Cloudera Docs

Migrating AMP based RAG Studio deployment to Runtime-based RAG Studio deployment

Follow the steps to manually migrate your current Accelerators for Machine Learning Projects (AMP) based RAG Studio deployment to the updated Runtime-based RAG Studio.

As no automated migration tool is currently available, the steps are required to successfully preserve and transfer your existing Knowledge Bases and Chat History.

Before starting the migration, ensure that the following conditions are met:

  1. An active AMP deployment is required that contains the Knowledge Bases and Chat History intended for migration.
  2. A newly deployed Runtime mode environment is required. This environment must be empty and contain no existing data.
  3. Ensure that the LLM providers and API keys in your Runtime mode deployment are configured to match the settings previously used in your AMP environment.
  4. Prior to starting the file transfer, stop the Runtime mode application. Do not delete the application; only stop the service.
  1. In the Cloudera console, click the Cloudera AI tile. The Cloudera AI Home page displays.
  2. Select the Project option in the left navigation pane if you want to open up an already existing project. Alternatively, you can create a new project, by launching RAG Studio.
  3. Select the required Project. The Overview option is displayed in the left navigation pane.
  4. Click the Overview option. The RAG Studio - Project window displays.
  5. Click the rag-studio file. The Files window displays.
  6. Select the rag-studio/databases file checkbox and click Download. The file is downloaded as a zip file.
  7. Create a root directory, in your local machine, for example, RagStudioMigration.
  8. Extract the downloaded zip file content into this root directory.
  9. Organise the extracted content into subfolders, for example: RagStudioMigration/rag-studio/databases/.
  10. Open the Overview of the same project, selected in Step 3. The RAG Studio - Project window displays.
  11. Select the rag-studio/databases file checkbox and delete the existing rag-studio file in the root directory of the project.
  12. Click the Upload button, to upload the rag-studio folder from your local RagStudioMigration directory into the project root.

    All the files from the RagStudioMigration directory on your local machine, are successfully uploaded into the project root.

  13. Click the Applications option, in the left navigation pane
  14. Click the option in the RagStudio tile and select Restart as to restart the RagStudio application.
  15. Click the RagStudio application tile. The RAG Studio window is displayed.
  16. Verify that the Knowledge Bases appear in the list and that your Chat History has been repopulated.

Post-Migration Troubleshooting:

When using Amazon Bedrock, you might encounter issues where LLMs, Embedding Models, or Reranking models appear with incorrect IDs. This can be caused by Bedrock renaming its IDs based on the Inference Profiles.

Follow the steps, if a model is reported as missing or invalid:

  1. Manually reselect the correct model or inference profile.
  2. Navigate to the relevant settings, such as chat settings or index settings, and choose the correct option from the available dropdown menus.