Forum Discussion
One Direct Lake (OneLake) table is empty despite data in warehouse
Hi all,
I created a direct lake (OneLake, not via SQL endpoint) semantic model that is based on a few warehouse tables. Everything worked fine and the report showed data until yesterday. All tables have data in them in the warehouse (verified with SQL) but one table (the fact table with less than 5k rows) does not have any data in the semantic model. I used the following dax expression and visuals to confirm that the fact table is empty:
How can I solve this issue?
New data was inserted with a SELECT from a view and there was an error that was introduced in the view. After fixing the view and truncating/inserting new data into the table, Power BI showed data. The other issue was that a MERGE statement was used which isn't supported in fabric. Either one of those two things must have caused the issues. It is now resolved, however.
7 Replies
Impressive
rohit1991Super User
Hi Anonymous
This can happen if the table is not picked up properly by Direct Lake, even though you see rows in the warehouse. A few things you can try:
-
Make sure the table is in a supported format (Direct Lake works best with Delta tables).
-
Remove the table from the model and add it back again. Sometimes it doesn’t load the first time.
-
Try creating a small test table in the same warehouse - if that loads, then the issue is with the fact table itself.
-
Since you already checked filters, relationships, and RLS, it’s not a modeling problem.
If it still stays empty, it’s most likely a backend limitation/bug in Fabric Direct Lake. In that case, raising a support ticket with Microsoft would be the next step.
-
- Shahid12523
Community Champion
- Check and convert unsupported types; rename columns to simple names.
- Add a unique ID column if missing.
- Refresh metadata in Desktop (remove & re-add table).
- Ensure OneLake storage permissions are correct.
- If needed, create a SQL view with clean columns and use that.
Hi Shahid12523
- data types are supported and there is no mismatch between warehouse and semantic model
- regarding unique column ID, renaming to simple names and creating a SQL View: up until yesterday, the report worked fine even without a unique column ID and the current names
- refreshing metadata doesn't change anything
- onelake storage permissions are correct because i can see the data when i use SELECT statements in the SQL endpoint
- creating a sql view wouldn't work because they need to be materialized for direct lake and I believe it's a preview feature for warehouses
Hi Anonymous ,
As none of the troubleshooting steps seemed to work, please consider reaching out to Microsoft Support. You can provide them with all the troubleshooting steps you've already taken, which will help them understand the issue better and provide a resolution. They might be able to identify something specific about your admin account setup or provide a solution that isn't immediately obvious.
Below is the link to create Microsoft Support ticket:
How to create a Fabric and Power BI Support ticket - Power BI | Microsoft Learn
Thank you
Hi rohit1991,
it's in delta format
tried removing and adding the fact table. No change
created small table in the warehouse and the data shows in the semantic model.
New data was inserted with a SELECT from a view and there was an error that was introduced in the view. After fixing the view and truncating/inserting new data into the table, Power BI showed data. The other issue was that a MERGE statement was used which isn't supported in fabric. Either one of those two things must have caused the issues. It is now resolved, however.
