search

databricks-demos / dbdemos

Demos to implement your Databricks Lakehouse
Other
255 stars 80 forks source link

02-Deploy-RAG-Chatbot-Model - Register chain to MLFlow #100

Closed Gasmane closed 5 months ago

Gasmane commented 5 months ago

Saving model to Unity Catalog registry step in the notebook.

I am trying to register a model for batch processing. However, I am encountering an issue related to accessing an AWS S3 bucket.

When I try to register the model using miflow langchain.log_model0, I encounter an AWS AccessDenied error during the GetBucketLocation operation.

ClientError: An error occurred (AccessDenied) when calling the GetBucketLocation operation: Access Denied

This operation is used to retrieve the AWS region where the S3 bucket is located. I have received AWS PowerUser access, but the issue persists. The Artifact URI uses DBFS paths, not direct S3 paths, meaning the underlying S3 bucket is managed by Databricks, not accessed directly.

Do you have any idea on how to solve this problem or what steps my Databricks administrator has to make (when it comes to accesses)? Please let me know if you need any additional information to help resolve these issues.

QuentinAmbard commented 5 months ago

Hi @Gasmane , I'm afraid this is a bit specific to your setup. I imagine your workspace isn't properly setup. I'd encourage reaching out to the support team or your account team, it's likely a role setup issue :/