CommandStep Class

Create an Azure ML Pipeline step that runs a command.

Create an Azure ML Pipeline step that runs a command.

Inheritance

azureml.pipeline.core._python_script_step_base._PythonScriptStepBase

CommandStep

Constructor

CommandStep(command=None, name=None, compute_target=None, runconfig=None, runconfig_pipeline_params=None, inputs=None, outputs=None, params=None, source_directory=None, allow_reuse=True, version=None)

Parameters

Name	Description
command	list or str The command to run or path of the executable/script relative to source_directory. It is required unless it is provided with runconfig. It can be specified with string arguments in a single string or with input/output/PipelineParameter in a list. default value: None
name	str The name of the step. If unspecified, the first word in the command is used. default value: None
compute_target	DsvmCompute or AmlCompute or RemoteCompute or HDInsightCompute or str or tuple The compute target to use. If unspecified, the target from the runconfig is used. This parameter may be specified as a compute target object or the string name of a compute target on the workspace. Optionally if the compute target is not available at pipeline creation time, you may specify a tuple of ('compute target name', 'compute target type') to avoid fetching the compute target object (AmlCompute type is 'AmlCompute' and RemoteCompute type is 'VirtualMachine'). default value: None
runconfig	ScriptRunConfig or RunConfiguration The optional configuration object which encapsulates the information necessary to submit a training run in an experiment. default value: None
runconfig_pipeline_params	<xref:<xref:{str: PipelineParameter}>> Overrides of runconfig properties at runtime using key-value pairs each with name of the runconfig property and PipelineParameter for that property. Supported values: 'NodeCount', 'MpiProcessCountPerNode', 'TensorflowWorkerCount', 'TensorflowParameterServerCount' default value: None
inputs	list[InputPortBinding or DataReference or PortDataReference or PipelineData or <xref:azureml.pipeline.core.pipeline_output_dataset.PipelineOutputDataset> or DatasetConsumptionConfig] A list of input port bindings. default value: None
outputs	list[PipelineData or OutputDatasetConfig or PipelineOutputAbstractDataset or OutputPortBinding] A list of output port bindings. default value: None
params	dict A dictionary of name-value pairs registered as environment variables with "AML_PARAMETER_". default value: None
source_directory	str A folder that contains scripts, conda env, and other resources used in the step. default value: None
allow_reuse	bool Indicates whether the step should reuse previous results when re-run with the same settings. Reuse is enabled by default. If the step contents (scripts/dependencies) as well as inputs and parameters remain unchanged, the output from the previous run of this step is reused. When reusing the step, instead of submitting the job to compute, the results from the previous run are immediately made available to any subsequent steps. If you use Azure Machine Learning datasets as inputs, reuse is determined by whether the dataset's definition has changed, not by whether the underlying data has changed. default value: True
version	str An optional version tag to denote a change in functionality for the step. default value: None
command Required	list or str The command to run or path of the executable/script relative to source_directory. It is required unless it is provided with runconfig. It can be specified with string arguments in a single string or with input/output/PipelineParameter in a list.
name Required	str The name of the step. If unspecified, the first word in the command is used.
compute_target Required	DsvmCompute or AmlCompute or RemoteCompute or HDInsightCompute or str or tuple The compute target to use. If unspecified, the target from the runconfig is used. This parameter may be specified as a compute target object or the string name of a compute target on the workspace. Optionally if the compute target is not available at pipeline creation time, you may specify a tuple of ('compute target name', 'compute target type') to avoid fetching the compute target object (AmlCompute type is 'AmlCompute' and RemoteCompute type is 'VirtualMachine').
runconfig Required	ScriptRunConfig or RunConfiguration The optional configuration object which encapsulates the information necessary to submit a training run in an experiment.
runconfig_pipeline_params Required	<xref:<xref:{str: PipelineParameter}>> Overrides of runconfig properties at runtime using key-value pairs each with name of the runconfig property and PipelineParameter for that property. Supported values: 'NodeCount', 'MpiProcessCountPerNode', 'TensorflowWorkerCount', 'TensorflowParameterServerCount'
inputs Required	list[InputPortBinding or DataReference or PortDataReference or PipelineData or <xref:azureml.pipeline.core.pipeline_output_dataset.PipelineOutputDataset> or DatasetConsumptionConfig] A list of input port bindings.
outputs Required	list[PipelineData or OutputDatasetConfig or PipelineOutputAbstractDataset or OutputPortBinding] A list of output port bindings.
params Required	dict A dictionary of name-value pairs registered as environment variables with "AML_PARAMETER_".
source_directory Required	str A folder that contains scripts, conda env, and other resources used in the step.
allow_reuse Required	bool Indicates whether the step should reuse previous results when re-run with the same settings. Reuse is enabled by default. If the step contents (scripts/dependencies) as well as inputs and parameters remain unchanged, the output from the previous run of this step is reused. When reusing the step, instead of submitting the job to compute, the results from the previous run are immediately made available to any subsequent steps. If you use Azure Machine Learning datasets as inputs, reuse is determined by whether the dataset's definition has changed, not by whether the underlying data has changed.
version Required	str An optional version tag to denote a change in functionality for the step.

Remarks

An CommandStep is a basic, built-in step to run a command on the given compute target. It takes a command as a parameter or from other parameters like runconfig. It also takes other optional parameters like compute target, inputs and outputs. You should use a ScriptRunConfig or RunConfiguration to specify requirements for the CommandStep, such as custom docker image.

The best practice for working with CommandStep is to use a separate folder for the executable or script to run any dependent files associated with the step, and specify that folder with the source_directory parameter. Following this best practice has two benefits. First, it helps reduce the size of the snapshot created for the step because only what is needed for the step is snapshotted. Second, the step's output from a previous run can be reused if there are no changes to the source_directory that would trigger a re-upload of the snapshot.

For the system-known commands source_directory is not required but you can still provide it with any dependent files associated with the step.

The following code example shows how to use a CommandStep in a machine learning training scenario. To list files in linux:

   from azureml.pipeline.steps import CommandStep

   trainStep = CommandStep(name='list step',
                           command='ls -lrt',
                           compute_target=compute_target)

To run a python script:

   from azureml.pipeline.steps import CommandStep

   trainStep = CommandStep(name='train step',
                           command='python train.py arg1 arg2',
                           source_directory=project_folder,
                           compute_target=compute_target)

To run a python script via ScriptRunConfig:

   from azureml.core import ScriptRunConfig
   from azureml.pipeline.steps import CommandStep

   train_src = ScriptRunConfig(source_directory=script_folder,
                               command='python train.py arg1 arg2',
                               environment=my_env)
   trainStep = CommandStep(name='train step',
                           runconfig=train_src)

See https://aka.ms/pl-first-pipeline for more details on creating pipelines in general.

Methods

create_node

Create a node for CommandStep and add it to the specified graph. This method is not intended to be used directly. When a pipeline is instantiated with this step, Azure ML automatically passes the parameters required through this method so that step can be added to a pipeline graph that represents the workflow.

create_node

Create a node for CommandStep and add it to the specified graph.

This method is not intended to be used directly. When a pipeline is instantiated with this step, Azure ML automatically passes the parameters required through this method so that step can be added to a pipeline graph that represents the workflow.

create_node(graph, default_datastore, context)

Parameters

Name	Description
graph Required	Graph The graph object to add the node to.
default_datastore Required	AbstractAzureStorageDatastore or AzureDataLakeDatastore The default datastore.
context Required	<xref:_GraphContext> The graph context.

Returns

Type	Description
Node	The created node.