Forum Discussion
Dataflow publish failed - Something went wrong (gen2)
Power BI gen2 dataflows can be so frustrating. Here is the entire error message presented in Power BI:
"Something went wrong"
Here are the full problem details, after expanding:
How can one argue that something went wrong? The main thing that went wrong is when we started using GEN2 dataflows for this ETL, instead of a more reliable tool.
I'm guessing that the next step is to enable detailed logging in the gateway and find out what is failing? I'm guessing the bugs are not in our own custom code, but on the Microsoft side. They aren't likely to tell me why their stuff is falling over, unfortunately. (Otherwise they would have shared those details in this error message.)
How do others troubleshoot this sort of thing? Retries?
This error happens during the publish, so am I expected to publish over and over until it succeeds?
EDIT: I enabled the detailed logging in the gateway, and the mashup log appears to be showing a recurring attempt to reach the data source every ten minutes. I'm guessing that there is some sort of timeout error, and a retry loop managed in the mashup. I think the gateway is flipping out and failing after the retries expire in the mashup.
Hi dbeavon3,
We appreciate your inquiry on the Microsoft Fabric Community Forum.
Thank you for reaching out and providing a detailed explanation of the issue you are experiencing with Power BI Gen2 dataflows. We understand how challenging it can be to encounter error messages such as "Something went wrong," especially when relying on these tools for critical ETL processes.
To address this issue, we recommend the following steps as a potential workaround:
- Retry the publish operation after a short interval to check if the issue resolves on its own. This approach can sometimes help overcome temporary service-related disruptions.
- Break down the dataflow into smaller segments to determine whether this reduces the likelihood of timeouts or failures.
If you find our response helpful, we kindly request you to mark it as the accepted solution and provide kudos. Your feedback will assist other community members facing similar challenges.
Thank you.
6 Replies
I think the failures are on the Microsoft side . In the past whenever the error messages are so vague it is intentional (ie. Power BI doesn't want to surface the underlying cause of these failures, since someone might ask them to actually fix the underlying root cause).
I've found that even when looking at the gateway logs ("mashup") there is no failure in on the PQ side. The custom mashup logs have the same appearance for a failed publish operation as a successful one. I believe the bugs are not related to the mashup, but are more often related to the spooling of the data out to the service (ie. the Microsoft side of things). Those are the types of errors which aren't presented in a transparent way to the customer, for various reasons.
Here are three other examples of the same meaningless error message, and some indicate that the problem is intermittent (ie. almost certainly a Microsoft bug).
https://community.fabric.microsoft.com/t5/Service/Fabric-Issues-Refreshing-DataFlow-Gen2/m-p/3846244
https://community.fabric.microsoft.com/t5/Dataflow/Data-flow-gen-2-not-publishing/m-p/3940882
Is anyone aware of a service outage or service degradation at the moment?
I enabled the detailed logging in the gateway, and the mashup log appears to be showing a recurring attempt to reach the data source every ten minutes. I'm guessing that there is some sort of timeout error, and a retry loop managed in the mashup. I think the gateway is flipping out and failing after the retries expire in the mashup.
Given these obvious problems happening in the mashup, you would think the "publish" of the dataflow would have a bit more to say than "something went wrong".
... more to follow...Hi dbeavon3,
We appreciate your inquiry on the Microsoft Fabric Community Forum.
Thank you for reaching out and providing a detailed explanation of the issue you are experiencing with Power BI Gen2 dataflows. We understand how challenging it can be to encounter error messages such as "Something went wrong," especially when relying on these tools for critical ETL processes.
To address this issue, we recommend the following steps as a potential workaround:
- Retry the publish operation after a short interval to check if the issue resolves on its own. This approach can sometimes help overcome temporary service-related disruptions.
- Break down the dataflow into smaller segments to determine whether this reduces the likelihood of timeouts or failures.
If you find our response helpful, we kindly request you to mark it as the accepted solution and provide kudos. Your feedback will assist other community members facing similar challenges.
Thank you.
Hi dbeavon3,
We are following up to see if your query has been resolved. Should you have identified a solution, we kindly request you to share it with the community to assist others facing similar issues.
If our response was helpful, please mark it as the accepted solution and provide kudos, as this helps the broader community.
Thank you.Hi dbeavon3,
We are following up to check if your query has been resolved. If you have found a solution, we would appreciate it if you could share it with the community to assist others encountering similar challenges.
If our response was helpful, kindly mark it as the accepted solution and give kudos, as this will benefit the wider community.
Thank you.
Hi dbeavon3,
We are following up to check if your query has been resolved. If you have found a solution, we would appreciate it if you could share it with the community to assist others encountering similar challenges.
If our response was helpful, kindly mark it as the accepted solution and give kudos, as this will benefit the wider community.
Thank you.