Forum Discussion

GLWY's avatar
GLWY
Frequent Visitor
2 years ago
Solved

PBI Desktop query caused issues in Power BI Service

Hi,

 

We recently encounter a very intersting issue. On Monday, our Power BI Service went down for about 2 hours and nobody could run a report. We investigated and found that the throttling % went over 5 times its limit. Meaning it was processing a query and did not allocate any CU for incoming requests.  

 

We found that the issue was a query sent from Power BI desktop. We found that it happened when a user downloaded an old report and tried to run it in Power BI desktop. It does not behave the same way as before and is now by-pass all limitations due to the Query-limit simulation feature introduced last year. 

 

The feature sets all old reports to "No query limits" and because we are using live connection to Semantic Models, it seems to by-pass the limits we have in Power BI Service. However new reports are set to Auto which is the recommended option.

 

 

 

My question is, is there a way to update all reports in Power BI service and set them to Auto? so that we don't end up with this situation again

 

  • My question is, is there a way to update all reports in Power BI service and set them to Auto? so that we don't end up with this situation again

    no, there is not.  This is not your only issue though. Keep in mind that dataflow refreshes also do not have any duration timeouts. A single rogue dataflow can bring down your entire capacity for a long time when it completes.  We have now setup a monitoring process to automatically cancel all dataflows that run longer than five hours. It's not foolproof though as cancelation requests may or may not be honored by the Power BI service.

2 Replies

  • My question is, is there a way to update all reports in Power BI service and set them to Auto? so that we don't end up with this situation again

    no, there is not.  This is not your only issue though. Keep in mind that dataflow refreshes also do not have any duration timeouts. A single rogue dataflow can bring down your entire capacity for a long time when it completes.  We have now setup a monitoring process to automatically cancel all dataflows that run longer than five hours. It's not foolproof though as cancelation requests may or may not be honored by the Power BI service.