Unable to create updated version of component in Azure ML Registry

[bastbu]

• Package Name: azure-ai-ml
• Package Version: 1.19.0
• Operating System: Dev Container (Ubuntu)
• Python Version: Python 3.11.10

Describe the bug

When creating the same component in an AML Registry twice (the second time with an updated version and code, but the same environment), the update of such a component fails.

The following error message gets thrown:

Exception: ValueError: No value for given attribute
Stack:
  
  [REDACTED]
  
    _ = client.components.create_or_update(component)
        ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_telemetry/activity.py", line 372, in wrapper
    return_value = f(*args, **kwargs)
                   ^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_component_operations.py", line 619, in create_or_update
    self._resolve_arm_id_or_upload_dependencies(component)
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_component_operations.py", line 795, in _resolve_arm_id_or_upload_dependencies
    self._resolve_dependencies_for_component(component, resolver)
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_component_operations.py", line 823, in _resolve_dependencies_for_component
    self._try_resolve_environment_for_component(
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_component_operations.py", line 783, in _try_resolve_environment_for_component
    parent.environment = resolver(parent.environment, azureml_type=AzureMLResourceType.ENVIRONMENT)
                         ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_operation_orchestrator.py", line 240, in get_asset_arm_id
    result = self._get_environment_arm_id(asset, register_asset=register_asset)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_operation_orchestrator.py", line 318, in _get_environment_arm_id
    env_response = self._environments.create_or_update(environment)  # type: ignore[attr-defined]
                   ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_telemetry/activity.py", line 292, in wrapper
    return f(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_environment_operations.py", line 205, in create_or_update
    raise ex
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_environment_operations.py", line 172, in create_or_update
    environment = _check_and_upload_env_build_context(
                  ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_artifacts/_artifact_utilities.py", line 538, in _check_and_upload_env_build_context
    uploaded_artifact = _upload_to_datastore(
                        ^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_artifacts/_artifact_utilities.py", line 384, in _upload_to_datastore
    artifact = upload_artifact(
               ^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_artifacts/_artifact_utilities.py", line 240, in upload_artifact
    datastore_info = get_datastore_info(datastore_operation, datastore_name)
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_artifacts/_artifact_utilities.py", line 99, in get_datastore_info
    datastore = operations.get(name) if name else operations.get_default()
                ^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_telemetry/activity.py", line 292, in wrapper
    return f(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/operations/_datastore_operations.py", line 155, in get
    datastore_resource = self._operation.get(
                         ^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/core/tracing/decorator.py", line 94, in wrapper_use_tracer
    return func(*args, **kwargs)
           ^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_restclient/v2024_07_01_preview/operations/_datastores_operations.py", line 496, in get
    request = build_get_request(
              ^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/azure/ai/ml/_restclient/v2024_07_01_preview/operations/_datastores_operations.py", line 147, in build_get_request
    "workspaceName": _SERIALIZER.url("workspace_name", workspace_name, 'str', pattern=r'^[a-zA-Z0-9][a-zA-Z0-9_-]{2,32}$'),
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/msrest/serialization.py", line 652, in url
    output = self.serialize_data(data, data_type, **kwargs)
             ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "/workspaces/.venv/lib/python3.11/site-packages/msrest/serialization.py", line 760, in serialize_data
    raise ValueError("No value for given attribute")

To Reproduce

Use the following component definition:

$schema: https://azuremlschemas.azureedge.net/latest/commandComponent.schema.json
type: command

name: <anything>

inputs:
  <...>

code: ./code
environment:
  build:
    path: ./environment
    dockerfile_path: Dockerfile

command: >-
  python main.py <...>

We use the following code to create a component:

from azure.ai.ml import load_component

component = load_component(<path-to-component>)
_ = client.components.create_or_update(component)

where client is an instance of MLClient that is configured to point to a registry.

Steps to reproduce the behavior:

1. Upload a component
2. Change the code and component version, upload the component again.
3. The attached error message gets thrown.

Attempted workaround

We currently work around this by setting the environment name and version in the CommandComponent, as the properties are public and mutable, i.e.

component.environment.name = component.name
component.environment.version = component.version

With this the component creation is successful, but we get the following warning, which indicates that something is off:

Warning: the provided asset name '<component_name>' will not be used for anonymous registration

Is this workaround appropriate or do you have another suggestion?

Expected behavior

We should be able to upload new component versions into the Azure ML Registry.

[github-actions]

Thanks for the feedback! We are routing this to the appropriate team for follow-up. cc @Azure/azure-ml-sdk @azureml-github.