Error During Backward Implementation in Power BI Deploy Pipeline

[igor_r_arangues]

I'm currently trying to introduce the use of Power BI Deploy Pipelines in my company. At the moment, we only have a Production Workspace, and my goal is to reconstruct the pipeline backwards, by copying existing reports, semantic models, and dataflows from Prod to Test and Dev workspaces.

We have around 220 items (including 6 dataflows and 107 reports/semantic models). Every time I attempt this backward implementation, the process runs for about 2 hours and 10 minutes, successfully deploying all dataflows and almost all semantic models — but it always fails before reaching the report deployment stage.

As a result, no reports are ever copied to the previous stages, and I have to manually delete the partially deployed items before trying again.

At this point, I’m not sure what else to try.

• Has anyone experienced something similar?

• Are there known limitations or best practices when doing this kind of reverse pipeline setup?

• Should I avoid backward implementation and start our use with Dev and Test empty?

Any advice would be appreciated!

[rohit1991]

Hi @vijaymuthukrishna,

Power BI Deploy Pipelines don’t support backward deployment (e.g. Test → Dev) well. The failures you're seeing are mostly due to item bindings and lineage issues across workspaces.

I would say safe way is to manually download the .pbix from Test or Prod and republish it to Dev. It’s not ideal, but it avoids the broken lineage errors. Also, for future rollbacks, it helps to keep versioned copies in OneDrive or source control.

[shashiPaul1570_ · Accepted solution]

Hi @igor_r_arangues 

You're encountering a known challenge when attempting backward deployment in Power BI deployment pipelines, especially with a large number of artifacts.

Why This Happens

Power BI's deployment pipeline UI requires copying all artifacts at once when deploying backward (e.g., from Production to Test or Dev). With ~220 items, you're likely hitting an internal timeout or resource limit — especially since the process stops just before reaching the report deployment stage.

Even though Power BI doesn’t always surface a clear error, long-running backward deployments (typically over 100–150 items) often fail silently due to

• Query timeout limits

• Memory/resource constraints on artifact processing

• Incomplete dependency bindings between datasets and reports

Best Methods to overcome this challenge

Avoid Full Backward Deployment via UI
Instead of copying all items at once:

• Start with empty Dev and Test workspaces

• Manually export/import PBIX files, dataflows, and datasets in manageable batches

• Once aligned, connect all three stages to a pipeline and move forward from Dev → Test → Prod

OR

Use REST API for Phased Backward Deployment

Please find these links to get this done easily. 

https://learn.microsoft.com/en-us/fabric/cicd/deployment-pipelines/pipeline-automation

Also there is similar kind of solution over fabric community

https://community.fabric.microsoft.com/t5/Service/Backwards-deployment-through-deployment-pipeline/t...

Please give this a try and let me know if it helps
If you find this post helpful, please mark it as a solution so others can benefit as well.

Thanks,
Shashi Paul