Forum Discussion
Subject: Dataflow Gen2 refresh fails with ModelBuilderEvaluationCancellationResult
Hi all,
I'm hitting a persistent refresh failure on a Dataflow Gen2 query and I'm trying to narrow down the root cause. Hoping someone recognizes this pattern.
Error: "A problem occurred while refreshing the dataflow: {errorMessage}. Error code: ModelBuilderEvaluationCancellationResult.").
Note that {errorMessage} is showing up literally as an unresolved placeholder, not an actual error description — the request ID is also all zeros. This suggests the mashup evaluation is being killed/cancelled rather than failing with a normal, descriptive error.
What I've already ruled out:
- Capacity throttling/overage — I checked the Fabric Capacity Metrics app and found a large cumulative overage (~20,000% carryforward) building up over several days, with active "Add %" spikes still occurring. I paused and resumed the capacity to reset the metering/smoothing state entirely. The refresh still fails with the exact same error immediately after.
- Data volume — I reduced the query's scope significantly (from a full year of months down to 3 months, cutting the row count in the heaviest cross-join/aggregation steps by roughly 4x). The refresh still fails identically.
So: reduced data volume + a freshly reset capacity all produce the exact same cancellation error. This makes me think the issue isn't about memory/timeout pressure from this specific query's complexity, but something else.
Questions:
- Has anyone seen ModelBuilderEvaluationCancellationResult occur with an empty {errorMessage} too?
- Is there a way to get more detail/logging on what's actually being cancelled here, beyond what's shown in the refresh history UI?
Any pointers appreciated.
Thanks and regards, Wesley.
Hi,
Thanks for pointing me in this direction. Unfortunately, I wasn't able to find any useful information there that helped identify the cause of the issue.
As a temporary workaround, I duplicated the original dataflow in its original state, and that version appears to be working correctly. What's strange is that it still contains the joins and more complex calculations, yet it doesn't fail immediately after publication like the modified version does.
This makes it even more difficult to determine what is actually causing the issue.
Thanks again.
8 Replies
Hi Wesley0107 ,
Thanks for reaching out to fabric community.
Based on the information you have shared, you have already ruled out two of the most common causes capacity pressure and data volume. Since the refresh still fails with the same ModelBuilderEvaluationCancellationResult after reducing the dataset and resetting the capacity, the issue is unlikely to be caused by resource exhaustion alone.
As a next step, we recommend downloading the detailed refresh logs for the Dataflow Gen2 refresh:
1. Open the Dataflow Gen2 item and navigate to Recent runs.
2. Select the failed refresh.
3. Choose Download detailed logs.The downloaded logs contain the mashup engine diagnostics, which provide more detailed information than the refresh history and can help identify the operation that was cancelled.
You can also review the refresh execution in the Monitoring Hub to determine whether a specific activity or transformation is consistently failing.
If the detailed logs still do not reveal the underlying exception and only show ModelBuilderEvaluationCancellationResult, we recommend opening a Microsoft Support ticket( https://learn.microsoft.com/en-us/power-bi/support/create-support-ticket )
Reference documentation:
Dataflow Gen2 monitoring and detailed refresh logs: https://learn.microsoft.com/fabric/data-factory/dataflows-gen2-monitor
Dataflow Gen2 refresh: https://learn.microsoft.com/fabric/data-factory/dataflow-gen2-refreshThanks!!
Hello,
Thanks for your response, I really appreciate it.
I've downloaded the detailed logs, which appear to be in Excel format? However, they don't seem to provide any additional information beyond what was already shown in the original error message.
Regarding the Monitoring Hub - is that the same as the Admin monitoring workspace?
Thanks a lot.
Hi Wesley0107 ,
Monitoring hub is a Fabric experience that allows you to monitor and track the status of Fabric activities, including Dataflow Gen2 refreshes. Below shows the monitoring hub details.
https://learn.microsoft.com/en-us/fabric/admin/monitoring-hub
Please raise a support ticket ( https://learn.microsoft.com/en-us/power-bi/support/create-support-ticket ) with microsoft so that they will troubleshoot the issue and fix it.
Thanks!!
Hi Wesley,
Based on what you've already tested, you've done a good job ruling out the two most common causes (capacity pressure and excessive data volume).
Since the refresh still fails after reducing the dataset and resetting the capacity, the issue may be related to the Dataflow Gen2 execution engine rather than the query itself. A few things I'd recommend checking:
Review the Monitor hub and the Dataflow refresh history for any additional activity details or correlated failures.
Try isolating the query by disabling query folding-intensive or complex transformation steps (joins, custom functions, cross joins, etc.) to identify whether a specific step is triggering the cancellation.
If the data source is external, verify there are no authentication, gateway, or transient connectivity issues that could be causing the evaluation to terminate unexpectedly.
If possible, create a minimal copy of the dataflow containing only the failing query to determine whether the behavior is specific to the dataflow or a broader service issue.
The fact that the refresh history displays a literal {errorMessage} placeholder and an all-zero request ID suggests the operation may be terminating before a proper exception is generated, which could indicate a service-side issue.
If the problem persists after isolating the query, I'd recommend opening a Microsoft Support ticket and providing the Activity ID, workspace ID, capacity details, and the exact refresh timestamp. This will allow the product team to inspect backend logs that aren't exposed in the UI.
It would also be helpful if you could share:
What is the data source (SQL, Lakehouse, SharePoint, etc.)?
Does the failure occur during evaluation or after data starts loading?
Is this a newly introduced issue, or did the same dataflow refresh successfully in the past?
If you identify the root cause or receive guidance from Microsoft Support, please consider sharing the resolution here—it would be valuable for others encountering the same ModelBuilderEvaluationCancellationResult error.
If you found this reply helpful, please consider giving it a Kudos. If it helps resolve the issue, marking it as the Accepted Solution will make it easier for others to find.
Hi Prince,
Thanks for your reply, it is much appreciated.
I've downloaded the detailed logs, which appear to be in Excel format? However, they don't seem to provide any additional information beyond what was already shown in the original error message.
Regarding the Monitoring Hub - is that the same as the Admin monitoring workspace?
I've also tried reducing the complexity of the queries by removing joins and complex calculations. In addition, I've split the dataflow into multiple queries to reduce the overall workload. Unfortunately, the issue still persists: the dataflow fails immediately with the same error message.
It may also be helpful to know that this issue did not occur previously. It only started after I made a few changes to the M code. However, I haven't been able to identify which specific change may have triggered the problem.
The data source is indeed external. However, this issue only occurs in my Gold dataflow. The Bronze and Silver dataflows refresh successfully without any authentication issues, so the connection itself appears to be working correctly.
Do you have any other suggestions based on this?
Thanks and regards,
Wesley
Hi Wesley0107 ,
Could you review the suggestion provided above and let us know if you have any additional questions, we are happy to address.
Thanks!!
