Hello Tableau Community,
I'm looking for advice on a Hive migration that touches a large number of published workbooks on Tableau Server. I've tested several approaches and found a limitation I'd like to confirm or get around.
ENVIRONMENT
- Tableau Server 2025.1.11 on Linux (Oracle Enterprise Linux 8.10) migrating to AWS EC2 Linux
- Data source type: Cloudera Hadoop / Hive (class='hadoophive'), authenticated with Kerberos (Kerberos delegation, embedded credentials in some connections)
- Content: about 1,500 workbooks and 100+ data sources; many workbooks are extract-based and some extracts are 50-100 GB
- 10 TB overall data
GOAL
The Hive source is moving to a new server in a different Kerberos realm. Every connection needs to point to the new Hive host and use the new Kerberos details:
- Server / port: old host -> <new-hive-host>:10000
- Kerberos realm: <OLD_REALM> -> <NEW_REALM>
- Kerberos host FQDN: <old-hive-host-fqdn> -> <new-hive-host-fqdn>
- Kerberos service name: Hive -> hive
I'd like to do this in bulk and, if possible, without downloading and republishing the full-size workbooks, because of the extract sizes and the time that would take.
Any experience with a similar migration, or pointers to documentation, would be very much appreciated.
Best Regards,
Rami Reddy Jaggavarapu,
Tableau Server administrator.
#Tableau Server
to keep it short, the connection details and authentication type information you are targeting are existing in one place only: in the xml of the content, meaning tds/tdsx for data sources, and this xml is stored in the Tableau Server postgres internal repository as blob.
So I would personally download the data source without extract using the REST API (without extract), manipulate the xml to match what you want and republish the datasource.
I'm not aware of any REST API endpoint that would allow to change all of that, especially with Kerberos / viewer credentials delegation type.
