def execute(config, docker_registry, base_image, namespace=None):
"""
Runs the LightGBM CLI in a single pod in user's Kubeflow cluster.
Users can configure it to be a train, predict, and other supported tasks
by using the right config.
Please refere https://github.com/microsoft/LightGBM/blob/master/docs/Parameters.rst
for more information on config options.
Attributes:
config: LightGBM config - Ref https://github.com/microsoft/LightGBM/blob/master/docs/Parameters.rst
docker_registry: registry to push the built docker image
base_image: base image to use for this job. It should have lightgbm installed and should be in PATH variable.
namespace: Kubernetes namespace to use
"""
if namespace is None:
namespace = "kubeflow"
config_file_name = None
if isinstance(config, str):
config_file_name = config
config = _load_config_file(config)
elif isinstance(config, dict):
config_file_name = _save_to_config_file(config)
else:
raise RuntimeError("config should be of type dict or string(filepath) "
"but got {}".format(type(dict)))
output_map = generate_context_files(config, config_file_name)
preprocessor = BasePreProcessor(
command=[ENTRYPOINT], output_map=output_map)
builder = AppendBuilder(registry=docker_registry,
base_image=base_image, preprocessor=preprocessor)
builder.build()
pod_spec = builder.generate_pod_spec()
deployer = Job(namespace=namespace, pod_spec_mutators=[
fairing.cloud.gcp.add_gcp_credentials_if_exists])
deployer.deploy(pod_spec)
def get_training_deployer(self):
return Job(namespace=self._namespace,
pod_spec_mutators=[gcp.add_gcp_credentials_if_exists])
def get_training_deployer(self):
return Job(namespace=self._namespace)
def get_training_deployer(self, pod_spec_mutators=None):
return Job(self._namespace, pod_spec_mutators=pod_spec_mutators)
def get_training_deployer(self, pod_spec_mutators=None):
pod_spec_mutators = pod_spec_mutators or []
pod_spec_mutators.append(aws.add_aws_credentials_if_exists)
return Job(namespace=self._namespace,
pod_spec_mutators=pod_spec_mutators)
def execute(config,
docker_registry,
base_image="gcr.io/kubeflow-fairing/lightgbm:latest",
namespace="kubeflow",
stream_log=True,
cores_per_worker=None,
memory_per_worker=None,
pod_spec_mutators=None):
"""
Runs the LightGBM CLI in a single pod in user's Kubeflow cluster.
Users can configure it to be a train, predict, and other supported tasks
by using the right config.
Please refere https://github.com/microsoft/LightGBM/blob/master/docs/Parameters.rst
for more information on config options.
Attributes:
config: LightGBM config - Ref https://github.com/microsoft/LightGBM/blob/master/docs/Parameters.rst
docker_registry: registry to push the built docker image
base_image: base image to use for this job. It should have lightgbm installed and should be in PATH variable.
namespace: Kubernetes namespace to use
stream_log: True - streams logs from the first worker in the training job after job launch till the training is finished.
Flase - no logs are streamed after the job launch. An async job launch use case.
cores_per_worker: #cpu cores allocated per worker
memory_per_worker: memory allocated per worker in GB, it can be fractional.
pod_spec_mutators: list of functions that is used to mutate the podsspec. e.g. fairing.cloud.gcp.add_gcp_credentials_if_exists
This can used to set things like volumes and security context.
