This is ignorable and has nothing to do with the date entries. For reasons not clear to me it tends to get reported only when doing full processing, so it can appear to go away when reprocessing a previously processed dimension because the default processing type in ssas is not a full process.
The explanation for this is when ssas joins a dimension to a measure group if the join column in the measure group is null ssas incorrectly, at least in my opinion, logs a null key error. This should really be a warning not an error as nulls are allowed in join fields. Lots of people were seeing these null key errors and would call up tech support to ask what they mean. To avoid having all these null errors and to speed up processing (which was impact by having to log all those messages), in the data source view, for every dimension query, an extra row is added per company where the values for the row are all set to N/A for strings or 0 for numeric types. When grouping by the dimension, any row in the measure group that have NULLs in the join columns are displayed under the attribute value N/A.
This has a side effect on the DataAreaId attribute.
The query for that attribute ends up with two values per company, one where the company name is N/A and one where the company name is the real name of the company. The reason a duplicate key warning is reported for this attribute is because of the presence of the extra row linked to the company that has N/A for the company value but is linked to the company.
The impact is that when a query groups by dataareaid attribute, the data rows will be linked to the first row for the company, which is what one would want anyway, and not to the N/A row.
You can see the N/A row is triggeing this by getting the query ssas uses for the dataareaid attribute and running it.
When you process the dimension from sql management studio, if you expand the dimension entry in the results and then expand the result for the attribute, you should be able to see the query for the attribute.
I get this query. Yours should be similar except that the filter value for the partition field will probably be a different number.
SELECT
DISTINCT
[WRKCTRTABLEEXPANDED].[DATAAREAID] AS [WRKCTRTABLEEXPANDEDDATAAREAID0_0],[WRKCTRTABLEEXPANDED].[DATAAREAID_NAMECOLUMN] AS [WRKCTRTABLEEXPANDEDDATAAREAID_NAMECOLUMN0_1]
FROM
(
SELECT WORKCENTERSSITE, WORKCENTERS, WORKCENTERGROUP, CAPACITYUNIT, WORKCENTER, WRKCTRID, CATEGORYNAME, COSTCATEGORYID, CATEGORYTABLE, SITE, PRODUCTIONUNITS, PRODUCTUNIT, DATAAREAID, PARTITION, RECID, NAME, RESOURCEGROUPID, DATAAREAID AS DATAAREAID_NAMECOLUMN, 0 AS BI_ISNOTAPPLICABLE FROM [DBO].WRKCTRTABLEEXPANDED WHERE WRKCTRTABLEEXPANDED.PARTITION = 5637144829 UNION ALL SELECT NOTAPPLICABLE.WORKCENTERSSITE AS WORKCENTERSSITE, NOTAPPLICABLE.WORKCENTERS AS WORKCENTERS, NOTAPPLICABLE.WORKCENTERGROUP AS WORKCENTERGROUP, NOTAPPLICABLE.CAPACITYUNIT AS CAPACITYUNIT, NOTAPPLICABLE.WORKCENTER AS WORKCENTER, NOTAPPLICABLE.WRKCTRID AS WRKCTRID, NOTAPPLICABLE.CATEGORYNAME AS CATEGORYNAME, NOTAPPLICABLE.COSTCATEGORYID AS COSTCATEGORYID, NOTAPPLICABLE.CATEGORYTABLE AS CATEGORYTABLE, NOTAPPLICABLE.SITE AS SITE, NOTAPPLICABLE.PRODUCTIONUNITS AS PRODUCTIONUNITS, NOTAPPLICABLE.PRODUCTUNIT AS PRODUCTUNIT, KEYCOLUMNS.DATAAREAID AS DATAAREAID, NOTAPPLICABLE.PARTITION AS PARTITION, NOTAPPLICABLE.RECID AS RECID, NOTAPPLICABLE.NAME AS NAME, NOTAPPLICABLE.RESOURCEGROUPID AS RESOURCEGROUPID, NOTAPPLICABLE.DATAAREAID_NAMECOLUMN AS DATAAREAID_NAMECOLUMN, 1 AS BI_ISNOTAPPLICABLE FROM ( SELECT DISTINCT DATAAREAID FROM [DBO].WRKCTRTABLEEXPANDED WHERE PARTITION = 5637144829 ) AS KEYCOLUMNS CROSS JOIN ( SELECT CAST(N'N/A' AS NVARCHAR(3)) AS WORKCENTERSSITE, '' AS WORKCENTERS, CAST(N'N/A' AS NVARCHAR(3)) AS WORKCENTERGROUP, NULL AS CAPACITYUNIT, CAST(N'N/A' AS NVARCHAR(3)) AS WORKCENTER, CAST(N'N/A' AS NVARCHAR(3)) AS WRKCTRID, CAST(N'N/A' AS NVARCHAR(3)) AS CATEGORYNAME, CAST(N'N/A' AS NVARCHAR(3)) AS COSTCATEGORYID, '' AS CATEGORYTABLE, CAST(N'N/A' AS NVARCHAR(3)) AS SITE, CAST(N'N/A' AS NVARCHAR(3)) AS PRODUCTIONUNITS, CAST(N'N/A' AS NVARCHAR(3)) AS PRODUCTUNIT, 0 AS PARTITION, 0 AS RECID, '' AS NAME, 0 AS RESOURCEGROUPID, CAST(N'N/A' AS NVARCHAR(3)) AS DATAAREAID_NAMECOLUMN, 1 AS BI_ISNOTAPPLICABLE ) AS NOTAPPLICABLE
)
AS [WRKCTRTABLEEXPANDED]
When I run this query I get back duplicates too.
WRKCTRTABLEEXPANDEDDATAAREAID0_0 WRKCTRTABLEEXPANDEDDATAAREAID_NAMECOLUMN0_1
dat dat
dat N/A
Hope this helps understand what is happening,
--Lance
