def queue_job(self, job_wrapper):
job_destination = job_wrapper.job_destination
self._populate_parameter_defaults(job_destination)
command_line, client, remote_job_config, compute_environment = self.__prepare_job(
job_wrapper, job_destination)
if not command_line:
return
try:
dependencies_description = PulsarJobRunner.__dependencies_description(
client, job_wrapper)
rewrite_paths = not PulsarJobRunner.__rewrite_parameters(client)
unstructured_path_rewrites = {}
output_names = []
if compute_environment:
unstructured_path_rewrites = compute_environment.unstructured_path_rewrites
output_names = compute_environment.output_names()
if self.app.config.metadata_strategy == "legacy":
# Drop this branch in 19.09.
metadata_directory = job_wrapper.working_directory
else:
metadata_directory = os.path.join(
job_wrapper.working_directory, "metadata")
client_job_description = ClientJobDescription(
command_line=command_line,
input_files=self.get_input_files(job_wrapper),
client_outputs=self.__client_outputs(client, job_wrapper),
working_directory=job_wrapper.tool_working_directory,
metadata_directory=metadata_directory,
tool=job_wrapper.tool,
config_files=job_wrapper.extra_filenames,
dependencies_description=dependencies_description,
env=client.env,
rewrite_paths=rewrite_paths,
arbitrary_files=unstructured_path_rewrites,
touch_outputs=output_names,
)
job_id = pulsar_submit_job(client, client_job_description,
remote_job_config)
log.info("Pulsar job submitted with job_id %s" % job_id)
job_wrapper.set_job_destination(job_destination, job_id)
job_wrapper.change_state(model.Job.states.QUEUED)
except Exception:
job_wrapper.fail("failure running job", exception=True)
log.exception("failure running job %d", job_wrapper.job_id)
return
pulsar_job_state = AsynchronousJobState()
pulsar_job_state.job_wrapper = job_wrapper
pulsar_job_state.job_id = job_id
pulsar_job_state.old_state = True
pulsar_job_state.running = False
pulsar_job_state.job_destination = job_destination
self.monitor_job(pulsar_job_state)
def queue_job(self, job_wrapper):
job_destination = job_wrapper.job_destination
self._populate_parameter_defaults(job_destination)
command_line, client, remote_job_config, compute_environment = self.__prepare_job(job_wrapper, job_destination)
if not command_line:
return
try:
dependencies_description = PulsarJobRunner.__dependencies_description(client, job_wrapper)
rewrite_paths = not PulsarJobRunner.__rewrite_parameters(client)
unstructured_path_rewrites = {}
output_names = []
if compute_environment:
unstructured_path_rewrites = compute_environment.unstructured_path_rewrites
output_names = compute_environment.output_names()
client_job_description = ClientJobDescription(
command_line=command_line,
input_files=self.get_input_files(job_wrapper),
client_outputs=self.__client_outputs(client, job_wrapper),
working_directory=job_wrapper.tool_working_directory,
metadata_directory=job_wrapper.working_directory,
tool=job_wrapper.tool,
config_files=job_wrapper.extra_filenames,
dependencies_description=dependencies_description,
env=client.env,
rewrite_paths=rewrite_paths,
arbitrary_files=unstructured_path_rewrites,
touch_outputs=output_names,
)
job_id = pulsar_submit_job(client, client_job_description, remote_job_config)
log.info("Pulsar job submitted with job_id %s" % job_id)
job_wrapper.set_job_destination(job_destination, job_id)
job_wrapper.change_state(model.Job.states.QUEUED)
except Exception:
job_wrapper.fail("failure running job", exception=True)
log.exception("failure running job %d", job_wrapper.job_id)
return
pulsar_job_state = AsynchronousJobState()
pulsar_job_state.job_wrapper = job_wrapper
pulsar_job_state.job_id = job_id
pulsar_job_state.old_state = True
pulsar_job_state.running = False
pulsar_job_state.job_destination = job_destination
self.monitor_job(pulsar_job_state)
def queue_job(self, job_wrapper):
job_destination = job_wrapper.job_destination
self._populate_parameter_defaults(job_destination)
command_line, client, remote_job_config, compute_environment, remote_container = self.__prepare_job(job_wrapper, job_destination)
if not command_line:
return
try:
dependencies_description = PulsarJobRunner.__dependencies_description(client, job_wrapper)
rewrite_paths = not PulsarJobRunner.__rewrite_parameters(client)
path_rewrites_unstructured = {}
output_names = []
if compute_environment:
path_rewrites_unstructured = compute_environment.path_rewrites_unstructured
output_names = compute_environment.output_names()
client_inputs_list = []
for input_dataset_wrapper in job_wrapper.get_input_paths():
