Forum Discussion
Automatic Aggregations with Direct Query not working
We are facing an issue with our semantic models after the move to ADBC connector. Our automatic aggregations do not generate any cache.
It was working when we initially published it, but the size kept reducing and now it's gone completely.
We ran trainings multiple times but it returns this error - Workload processing finished training without generating aggregations because query log processing did not identify any meaningful aggregation candidates this time. The calculated ratio was '0.0197265515592077' while the expected ratio is '0.75'. The next training might determine better aggregation candidates based on the continuously updated query log.
We need to understand why this is happening as none of my new models are generating any queries and its affecting performance.
3 Replies
For one of the models the error is - No SystemManaged table found after PA training
v-achippaCommunity Support
Hi V_S27,
Thank you for reaching out to Microsoft Fabric Community.
Here the error message means that the automatic aggregations training completed but did not find suitable query patterns, so no system managed aggregation cache was created.
Run the affected reports normally to generate representative DirectQuery queries and then use Train and Refresh Now once more. Automatic Aggregations use the query log to identify aggregation candidates, so the model needs representative query activity before training.
For reference, follow the below documentation:
Automatic aggregations overview - Microsoft Fabric | Microsoft Learn
Configure automatic aggregations - Microsoft Fabric | Microsoft LearnThanks and regards,
Anjan Kumar Chippa- Jai-Rathinavel
Super User
Hi V_S27 , please check this detailed troubleshooting Guide and let me know if it is what you are looking for
Troubleshooting automatic aggregations to optimize DirectQuery performance | Microsoft Fabric Community
