This is best Fabric, Power BI, SQL and AI community event. How do we know? The last event sold out! Save €200 with code FABCMTY200.
Register nowA new Data Days event is coming soon! This time we’re going bigger than ever. Fabric, Power BI, SQL, AI and more. Don't miss out.
I have 2 semantic models and one is pulling some data from the other including just the refresh time.
It seemed ot be working fine just pulling in a table of data, but I just added the refresh time and getting this error just where the refresh date is displayed? I'm not getting any error for the other data being pulled from the other model
Underlying Error: OnPremiseServiceException
Activity ID: 07d03594-b96f-48b8-b397-20f2d2940ad8
Correlation ID: 3c53e663-b2e4-075a-2f06-2923fd3d9171
Request ID: 3c7ab2e8-b6bf-44c1-831b-03490299a92e
Time: Fri May 29 2026 10:12:21 GMT+0100 (British Summer Time)
Service version: 13.0.28503.502
Client version: 2605.4.29440-train
Cluster URI: https://wabi-uk-south-b-primary-redirect.analysis.windows.net/
Any clues what this means? the pbix works fine no issues.
Hi @masplin ,
We’d like to follow up regarding the recent concern. Kindly confirm whether the issue has been resolved, or if further assistance is still required. We are available to support you and are committed to helping you reach a resolution.
Thank you.
So i really cant understand what is going on here
I have 2 tables coming from Model 1; Client Bookings and the Refresh date. Both tables are connected by Direct query to Model 2
If I create a measure e.g. MAX using this refresh date column in model 2 it causes this error
If I add a column in Model 1 on the Client Booking table that just contains this date, then create a measure in model 2 e.g. MAX of this column it causes the error
If I try to add any column of data from the Model 1 [Client Booking] table to a visual in model 2 it errors
However if I add a column to my model 2 [Client] table = MAX(Refresh date) then add a measure to get the MAX of that column it works fine
So it seems you cannot use any data coming from the connected model directly (either as a row or column of a visual or in a measure). First you have to read the data into a table of the current model using a column calculation then you can use it. This seems really odd behaviour.
Am I doing something wrong with having direct query conenctions ottthe other model or something in the permissioning
So it seems that you cant
Hi @masplin
Can you try republishing both models, refreshing credentials or gateway mappings, and replacing refresh time logic with a calculated table or measure instead of metadata / DMV based queries. Also test with simple measure from the source model to confirm cross model connection itself is working
Hi @masplin
I think this OnPremiseServiceException error happens because your second semantic model is configured with a local timezone format (like British Summer Time) that conflicts with the strict cloud format requirements when passed between chained datasets. When you pull a dynamic refresh timestamp locally in Power BI Desktop, the engine handles the data type flexibly, but once published, the Power BI Service requires datetime strings coming from a cross-dataset query to be explicitly locked down to an absolute ISO standard or UTC format to prevent the backend gateway layer from crashing.
To fix this, go back into your first semantic model's Power Query editor where the refresh time is generated, and explicitly convert that timestamp column to text or force it to UTC using DateTimeZone.ToUtc(DateTimeZone.LocalNow()). If you convert it to a plain text string before passing it to the second model, it completely bypasses the cloud timezone validation check, which will resolve the error instantly on the Service without breaking any of your other cross-model table links.
Well that didnt work. I created a seperate table so I could format the date as text as you suggested. I'm actually picking up the modified date of a set of excel spreadhseets.
so in my source model i just have
let
Source = SharePoint.Files("https://consolxxxxthcareltd.sharepoint.com/sites/PowerBIReporting/",[ApiVersion=15]),
#"Filtered Rows" = Table.SelectRows(Source, each ([Folder Path] = "https://consolidatedhealthcareltd.sharepoint.com/sites/PowerBIReporting/Shared Documents/Trinity New Model/Date Exports/Transaction Data/")),
#"Sorted Rows" = Table.Sort(#"Filtered Rows",{{"Date modified", Order.Descending}}),
#"Kept First Rows" = Table.FirstN(#"Sorted Rows",1),
#"Removed Other Columns" = Table.SelectColumns(#"Kept First Rows",{"Date modified"}),
#"Changed Type2" = Table.TransformColumnTypes(#"Removed Other Columns",{{"Date modified", type datetime}}),
#"Changed Type3" = Table.TransformColumnTypes(#"Changed Type2",{{"Date modified", type text}})
in
#"Changed Type3"this which sorts the files to get the latest one then converts the modified date to text.
I deleted the previous table from my second model and just connected to this new table. I used VALUES to turn the column data into a measure so i could add it to the multi card.
Exactly same error. Also there is a pink ribbon saying "an error occured in Power BI Preium backend" but we dont have power bi premium?
Everything else seems to be fine and think I am connected to the other table coming from the first model no problem. Its just this date that is the issue.
Wierdly I got a blank pbix and connected to the first model. No problem adding the date as a column or a measure. no errors. So the issue is some conflict between the data in the second model and having connection to the first model. As I say I have an existing connection to the first model pulling in a whole table of data for comparison and that seems to be fine.
Really stumped
Hi @masplin,
at least the error message states something regarding "onpremiseservice". I would check the following:
Hope this helps to troubleshoot.
Best regards!
Check out the May 2026 Power BI update to learn about new features.
Sign up to receive a private message when registration opens and key events begin.
If you have recently started exploring Fabric, we'd love to hear how it's going. Your feedback can help with product improvements.
| User | Count |
|---|---|
| 10 | |
| 9 | |
| 9 | |
| 7 | |
| 7 |
| User | Count |
|---|---|
| 22 | |
| 21 | |
| 19 | |
| 17 | |
| 15 |