Looker Embedded User Authentication Fails with Gemini Conversational Data Analytics

Google Cloud Build with AI Conversational Agents (Dialogflow CX)
1

we are integrating Looker with Gemini Data Analytics and running into an issue that happens only for embedded Looker users—standard Looker users work fine.
Here’s the simplified code snippet:

access_token = sdk.login_user(user_id=user_id).access_token

looker_credentials = geminidataanalytics.Credentials(
    oauth=geminidataanalytics.OAuthCredentials(
        token=geminidataanalytics.OAuthCredentials.TokenBased(
            access_token=access_token
        )
    )
)

conversation_reference = geminidataanalytics.ConversationReference(
    conversation=data_chat_client.conversation_path(
        billing_project, location, conversation_id
    ),
    data_agent_context=geminidataanalytics.DataAgentContext(
        data_agent=data_chat_client.data_agent_path(
            billing_project, location, data_agent_id
        ),
        credentials=looker_credentials
    ),
)

When I use an embedded Looker user, the chat request fails with:

NotFound: 404 SCHEMA_RESOLUTION: failed due to NOT_FOUND error
...

If I run the exact same code with a standard Looker user, it works without errors.
Could you advise on:

  1. Do we need to provide any roles or specific permission that are specific to embed users ?

Please let me know if more details (logs, existing permissions.., etc.) would help.
Thanks in advance,

2

@Vamsi_Panchada have you found a solution for this issue?

3

@Pawel_Malon - it does looks like this is the issue at google system so at a moment the feature to auth with embed user is not available. But there is a workaround which is not optimal, like you can create and delete a standard user using looker_sdk on the fly. It works but it’s optimal ..

1 Like