Skip to content

Unable to use Env Variables from PipelineComponentBatchDeployment #42255

@TopAgrume

Description

@TopAgrume
  • Package Name: azure-ai-ml
  • Package Version: 1.10.1
  • Operating System: Linux
  • Python Version: 3.12

Describe the bug
Environment variables specified during the deployment of batch pipeline components are not being set correctly. When attempting to retrieve these variables, they appear as an empty dictionary.

To Reproduce (code example)

training_component = training_pipeline().component
pipeline_component = client.components.create_or_update(training_component)

batch_deployment = PipelineComponentBatchDeployment(
    name="deploy-name",
    description="Test",
    endpoint_name="endpoint_name",
    component=pipeline_component,
    settings={
        "continue_on_step_failure": "False",
        "default_compute": "gpu-config",
    },
    environment_variables={"test1": 1, "test2": 2}
)
# Here: batch_deployment.environment_variables == {"test1": 1, "test2": 2}
client.batch_deployments.begin_create_or_update(batch_deployment).result()

# Test environment variables:
deployment_test = ml_client.batch_deployments.get("deploy-name", "endpoint_name")
deployment_test.environment_variables   # Expected: {"test1": 1, "test2": 2}, Actual: EMPTY DICT

# In another script, when triggering the batch endpoint and printing all the environment variables inside running components, "test1"=1 and "test2"=2 do not appear!

Expected behavior
The environment variables specified during the deployment should be accessible and correctly set when running batch pipeline components. This allows for the use of these variables within the components during execution.

Doc link:
https://learn.microsoft.com/en-us/python/api/azure-ai-ml/azure.ai.ml.entities.batchdeployment?view=azure-python

Metadata

Metadata

Labels

ClientThis issue points to a problem in the data-plane of the library.Machine LearningService AttentionWorkflow: This issue is responsible by Azure service team.customer-reportedIssues that are reported by GitHub users external to the Azure organization.issue-addressedWorkflow: The Azure SDK team believes it to be addressed and ready to close.questionThe issue doesn't require a change to the product in order to be resolved. Most issues start as that

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions