Direct lake Semantic model changed, but recognized as a new model in the deployment pipeline

[Tiger_an]

Hi all,

Recently, I changed my direct lake semantic model, but when I try to deploy it from my test workspace to production workspace, it is recognized as a new model instead of different. 

I use tabular editor to check the metadata for the two models, I see some difference in the annotation part:

model in test:

model in production:

I am not sure if this is the root cause, if yes, how can I fix this?

Any suggestion will be greatly appreciated!

[v-kpoloju-msft]

Hi @Tiger_an,

Thank you for reaching out to the Microsoft fabric community forum. Thank you @Akash_Varuna, for your inputs on this thread. After thoroughly reviewing the details you provided, I have identified few workarounds that may help resolve the issue. Please follow these steps:

The annotation "PBIService_Versioning_SaveVersion" in your production model indicates that the model was saved or modified in Power BI Service, which may have changed its internal metadata. This could be causing inconsistencies when comparing the test and production models during deployment.

For deployment pipelines to work smoothly, the metadata of the test and production models must match. If there are differences, Fabric might not recognize them as the same model. This can result in the test model being treated as a completely new model instead of an update, leading to deployment issues.

To fix this, try opening the production model in Power BI Desktop, saving it using the same version as the test model, and then re-uploading it to the production workspace. If the issue persists, use Tabular Editor to compare annotations between the models. Look for any discrepancies especially in entries like "PBIService_Versioning_SaveVersion" and remove or adjust them as needed before saving and redeploying.

Additionally, check the Model ID, Annotations, and any unique identifiers in Tabular Editor to ensure they match between the test and production models. If there are unintended differences, consider reverting the metadata to align with the production model. If the model was redeployed or duplicated in the test workspace, Power BI might have assigned it a new ID. Always verify that you're working with the same dataset rather than unintentionally creating a new one.

Yes, the metadata differences, particularly the __PBIService_Versioning_SaveVersion, are likely causing the deployment pipeline to recognize the modified model as a new one. Aligning the metadata using Tabular Editor and ensuring consistency in the Power BI Desktop version across test and production environments should resolve the issue.

Kindly refer to the below following links for more information:

Manage Direct Lake semantic models - Microsoft Fabric | Microsoft Learn
Develop Direct Lake semantic models - Microsoft Fabric | Microsoft Learn

If this post helps, then please give us ‘Kudos’ and consider Accept it as a solution to help the other members find it more quickly.

Thank you.

[v-kpoloju-msft]

Hi @Tiger_an,

May I ask if you have resolved this issue? If so, please mark the helpful reply and accept it as the solution. This will be helpful for other community members who have similar problems to solve it faster.

Thank you.

[Tiger_an]

Hi @v-kpoloju-msft  @Akash_Varuna Thanks for your help.

I tried to update the meta date with tabular editor to my test dataset, now I see the annotations are exactly the same. I also compared other items in the metadata tab, all the same.

But I still can't link them in the pipeline. The test version still recoganized as a new one.

Any other information will be aprrecitated.

Best Regards.

Tiger

[v-kpoloju-msft]

Hi @Tiger_an,

After reviewing the details you provided, I have identified few alternative workarounds that may help resolve the issue. Please follow these steps:

The ModelID annotation is crucial for tracking dataset versions. If the ModelID differs between the test and production versions, Power BI may consider them as distinct datasets. Please ensure the ModelID is identical in both versions.

Even if the metadata in Tabular Editor matches, the dataset may have a different internal Dataset ID in the Power BI Service. To confirm, go to the Power BI Service, navigate to your workspace, and compare the dataset GUID/ID under Settings > Dataset.

If any modifications were made to table structures, relationships, or calculated columns, Power BI might recognize it as a distinct model. Instead of manually updating metadata, consider deploying the dataset using Tabular Editor’s "Save to Power BI Service" feature. Make sure to select the correct dataset during deployment.

