"Invalid object name 'ExtendedCatalog'" after connecting a Jan 2025 PBIRS instance in scale-out

[daniel_doboseru]

Hi. After trying to scale-out our PBIRS instance, we started to see the dreaded "An error occured within the report server database. This may be due to a connection failure, timeout or low disk condition within the database."

Our setup:
1. one PBIRS instance, May 2024, with ReportServer database hosted on the same server (MS-SQL 2022 Standard)

2. installed an additional PBIRS instance on a new server, Jan 2025, without configuration

3. configured the new instance to use the ReportServer database of the first instance, and enabled scale-out

4. both instances work (paginated reports, subscriptions), except the fact that now both of them are randomly throwing errors of the mentioned type (when listing the directories, managing reports, even running etc)

Sometimes it's the one about ExtendedCatalog, sometimes it's just a generic error message with no details, and even an error regarding a stored procedure, GetUserNameBySid if I'm not mistaken - usually the exact same action works fine after refreshing the page.

My question is: it's possible that after connecting the Jan 2025 instance on the existing RS database, some objects might have been modified and because two instances with different versions (May 2024 + Jan 2025) are using it, it's in an inconsistent state for both and thus the random errors? Or there shouldn't be compatibility issues between these two versions regarding the RS database?

@Petebro would really appreciate an input from your end. Thank you!

[v-jianpeng-msft · Accepted solution]

Hi, @daniel_doboseru 

All nodes in a scale-out deployment must use the same version of Reporting Services, because a database schema upgrade is automatically triggered when a new version instance connects to the old database, and the old version instance is not compatible with the upgraded schema.
Priority: Unify PBIRS versions and rebuild scale-out clusters
Downgrade or upgrade to the unified version
If you can't roll back a database upgrade for a 2025.01 instance:
1. Upgrade the 2024.05 instance to the 2025.01 version
2. Reconfigure the scale-out deployment on the new server
3. Migrate the original database to a new cluster
If you can find the backup before the database upgrade:
1. Roll back the 'ReportServer' database to its pre-upgrade state
2. Uninstall the 2025.01 instance and reinstall the 2024.05 version
3. Reconfigure scale-out

Best Regards

Jianpeng Li

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

[daniel_doboseru]

Hello. Thank you for your answer. We will indeed prioritize unifying the two scale-out nodes to the newer version.