Yesterday I tested out a filter the Sample Sales Lite subject area. For my selected columns I chose Time Hierarchy, Products Hierarchy, and the Revenue base fact. I then applied a filter on the Product column where it is not equal to / is not in "V5x Flip Phone; CompCell RX3; Touch-Screen T5; KeyMax S-Phone; PocketFun ES; MaxiFun 2000". Going to Results, it appears to show the proper results with a new aggregation for the Total Time and All Products level. However, if the Product column is selected, then it gives entirely different results with an incorrect aggregation.
I then attempted to apply the same type of filter (2 dimensions, 1 measurement, filter on an attribute of one of the two dimensions) to my own cube that was made in AWM, but I get the following error: "[nQSError: 43113] Message returned from OBIS. [nQSError: 59137] Filter level STUDENT_TYPE is below the projected level ALL_STUDENT_TYPES on dimension STUDENT_TYPE_2 while an externally aggregated measure is present.Please have your System Administrator look at the log for more details on this error. (HY000)"
I am suspecting that the cause for the difference of behaviour between the two cubes is due to the fact that the Sample Sales Lite cube is XML based and is essentially local to the BI environment, whereas my cube resides within an Oracle database. Is it just not possible to apply a filter on an Oracle-database supported cube in BI? We are running Business Intelligence 18.104.22.168.0.
One more thing.
Screenshots are always helpful. So if you can upload the screenshots with some details to a website (like dropbox) and provide the link to that doc in your post then it improves your chances of getting an answer quickly.