Error Exporting Data Wrangler Flow to Inference Pipeline

Question

I have created a flow in the Data Wrangler component in AWS SageMaker Canvas.  When I click the last node to Export -> Export via Jupyter notebook to -> SageMaker Inference Pipeline; I get an error.
The exact error message is: "An unexpected error occurred when preparing your flow for inference."

Unfortunately, this error is not very descriptive so I cannot diagnose the problem.  I am not using a transform that is not permitted to be exported like "group by" etc.  At this point the error could be anything.  Is there a way to diagnose what the problem is?

My flow has one custom transformation in which I measure the number of days between two dates, this transform is done with pandas as opposed to pySpark as my data is very small < 10K rows and 30 columns.

Answer

Hi there,

Thanks for using AWS Sagemaker.

I understand that you are facing the below issue when trying to export the data flow to inference pipeline.

```
An unexpected error occurred when preparing your flow for inference
```

Please note that the above error can occur if any step types or transformations are not supported by inference endpoint data wrangler flow. Hence, to solve this issue, can you please re-check the transformation steps and retry.

When you choose  Export to --> Sagemaker inference pipeline. Sagemaker will try to validate it and see if the data is good for inference, it should be supported.

Also can you check if you have any empty columns in the input dataset?, one of the workarounds is to drop all the cols where there are null values.

To further understand the issue more in depth as I have limited visibility on your setup, I'd recommend you to reach to AWS Support by creating a support case[+] so that the engineer can investigate further and help you overcome the issue.

[+] Open a support case with AWS using the link: https://console.aws.amazon.com/support/home?#/case/create

Error Exporting Data Wrangler Flow to Inference Pipeline

Relevant content