My SSAS db has a server time dimension. When I add the dimension to my cube and relate the date (key) attribute to a datetime measure group column of my fact table that has date and time I get an error when I process the cube. I've set the unknown member property to visible in the time dimension and set null processing to unknown member in the dimension usage as you suggest.
This is the error: " ...The attribute key cannot be found when processing: Table: 'some_table', Column: 'some_date',Value: '5/2/2008 2:48:40 PM'. The attribute is 'Date'. Errors in the OLAP storage engine: The attribute key was converted to an unknown member because the attribute key was not found."
This error is the reason I converted the fact table to a named query in the DSV and stripped off the time from the date field.
I suspect that the attribute key members of the server time dimension do not have hh:mm:ss precision. Looking at the Type property of the server time key attribute "Date" I can see that it is set to days. If I change the Type from days to date, SSAS warns me that "...DimensionAttribute [Time].[Date] : The 'Type' property should be 'Days' for the key attribute of a time dimension." With that in mind, do you know if it is possible to get hour precision from a server time dimension, or will I have to create a time dimension from scratch? I read some article that suggested that a time dimesion should really be a snowflake made up of year, semester, quarter, month, day... tables.
Also, do you know if when the cube is processed in SSAS is SSAS is doing some kind of implicit conversion of the measure group column like from datetime to int or decimal?
sorry for the long winded post