# str here to resolve false_path if set on a DatasetPath object.
path = str(input_dataset_wrapper)
object_store_ref = {
"dataset_id": input_dataset_wrapper.dataset_id,
"dataset_uuid": str(input_dataset_wrapper.dataset_uuid),
"object_store_id": input_dataset_wrapper.object_store_id,
}
client_inputs_list.append(ClientInput(path, CLIENT_INPUT_PATH_TYPES.INPUT_PATH, object_store_ref=object_store_ref))
for input_extra_path in compute_environment.path_rewrites_input_extra.keys():
# TODO: track dataset for object_Store_ref...
client_inputs_list.append(ClientInput(input_extra_path, CLIENT_INPUT_PATH_TYPES.INPUT_EXTRA_FILES_PATH))
for input_metadata_path in compute_environment.path_rewrites_input_metadata.keys():
# TODO: track dataset for object_Store_ref...
client_inputs_list.append(ClientInput(input_metadata_path, CLIENT_INPUT_PATH_TYPES.INPUT_METADATA_PATH))
input_files = None
client_inputs = ClientInputs(client_inputs_list)
else:
input_files = self.get_input_files(job_wrapper)
client_inputs = None
if self.app.config.metadata_strategy == "legacy":
# Drop this branch in 19.09.
metadata_directory = job_wrapper.working_directory
else:
metadata_directory = os.path.join(job_wrapper.working_directory, "metadata")
remote_pulsar_app_config = job_destination.params.get("pulsar_app_config", {})
job_directory_files = []
config_files = job_wrapper.extra_filenames
tool_script = os.path.join(job_wrapper.working_directory, "tool_script.sh")
if os.path.exists(tool_script):
log.debug("Registering tool_script for Pulsar transfer [%s]" % tool_script)
job_directory_files.append(tool_script)
client_job_description = ClientJobDescription(
command_line=command_line,
input_files=input_files,
client_inputs=client_inputs, # Only one of these input defs should be non-None
client_outputs=self.__client_outputs(client, job_wrapper),
working_directory=job_wrapper.tool_working_directory,
metadata_directory=metadata_directory,
tool=job_wrapper.tool,
config_files=config_files,
dependencies_description=dependencies_description,
env=client.env,
rewrite_paths=rewrite_paths,
arbitrary_files=path_rewrites_unstructured,
touch_outputs=output_names,
remote_pulsar_app_config=remote_pulsar_app_config,
job_directory_files=job_directory_files,
container=None if not remote_container else remote_container.container_id,
)
job_id = pulsar_submit_job(client, client_job_description, remote_job_config)
log.info("Pulsar job submitted with job_id %s" % job_id)
job_wrapper.set_job_destination(job_destination, job_id)
job_wrapper.change_state(model.Job.states.QUEUED)
except Exception:
job_wrapper.fail("failure running job", exception=True)
log.exception("failure running job %d", job_wrapper.job_id)
return
pulsar_job_state = AsynchronousJobState()
pulsar_job_state.job_wrapper = job_wrapper
pulsar_job_state.job_id = job_id
pulsar_job_state.old_state = True
pulsar_job_state.running = False
pulsar_job_state.job_destination = job_destination
self.monitor_job(pulsar_job_state)
