`az ml job download` fails with python error

[john-terraform]

Describe the bug

Command Name
az ml job download (FYI: az feedback initially has set this field as Unknown for some reason)

Errors:

1. 'ml' is misspelled or not recognized by the system. (the error provided by `az feedback`)

2. `Type <class 'azext_mlv2.manual.vendored_curated_sdk.azure.ai.ml.entities._job.pipeline._io.base.NodeOutput'> is not supported for operation bool().` (the actual error in the terminal)

To Reproduce:

Steps to reproduce the behavior. Note that argument values have been redacted, as they may contain sensitive information.

• az extension add -n ml -y
• az ml job download --name {} --resource-group {} --workspace-name {} --download-path {}

Expected Behavior

The job should be downloaded successfully.

Environment Summary

Linux-5.15.0-1023-azure-x86_64-with-glibc2.31, Ubuntu 20.04.5 LTS
Python 3.10.8
Installer: DEB

azure-cli 2.42.0

Extensions:
ml 2.11.0

Dependencies:
msal 1.20.0
azure-mgmt-resource 21.1.0b1

Additional Context

[john-terraform]

I've managed to reproduce the error within 2 different environments that are hosted with Azure VMs. The error started appearing within the following period: ~Nov 9 - Nov 11.

[yonzhan]

route to CXP team

[john-terraform]

Any updates on this case so far?

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

Issue Details

---

Describe the bug

Command Name
az ml job download (FYI: az feedback initially has set this field as Unknown for some reason)

Errors:

1. 'ml' is misspelled or not recognized by the system. (the error provided by `az feedback`)

2. `Type <class 'azext_mlv2.manual.vendored_curated_sdk.azure.ai.ml.entities._job.pipeline._io.base.NodeOutput'> is not supported for operation bool().` (the actual error in the terminal)

To Reproduce:

Steps to reproduce the behavior. Note that argument values have been redacted, as they may contain sensitive information.

• az extension add -n ml -y
• az ml job download --name {} --resource-group {} --workspace-name {} --download-path {}

Expected Behavior

The job should be downloaded successfully.

Environment Summary

Linux-5.15.0-1023-azure-x86_64-with-glibc2.31, Ubuntu 20.04.5 LTS
Python 3.10.8
Installer: DEB

azure-cli 2.42.0

Extensions:
ml 2.11.0

Dependencies:
msal 1.20.0
azure-mgmt-resource 21.1.0b1

Additional Context

Author:	iddqdiddqd
Assignees:	-
Labels:	Service Attention, Machine Learning, customer-reported, Auto-Assign
Milestone:	-

[luigiw]

@iddqdiddqd , thx for reporting this issue. Can you repro this consistently? Can you try adding "--debug" after your command and share the results? Thank you.

[zhengfeiwang]

@iddqdiddqd , for second error, it should be a known bug and it has been already fixed in new release. Please try ml extension with version 2.12.0 and let me know if it works for you. Thanks!

[john-terraform]

Thank you for your replies @zhengfeiwang and @luigiw. Unfortunately, I cannot reproduce this issue since I've recreated my Python environment and installed az-cli from scratch. The mlv2 extension az ml job download command works out of the box now.

However, now I am not sure whether I should keep az-cli regularly updated along with the mlv2 extension in order not to face such the issues in future.

[john-terraform]

@zhengfeiwang, is there any list where I can find pinned pip dependencies for a specific az-cli release?

[zhengfeiwang]

@zhengfeiwang, is there any list where I can find pinned pip dependencies for a specific az-cli release?

@luigiw do we have any public page on this? I remember that we have one page in internal DevOps repo.