Deployment Pipeline - Random Report Overwritten when I Publish new Semantic Model

[DavidKuhry]

Hello all,
I have configured a Deployment Pipeline and overall, the process seems to work fine.
I have 1 Pipeline that contains the Master Data Model, and a number of live connected reports.

When I published a new version of the Model (which has a couple of random report pages in it), instead of ending up with a new report of the same name as I would expect (which I could either delete or ignore), one of my live connected reports was over written with the pages from the Data Model Report. I had to restore it by deploying the report again from an earlier stage. Then when I download the Data Model file, the orginal report pages are gone and it has been replaced by the report. It 

It seems that the report has become binded to the Data Model file instead of being live connected to it. 

How can I fix this?

I've read that having separate Pipelines for Models and Reports is Best Practice. But surely, I should be able to do it with one, and not have it break my content.

[DavidKuhry · Accepted solution]

SOLVED: So the issue is related to work I did a week ago. All of the reports had been created with their own semantic model. The models did nothing but hold a measure or two that were made for the report, or nothing at all. I used a script a REST API tool to rebind all of the reports directly to the Master Data Model, then delete the no longer needed semantic models. I also deleted the report generated by the Data Model itself.

Because of this, it seems that when I re-published the Master Data Model, it - why I don't really know - picked one of the reports that was bound to it, and over wrote it with the report pages from the Master Data Model.

SOLUTION: I had to download each of the reports, ensure connection to correct live data model, and republished them. When I then attempted to Re-Publish the Master Data Model, it created a new Report for itself, and everything else was left alone. As I would expect.

Very strange situation. It probably totally makes sense if you better understand the underlying architecture, but I was confounded.

[v-zhengdxu-msft]

Hi @DavidKuhry 
Congratulations on solving this problem!

I've had this problem before, but I didn't think of a solution like this, thank you very much for the solution!

^{Best Regards,}

^{Zhengdong Xu}