[BUG]: Error running "migrate-tables" - failed for getting token from AzureAD - UBER PRINCIPAL

[gregorymiguel]

Is there an existing issue for this?

• I have searched the existing issues

Current Behavior

I found an issue when trying to execute the migrate-tables workflow.
The uber principal has been created and the policy was updated correctly with the uber credentials info.
All the 4 tasks returns with the same error:

Both clusters (main and migrate_tables) are using the policy created by create-uber-principal task.

Policy:

Expected Behavior

The cluster should be able to get AAD token using the uber principal credentials.

Steps To Reproduce

No response

Cloud

Azure

Operating System

macOS

Version

latest via Databricks CLI

Relevant log output

No response

[nkvuong]

@gregorymiguel this is similar to #1625 - could you also update the spn details for sql warehouse configuration in the meantime?

[gregorymiguel]

Hi @nkvuong ,

I added the spn config to the sql warehouse before opening this issue. It didn't work for me.
I don't think the issue reported in #1625 is correct. The migrate-tables task doesn't use sql warehouse compute to execute the inner tasks. The clusters related to migrate-tables are Job Clusters and assigned to ucx compute policy.
I'm debugging the issue. I hope bringing you more information soon. Wish me luck! =)

[JCZuurmond]

Hi @gregorymiguel , thank you for being elaborate in your Gihub issues :) it's much appreciated. How are you doing on this bug?

I was thinking: could it be something simple like the single curly brackets in the secret instead of double curly brackets:

{secrets/ucx/uber_principal_secret}

should be

{{secrets/ucx/uber_principal_secret}}

[JCZuurmond]

@gregorymiguel : Could you confirm that the proposed solution solves your issue? Otherwise we need to investigate the issue further

[gregorymiguel]

Hi @JCZuurmond ,
Thank you! The solution proposed worked!