution_plan.execution_plan_snapshot
except DagsterSubprocessError as e:
errors.extend(e.subprocess_error_infos)
except Exception as e: # pylint: disable=broad-except
errors.append(serializable_error_info_from_exc_info(sys.exc_info()))
pipeline_tags = external_pipeline.tags or {}
check_tags(pipeline_tags, "pipeline_tags")
tags = merge_dicts(pipeline_tags, schedule_tags)
# Enter the run in the DB with the information we have
possibly_invalid_pipeline_run = instance.create_run(
pipeline_name=external_schedule.pipeline_name,
run_id=None,
run_config=run_config,
mode=external_schedule.mode,
solids_to_execute=external_pipeline.solids_to_execute,
step_keys_to_execute=None,
solid_selection=external_pipeline.solid_selection,
status=None,
root_run_id=None,
parent_run_id=None,
tags=tags,
pipeline_snapshot=external_pipeline.pipeline_snapshot,
execution_plan_snapshot=execution_plan_snapshot,
parent_pipeline_snapshot=external_pipeline.parent_pipeline_snapshot,
external_pipeline_origin=external_pipeline.get_external_origin(),
)
tick_context.add_run(run_id=possibly_invalid_pipeline_run.run_id,
run_key=run_request.run_key)
# If there were errors, inject them into the event log and fail the run
if len(errors) > 0:
for error in errors:
instance.report_engine_event(
error.message,
possibly_invalid_pipeline_run,
EngineEventData.engine_error(error),
)
instance.report_run_failed(possibly_invalid_pipeline_run)
tick_context.stream.send(
ScheduledExecutionFailed(
run_id=possibly_invalid_pipeline_run.run_id, errors=errors))
return
try:
launched_run = instance.submit_run(
possibly_invalid_pipeline_run.run_id, external_pipeline)
except Exception: # pylint: disable=broad-except
tick_context.stream.send(
ScheduledExecutionFailed(
run_id=possibly_invalid_pipeline_run.run_id,
errors=[serializable_error_info_from_exc_info(sys.exc_info())],
))
return
tick_context.stream.send(
ScheduledExecutionSuccess(run_id=launched_run.run_id))
def _launch_scheduled_execution(instance, repo_location, external_repo,
external_schedule, tick, stream):
pipeline_selector = PipelineSelector(
location_name=repo_location.name,
repository_name=external_repo.name,
pipeline_name=external_schedule.pipeline_name,
solid_selection=external_schedule.solid_selection,
)
subset_pipeline_result = repo_location.get_subset_external_pipeline_result(
pipeline_selector)
external_pipeline = ExternalPipeline(
subset_pipeline_result.external_pipeline_data,
external_repo.handle,
)
schedule_execution_data = repo_location.get_external_schedule_execution_data(
instance=instance,
repository_handle=external_repo.handle,
schedule_name=external_schedule.name,
schedule_execution_data_mode=ScheduleExecutionDataMode.
LAUNCH_SCHEDULED_EXECUTION,
scheduled_execution_time=
None, # No way to know this in general for this scheduler
)
run_config = {}
schedule_tags = {}
execution_plan_snapshot = None
errors = []
if isinstance(schedule_execution_data, ExternalScheduleExecutionErrorData):
error = schedule_execution_data.error
tick.update_with_status(ScheduleTickStatus.FAILURE, error=error)
stream.send(ScheduledExecutionFailed(run_id=None, errors=[error]))
return
elif not schedule_execution_data.should_execute:
# Update tick to skipped state and return
tick.update_with_status(ScheduleTickStatus.SKIPPED)
stream.send(ScheduledExecutionSkipped())
return
else:
run_config = schedule_execution_data.run_config
schedule_tags = schedule_execution_data.tags
try:
external_execution_plan = repo_location.get_external_execution_plan(
external_pipeline,
run_config,
external_schedule.mode,
step_keys_to_execute=None,
)
execution_plan_snapshot = external_execution_plan.execution_plan_snapshot
except DagsterSubprocessError as e:
errors.extend(e.subprocess_error_infos)
except Exception as e: # pylint: disable=broad-except
errors.append(serializable_error_info_from_exc_info(
sys.exc_info()))
pipeline_tags = external_pipeline.tags or {}
check_tags(pipeline_tags, "pipeline_tags")
tags = merge_dicts(pipeline_tags, schedule_tags)
# Enter the run in the DB with the information we have
possibly_invalid_pipeline_run = instance.create_run(
pipeline_name=external_schedule.pipeline_name,
run_id=None,
run_config=run_config,
mode=external_schedule.mode,
solids_to_execute=external_pipeline.solids_to_execute,
step_keys_to_execute=None,
solid_selection=external_pipeline.solid_selection,
status=None,
root_run_id=None,
parent_run_id=None,
tags=tags,
pipeline_snapshot=external_pipeline.pipeline_snapshot,
execution_plan_snapshot=execution_plan_snapshot,
parent_pipeline_snapshot=external_pipeline.parent_pipeline_snapshot,
)
tick.update_with_status(ScheduleTickStatus.SUCCESS,
run_id=possibly_invalid_pipeline_run.run_id)
# If there were errors, inject them into the event log and fail the run
if len(errors) > 0:
for error in errors:
instance.report_engine_event(
error.message,
possibly_invalid_pipeline_run,
EngineEventData.engine_error(error),
)
instance.report_run_failed(possibly_invalid_pipeline_run)
stream.send(
ScheduledExecutionFailed(
run_id=possibly_invalid_pipeline_run.run_id, errors=errors))
return
try:
launched_run = instance.launch_run(
possibly_invalid_pipeline_run.run_id, external_pipeline)
except Exception: # pylint: disable=broad-except
stream.send(
ScheduledExecutionFailed(
run_id=possibly_invalid_pipeline_run.run_id, errors=[error]))
return
stream.send(ScheduledExecutionSuccess(run_id=launched_run.run_id))
return
