Problem

You define a custom tag, custom_tags.<mytag>, in your fixed cluster policy. After you update the tag value in your policy, you notice newly launched AWS instances still display the old tag value instead of the updated one.

Cause

AWS instances still displaying the old tag value are relying on existing jobs created with the previous policy configuration. 

These existing jobs do not automatically inherit updates. They continue using the old tag values until they are manually updated. 

You can confirm an AWS instance is using an existing job by checking the job’s JSON configuration for the previous custom tag. For instructions, review the “View compute configuration as a JSON file” section of the Manage compute documentation. 

Solution

Check for outdated jobs and update them in the UI. 

1. Navigate to Compute in the Databricks UI.
2. Go to the Policies tab and select the cluster policy with the updated tag value.
3. Go to the Jobs tab.
4. Look for warning icons indicating which jobs require an update. 
5. To batch update all jobs with the warning icon, go to the Compliance column and click Fix all to apply the updated policy to all affected jobs.

Alternatively, update individual jobs by selecting them one at a time.

For additional information on creating and managing cluster policies in Databricks, refer to the Create and manage compute policies documentation.