Check if the deployment pipeline settings allow dataset updates rather than creating a new dataset. If other methods are unsuccessful, you may use Power BI’s Update Data source API to directly push the metadata update.

Kindly refer to the below following link for more information:

The Microsoft Fabric deployment pipelines process - Microsoft Fabric | Microsoft Learn

I hope this could resolve your issue, if you need any further assistance, feel free to reach out.
If this post helps, then please give us ‘Kudos’ and consider Accept it as a solution to help the other members find it more quickly.

Thank you for using Microsoft Community Forum.

[Tiger_an]

Hi @v-kpoloju-msft ,

Thanks for your response.

1. I can't see the model ID anotation in the master data tab with tabular editor, do you have a screenshot to show where I can find it?

2. Regarding "Even if the metadata in Tabular Editor matches, the dataset may have a different internal Dataset ID in the Power BI Service. To confirm, go to the Power BI Service, navigate to your workspace, and compare the dataset GUID/ID under Settings > Dataset." Isn't it every dataset has its own GUID? 

3. Regarding "If any modifications were made to table structures, relationships, or calculated columns, Power BI might recognize it as a distinct model." But these opreations are normall actions we will do to change the current model, right? Why it will be recognized as a new version?

Thanks.

Best Regards.

Tiger

[v-kpoloju-msft]

Hi @Tiger_an,

I apologize for any inconvenience caused. After reviewing the details you provided, I have outlined the following steps. Please follow the steps below:

Tabular Editor does not display the Model ID (Dataset GUID) directly because this identifier is assigned by the Power BI Service and is not stored within the Tabular Object Model (TOM) metadata that Tabular Editor accesses. To locate the Model ID (Dataset ID): Access Power BI Service, navigate to your workspace. Select the dataset and click on Settings. In the URL in your browser, the dataset ID (GUID) will be part of the URL.

Yes, each dataset in Power BI Service possesses a distinct GUID (Globally Unique Identifier). The complication arises when a dataset is altered or republished. In these instances, even if the metadata (including dataset name, tables, and columns) appears identical in tools like Tabular Editor, the internal Dataset ID (GUID) in Power BI Service may change.

Yes, modifying tables, relationships, or calculated columns is expected, but Power BI may recognize it as a new model if the dataset is republished incorrectly, structural changes alter the internal model ID, or XMLA tools track it differently. To prevent this, always publish updates to the same workspace and dataset, and verify the dataset ID in Power BI Service before and after changes.

If the issue persists, please consider raising a Microsoft support ticket. You can create one using the following link:
https://learn.microsoft.com/en-us/power-bi/support/create-support-ticket

If this post helps, then please give us ‘Kudos’ and consider Accept it as a solution to help the other members find it more quickly.

Thank you for using Microsoft Community Forum.

[v-kpoloju-msft · Accepted solution]

Hi @Tiger_an,
As previously mentioned, please use the link provided in my last post to raise a support ticket.

Thank you for using Microsoft Community Forum.

[Tiger_an]

Thanks for your information, I will submit the ticket.

[Tiger_an]

@v-kpoloju-msft 

Hi kpoloju,

Thanks for your explanation, then I understand what you mentioned of the GUID at the beggining. When we re-publish the dataset to the test workspace, the guid may changed due to metadata change, that will cause this issue.

But how can we fix this now? Do you have any suggestions for this? Or we can only submit a ticket to Microsoft?

Thanks a lot for your help.

Best Regards.

Tiger

[Akash_Varuna]

Hi @Tiger_an It might be the case could you please check these and see if it resolves 

• Compare Metadata: Use Tabular Editor to check differences in annotations, IDs, and dataset names between test and production models.

• Fix Annotations: Update the test model's annotations to match the production model.

• Check Pipeline Settings: Ensure the pipeline links the test model to the correct production workspace.

• Revalidate: Validate the updated test model to ensure consistency.