kfp package

class kfp.LocalClient(pipeline_root: str = None)[source]

Bases: object

class ExecutionMode(mode: str = 'docker', images_to_exclude: List[str] = [], ops_to_exclude: List[str] = [])[source]

Bases: object

Configuration to decide whether the client executes a component in docker or in local process.

DOCKER = 'docker'
LOCAL = 'local'
images_to_exclude
mode
ops_to_exclude
create_run_from_pipeline_func(pipeline_func: Callable, arguments: Mapping[str, str], execution_mode: kfp._local_client.LocalClient.ExecutionMode = <kfp._local_client.LocalClient.ExecutionMode object>)[source]

Runs a pipeline locally, either using Docker or in a local process.

Parameters:
  • pipeline_func – pipeline function
  • arguments – Arguments to the pipeline function provided as a dict, reference to kfp.client.create_run_from_pipeline_func
  • execution_mode – Configuration to decide whether the client executes component in docker or in local process.
kfp.run_pipeline_func_locally(pipeline_func: Callable, arguments: Mapping[str, str], local_client: kfp._local_client.LocalClient = None, execution_mode: kfp._local_client.LocalClient.ExecutionMode = <kfp._local_client.LocalClient.ExecutionMode object>)[source]

Runs a pipeline locally, either using Docker or in a local process.

Feature stage: [Alpha](https://github.com/kubeflow/pipelines/blob/master/docs/release/feature-stages.md#alpha)

In this alpha implementation, we support:
  • Control flow: Condition, ParallelFor
  • Data passing: InputValue, InputPath, OutputPath
And we don’t support:
  • Control flow: ExitHandler, Graph, SubGraph
  • ContainerOp with environment variables, init containers, sidecars, pvolumes
  • ResourceOp
  • VolumeOp
  • Caching
Parameters:
  • pipeline_func – A function that describes a pipeline by calling components and composing them into execution graph.
  • arguments – Arguments to the pipeline function provided as a dict. reference to kfp.client.create_run_from_pipeline_func
  • local_client – Optional. An instance of kfp.LocalClient
  • execution_mode – Configuration to decide whether the client executes component in docker or in local process.
kfp.run_pipeline_func_on_cluster(pipeline_func: Callable, arguments: Mapping[str, str], run_name: str = None, experiment_name: str = None, kfp_client: kfp._client.Client = None, pipeline_conf: kfp.dsl._pipeline.PipelineConf = None)[source]

Runs pipeline on KFP-enabled Kubernetes cluster.

This command compiles the pipeline function, creates or gets an experiment and submits the pipeline for execution.

Feature stage: [Alpha](https://github.com/kubeflow/pipelines/blob/07328e5094ac2981d3059314cc848fbb71437a76/docs/release/feature-stages.md#alpha)

Parameters:
  • pipeline_func – A function that describes a pipeline by calling components
  • composing them into execution graph. (and) –
  • arguments – Arguments to the pipeline function provided as a dict.
  • run_name – Optional. Name of the run to be shown in the UI.
  • experiment_name – Optional. Name of the experiment to add the run to.
  • kfp_client – Optional. An instance of kfp.Client configured for the desired KFP cluster.
  • pipeline_conf – Optional. kfp.dsl.PipelineConf instance. Can specify op transforms, image pull secrets and other pipeline-level configuration options.