Unable to upload pipeline using decorators to Azure

[zadaninck]

Describe the bug

Hi, I am setting up a ClearML pipeline using decorators. For one of the pipeline steps I get the following error:
clearml.storage - ERROR - Exception encountered while uploading Failed uploading object <azure_link> pipeline logic/custom pipeline logic artifacts/result_create_dataset_dataset_tags/result_create_dataset_dataset_tags.pkl (400): <?xml version="1.0" encoding="utf-8"?><Error><Code>InvalidQueryParameterValue</Code><Message>Value for one of the query parameters specified in the request URI is invalid.

The issue occurs with one of my arguments, which contains an array of tags that I want to give to my dataset. When removing this argument, everything works fine.

I have also recreated the pipeline using functions, and then it causes no issues, but preferably I would use the decorator method.

To reproduce

@PipelineDecorator.component(return_values=["dataset"], cache=True, task_type=TaskTypes.data_processing)
def create_dataset(path, dataset_name, dataset_project, dataset_tags, description):
    print("step_one", flush=True)

@PipelineDecorator.pipeline(name="custom pipeline logic", project="examples", version="0.0.5")
def executing_pipeline():
  dataset = create_dataset(
          path=path,
          dataset_name="dataset_name", 
          dataset_project="dataset_project",
          dataset_tags=["tag1", "tag2", "tag3", "tag4"],
          description="Description.",
   )

Environment

• Server type: self-hosted ClearML server with Azure data storage
• ClearML SDK Version: 1.17.0
• ClearML Server Version (Only for self hosted). WebApp: 1.17.1-554 • Server: 1.17.1-554 • API: 2.31
• Python Version: 3.10.12
• OS: Linux

[eugen-ajechiloae-clearml]

Hi @zadaninck ! I wasn't able to reproduce this unfortunately.
There should not be many differences when it comes to artifact uploading between pipelines from decorators and pipelines from functions.
Could it be that your blob storage/azure-storage-blob version does not do well with spaces (i.e. the spaces in custom pipeline logic) ?

[zadaninck]

Hi @eugen-ajechiloae-clearml, I have tried removing the spaces, but the same error still occurs. It seems to be related specifically to the dataset_tags argument.