Forum Discussion
Odd Failure error which doesnt make sense
My scheduled refresh is failing with this error?
Data source error: {"error":{"code":"DM_GWPipeline_Gateway_MashupDataAccessError","pbi.error"
{"code":"DM_GWPipeline_Gateway_MashupDataAccessError","parameters":{},"details":[{"code":"DM_ErrorDetailNameCode_UnderlyingErrorCode","detail":{"type":1,"value":"-2147467259"}},{"code":"DM_ErrorDetailNameCode_UnderlyingErrorMessage","detail":{"type":1,"value":"The key didn't match any rows in the table."}},{"code":"DM_ErrorDetailNameCode_UnderlyingHResult","detail":{"type":1,"value":"-2147467259"}},{"code":"Microsoft.Data.Mashup.ErrorCode","detail":{"type":1,"value":"10061"}},{"code":"Microsoft.Data.Mashup.ValueError.Key","detail":{"type":1,"value":"[entity = \"Scheduled Fees\", version = \"\"]"}},{"code":"Microsoft.Data.Mashup.ValueError.Reason","detail":{"type":1,"value":"Expression.Error"}},{"code":"Microsoft.Data.Mashup.ValueError.Table","detail":{"type":1,"value":"#table({\"entity\", \"dataCategory\", \"Data\", \"Tags\", \"entityName\", \"version\", \"ItemKind\", \"ItemName\", \"IsLeaf\"}, {})"}}],"exceptionCulprit":1}}}
Manual online refresh and pbix refresh are fine. Table is definitely there in Fabric.
Any ideas why it is failing on schedule?
Hi masplin
Thanks for checking the timings. Based on that, the DW refresh completed around 3:04am, while the semantic model refresh didn't start until 5:00am, so there doesn't appear to be a direct overlap.Since the scheduled refresh is working again this morning and there were no configuration changes, this could have been a transient issue at the time of the failed refresh rather than an ongoing configuration problem.
I would suggest monitoring the next few scheduled refreshes. If the same error occurs again, please check the refresh history for the detailed failure information and we can investigate further. Power BI recommends using the refresh history to compare the timing and error details of individual refresh attempts.
7 Replies
Hi masplin
Thanks for checking the timings. Based on that, the DW refresh completed around 3:04am, while the semantic model refresh didn't start until 5:00am, so there doesn't appear to be a direct overlap.Since the scheduled refresh is working again this morning and there were no configuration changes, this could have been a transient issue at the time of the failed refresh rather than an ongoing configuration problem.
I would suggest monitoring the next few scheduled refreshes. If the same error occurs again, please check the refresh history for the detailed failure information and we can investigate further. Power BI recommends using the refresh history to compare the timing and error details of individual refresh attempts.
Hi masplin,
When you say the manual refresh works, does it actually work?
Sometimes a refresh will "work" in the service, as in it won't fail, but if you look at the refresh history it will say that some tables failed.
Generally if the refresh works in the service, it works whether it is manual or scheduled. I have seen lots of cases where it works in desktop but not in the service, but I've not seen it work in the service manually but not scheduled.
I set a scheduled refresh for 10:30am which worked fine so wondering if its related to something happening at 5am making the table unavailable. Data looks like it has updated. See if it does it again tomorrow!!!
Hi masplin
What it looks like to me is that the SQL analytics endpoint has not been refreshed by the timely semantic model runs. I would recommend putting this as part of your data pipeline or using a notebook to refresh the SQL analytics endpoint before your semantic model. Be precious to see if that will solve your issue.
I wish I was intelligent enough to understand your advice!!! Sorry I am not a technology person so could you explain in idiot speak? Thanks
Hi masplin
In addition to GilbertQ tayloramy.
You're right to be suspicious of the error text. "The key didn't match any rows in the table" doesn't mean the Scheduled Fees table has been deleted.The useful part is actually the very end: #table({"entity", "dataCategory", ... }, {})
That's the list of tables Power BI got back, and the {} means it came back empty. If the table had simply been renamed or removed, the other tables would still be listed. Nothing was returned at all so the "key didn't match" message is a symptom, not the cause.
Those column names (entity, entityName, version) tell us the source is a dataflow. So the question becomes: why would the dataflow return no tables at that moment, but work fine when you refresh manually?
That's usually timing. The most likely explanation is that the dataflow is still refreshing when the semantic model's scheduled refresh starts. While a dataflow is mid-refresh its output can be briefly unavailable, so the model asks for the table list and gets nothing. Your manual refreshes happen at a normal time of day, well clear of that window.
What I would check first:
- Open the dataflow and check its Refresh history note the start and finish times.
- Compare those against the time the model refresh failed.
- Check whether the dataflow refresh itself failed or ran long that day. A failed dataflow leaves nothing for the model to read, producing exactly this empty list.
How to fix it:
- Short term, move the model's schedule later, leaving a generous gap dataflow durations vary day to day, so a few minutes usually isn't enough.
- The more reliable fix is to chain them, so the model refresh is triggered only after the dataflow finishes successfully. A Fabric pipeline or Power Automate flow can do this, so a slow dataflow day delays the model instead of breaking it.
- I would hold off on recreating the table or re-entering credentials for now. Since your manual refresh in the service works, and that uses the same stored connection as the scheduled refresh, the connection itself is unlikely to be the problem.
If the dataflow timings don't line up, let us know and we'll dig further.
https://learn.microsoft.com/power-bi/connect-data/refresh-data
Thanks,
C Srikanth
Community Support Team
The DW refreshed at 3am and take 4 minutes. The Failed Model refresh started at 5am and failed after 8 minutes so doesnt seme to be that? schedule work fine this morning. Seems jsut a bit of a hiccup?
Thanks
