Forum Discussion
Importing SSAS tabular data, Power BI stuck on evaluating
I sort of did, though I’m not sure if it applies to my original problem here. I ended up having various issues where Power BI Desktop would not import data from an SQL Server source either (using which was my first workaround), even when the datasets in question were very small. Ultimately the issue was that we were running Power BI in a Hyper-V virtual Windows machine, and the solution was to increase the number of virtual cores from one to at least two.
Power BI Desktop requirements are given as "CPU: 1 gigahertz (GHz) 64-bit (x64) processor or better recommended", with no mention of multicore being a requirement. Maybe rufengli or someone else at MS should amend this or look into it? I understand that 64-bit processors in practice may always be at least dual-core, but considering that virtualised environments are not uncommon, it would not hurt to clarify this. Also, it would have been helpful if Power BI could throw some sort of error in such cases, instead of running a query forever.
Hopefully this is helpful.
JargerBiirli Did you figure this out?
Are you an SSAS admin?
Perhaps you can run a profiler trace against the back-end database?
I'm having very odd issues as well, and the culprit seems to be the query,
select [MEASURE_UNIQUE_NAME], [MEASURE_CAPTION], [DATA_TYPE], [MEASUREGROUP_NAME], [MEASURE_DISPLAY_FOLDER] from $system.mdschema_measures where [CUBE_NAME] = @CubeName and [MEASURE_IS_VISIBLE]
For whatever reason, this metadata query seems to create a conflict in SSAS with processing operations. In general there should never be conflicts between SSAS readers and writers until the commit phase (which takes only a second). But something really weird is going on with this SQL metadata query....
... I'm finding that the conflicts happen much sooner than you would expect. I'm finding that in some cases when Power BI is trying to refresh data, then it may end up being locked for the entire duration of the processing operation (~10 mins), and subsequently cancelled during the commit phase (after another ~30 seconds of not completing).
These types of metadata queries should be instantaneous! So I find it highly suspicious that they would end up taking 10 mins, and then needing to be cancelled as part of the commit phase. We are running SSAS on SQL 2016 CU 12 (13.0.5698.0).
I sort of did, though I’m not sure if it applies to my original problem here. I ended up having various issues where Power BI Desktop would not import data from an SQL Server source either (using which was my first workaround), even when the datasets in question were very small. Ultimately the issue was that we were running Power BI in a Hyper-V virtual Windows machine, and the solution was to increase the number of virtual cores from one to at least two.
Power BI Desktop requirements are given as "CPU: 1 gigahertz (GHz) 64-bit (x64) processor or better recommended", with no mention of multicore being a requirement. Maybe rufengli or someone else at MS should amend this or look into it? I understand that 64-bit processors in practice may always be at least dual-core, but considering that virtualised environments are not uncommon, it would not hurt to clarify this. Also, it would have been helpful if Power BI could throw some sort of error in such cases, instead of running a query forever.
Hopefully this is helpful.