Advance your Data & AI career with 50 days of live learning, dataviz contests, hands-on challenges, study groups & certifications and more!
Get registeredGet Fabric Certified for FREE during Fabric Data Days. Don't miss your chance! Request now
Context
We currently have an issue with credentials in dataflows after they're taken over (usually after deployment via pipeline), so we wanted an automated check on that before our weekly refresh.
Our Solution
Since we cannot do it using REST API (dataflows API won't allow testing the connection on data sources) I thought of triggering a refresh and cancelling it a couple mins later (we have one dataflow per table, so it would fail immediately). We need to cancel them because tables are very large, so it'd impact our backend.
Problem
I can trigger the dataflow refresh via API, but cannot cancel it via API. It says "No refresh in progress".
I have the same issue even if I use the MS Learn editor. I can see the transaction in progress if I call Get Dataflow Transactions (API or browser). The scopes required for Cancel Transaction are the same as Refresh Dataflow, so I assume I have the proper permissions. Has anyone had this issue before?
We have no problems issuing cancellation requests for dataflows in flight.* We do this continuously on our tenant because we are limiting dataflows to the same 5 hours as the semantic model refresh limit.
* Remember that this is only a request. If and when it gets honored is rather random, and depends on where in the process the dataflow is. We have also anecdotal evidence that cancellations are not to be confused with rollbacks.
Hi @endel
Thanks for reaching out to Microsoft Fabric Community Forum.
Based on the above error the issue with canceling a dataflow refresh via the REST API after successfully triggering it. You're already able to trigger the refresh, but when you try to cancel it, you're receiving a No refresh in progress.
Here are some workaround which may help you.
Hope this helps !!
Thank You.
Actually, I'm triggering the dataflow refresh, then wait a few seconds and request the refresh history to get the ID of the one in progress, so I believe this discards the hypothesis of being out of sync.
Hi @endel
Thanks for confirming your setup. Since you’re already triggering the refresh and retrieving the history immediately after, the process should typically stay in sync.
If you’re still observing mismatched or missing refresh IDs despite this, it might require a deeper review from the Power BI service side. I’d recommend opening a Microsoft Fabric Support and Status | Microsoft Fabric.
Advance your Data & AI career with 50 days of live learning, contests, hands-on challenges, study groups & certifications and more!
Check out the October 2025 Power BI update to learn about new features.