What is expected behavior for reports after minor schema changes?

[ddelapasse]

I have a schema that is expected to have small changes (ie just column name changes - no types) periodically.  I am using the api to update the schema definition when this occurs and that is successful.  I thought that as long as I kept the lineage tag the same the report would be fine - but my test report says:  "Something's wrong with one or more fields... and lists the OLD name of my schema change as well as ALL the other columns in the table.  My only choice is "fix the report" which removes ALL the data from the table and wants me to rebuild it.

Isn't this the point of lineage tag?  What can I do to avoid ruining all existing reports that reference the column?

Just wanted to add that we are embedding PowerBI in a webapp so we are not using the desktop app.  Hopefully there's a way to make this work without the desktop app having to republish.  Do I need to re-bind the report to the model via the api somehow perhaps?

[v-veshwara-msft]

Hi @ddelapasse ,
We haven't heard from you since last response and just wanted to check whether the solutions provided met your needs. If yes, please accept as solution to help others benefit. If not, please reach out.
Thank you.

[v-veshwara-msft]

Hi @ddelapasse ,

Thank you for reaching out. Based on your description, renaming a column is treated by Power BI as a column removal and the addition of a new one, which can break visuals referencing the old column name. This is likely causing the error in your reports.

Lineage tags help track relationships between data sources and reports, but they do not automatically update renamed columns in reports. While lineage tags ensure relationships between objects remain intact, Power BI reports continue to reference the original column names. This can cause errors if those columns are renamed in the dataset.

As mentioned by @lbendlin , renaming a column is seen as removing the old column and adding a new one, triggering a metadata re-evaluation.
As suggested by @Poojara_D12 , lineage tags do not handle column renames automatically, and rebinding via the API is needed to link the report to the updated dataset.

Possible solutions:

1. Alias the Old Column Name: Instead of renaming, keep the old column name and add the new one as an alias. This helps your reports continue working without any disruption while you transition.
2. Rebind the Report via API: Since you're embedding Power BI in a web app, reports do not automatically update. As suggested by @Poojara_D12 , using the Rebind Report API will allow you to link the report to the updated dataset. Rebind Report - Power BI REST API.
3. Manually Update the Report: If needed, you can manually update visuals in Power BI Desktop and republish the report with the new column names.
   
   

Hope this helps. Please reach out for further assistance.

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly and a kudos would be appreciated.
Best regards,
Vinay.

[Poojara_D12]

Hi @ddelapasse 

When you update a schema by renaming columns, lineage tags may not always automatically map the new column names to existing reports, causing the error where old column names still appear. The lineage tag helps track relationships, but Power BI may require manual re-binding to ensure that reports are linked to the updated model. In embedded Power BI, the report is not automatically re-bound to the new data source after schema changes, and the API can be used to programmatically re-bind the report to the updated model. To avoid report breakage, you could alias old columns temporarily, use parameterized columns, or update visuals automatically. Ensuring reports remain intact requires re-binding or careful version management when changing schema elements like column names.

 

[lbendlin]

There's no column name change. What you have is the removal of a column and the addition of another column.  This sends Power Query in a tizzy, always.  (especially the removal).  It now panics and wants to absolutely re-read ALL the meta data - resulting in the dreaded "Evaluating..."  phase (that REALLY should be a t-shirt slogan!).