"""
config_file_name = None
if isinstance(config, str):
config_file_name = config
config = utils.load_properties_config_file(config)
elif isinstance(config, dict):
config_file_name = utils.save_properties_config_file(config)
else:
raise RuntimeError("config should be of type dict or string(filepath) "
"but got {}".format(type(dict)))
utils.scrub_fields(config, BLACKLISTED_FIELDS)
_, num_machines = utils.get_config_value(config, NUM_MACHINES_FILEDS)
num_machines = num_machines or 1
if num_machines:
try:
num_machines = int(num_machines)
except ValueError:
raise ValueError(
"num_machines value in config should be an int >= 1 "
"but got {}".format(config.get('num_machines')))
if num_machines < 1:
raise ValueError(
"num_machines value in config should >= 1 but got {}".format(
num_machines))
if num_machines > 1:
config['machine_list_file'] = "mlist.txt"
output_map = generate_context_files(config, config_file_name,
num_machines > 1)
preprocessor = BasePreProcessor(command=[ENTRYPOINT],
output_map=output_map)
builder = AppendBuilder(registry=docker_registry,
base_image=base_image,
preprocessor=preprocessor)
builder.build()
pod_spec = builder.generate_pod_spec()
pod_spec_mutators = pod_spec_mutators or []
pod_spec_mutators.append(fairing.cloud.gcp.add_gcp_credentials_if_exists)
pod_spec_mutators.append(
k8s_utils.get_resource_mutator(cores_per_worker, memory_per_worker))
if num_machines == 1:
# non-distributed mode
deployer = Job(namespace=namespace,
pod_spec_mutators=pod_spec_mutators,
stream_log=stream_log)
else:
# distributed mode
deployer = TfJob(namespace=namespace,
pod_spec_mutators=pod_spec_mutators,
chief_count=1,
worker_count=num_machines - 1,
stream_log=stream_log)
deployer.deploy(pod_spec)
return deployer
def execute(config,
docker_registry,
base_image="gcr.io/kubeflow-fairing/lightgbm:latest",
namespace="kubeflow"):
"""
Runs the LightGBM CLI in a single pod in user's Kubeflow cluster.
Users can configure it to be a train, predict, and other supported tasks
by using the right config.
Please refere https://github.com/microsoft/LightGBM/blob/master/docs/Parameters.rst
for more information on config options.
Attributes:
config: LightGBM config - Ref https://github.com/microsoft/LightGBM/blob/master/docs/Parameters.rst
docker_registry: registry to push the built docker image
base_image: base image to use for this job. It should have lightgbm installed and should be in PATH variable.
namespace: Kubernetes namespace to use
"""
config_file_name = None
if isinstance(config, str):
config_file_name = config
config = utils.load_properties_config_file(config)
elif isinstance(config, dict):
config_file_name = utils.save_properties_config_file(config)
else:
raise RuntimeError("config should be of type dict or string(filepath) "
"but got {}".format(type(dict)))
utils.scrub_fields(config, BLACKLISTED_FIELDS)
_, num_machines = utils.get_config_value(config, NUM_MACHINES_FILEDS)
num_machines = num_machines or 1
if num_machines:
try:
num_machines = int(num_machines)
except ValueError:
raise ValueError(
"num_machines value in config should be an int >= 1 "
"but got {}".format(config.get('num_machines')))
if num_machines < 1:
raise ValueError(
"num_machines value in config should >= 1 but got {}".format(
num_machines))
if num_machines > 1:
config['machine_list_file'] = "mlist.txt"
output_map = generate_context_files(config, config_file_name,
num_machines > 1)
preprocessor = BasePreProcessor(command=[ENTRYPOINT],
output_map=output_map)
builder = AppendBuilder(registry=docker_registry,
base_image=base_image,
preprocessor=preprocessor)
builder.build()
pod_spec = builder.generate_pod_spec()
if num_machines == 1:
# non-distributed mode
deployer = Job(namespace=namespace,
pod_spec_mutators=[
fairing.cloud.gcp.add_gcp_credentials_if_exists
])
else:
# distributed mode
deployer = TfJob(namespace=namespace,
pod_spec_mutators=[
fairing.cloud.gcp.add_gcp_credentials_if_exists
],
chief_count=1,
worker_count=num_machines - 1)
deployer.deploy(pod_spec)
def submit(self):
self._build()
deployer = Job()
deployer.deploy(self.pod_spec)