Pipeline code upload location invalid (standalone job pattern applied) for steps included in Condition step

[AndreiVoinovTR]

Describe the bug
I am deploying a SageMaker Pipeline with some Training and Processing steps included in a ConditionStep - as if_steps. I
am noticing that local code (scripts) for these Training and Processing steps is being uploaded to s3 to locations with patterns like s3://<bucket>/<prefix>/<base_job_name><tm_stamp>/... while it should be something like s3://<bucket>/<prefix>/<pipeline_name>/<code_hash>/....

I tried to trace it in the SDK code and I believe the issue is due to the fact that all the build_steps() calls for the whole list of steps (under this ConditionStep) are executed within a single _pipeline_config_manager context (of the 'parent' condition step) - with the code_hash property of None.

See https://github.com/aws/sagemaker-python-sdk/blob/master/src/sagemaker/workflow/utilities.py#L120

To reproduce
Create and deploy a Pipeline with Processing/Training steps (with local script code) wrapped in a Condition step. Observe the resulting code upload s3 location.

Expected behavior
Local code for pipeline steps should be uploaded to a pipeline-specific location in s3.

System information
SageMaker Python SDK version: 2.188.0

[qidewenwhen]

Add a note for following oncalls to avoid duplicate work:
This is issue has been fixed and internally reviewed. Will be published later

[AndreiVoinovTR]

@qidewenwhen this is good news, is there an estimated timeline for when the fix can be published?

[qidewenwhen]

Hi @AndreiVoinovTR sorry for the delay. The fix should have been released in this version: https://github.com/aws/sagemaker-python-sdk/releases/tag/v2.199.0

Could you upgrade the SDK to latest version and let us know if the issue resolved on your end? Thanks!

[qidewenwhen]

Closing the issue as we did not hear back in past 3 weeks. Feel free to reopen if you have further questions. Thanks!