Semantic Model

Hello,

I got the following error while trying to create a report model:

The semantic model is not valid. Details: The element 'Entity' in namespace 'http://schemas.microsoft.com/sqlserver/2004/10/semanticmodeling' has incomplete content. List of possible elements expected: 'IdentifyingAttributes' in namespace 'http://schemas.microsoft.com/sqlserver/2004/10/semanticmodeling'. Line 943, position 3. (rsModelingError) (Report Services SOAP Proxy Source)

------------------------------
For help, click: http://go.microsoft.com/fwlink/ LinkId=20476&EvtSrc=Microsoft.ReportingServices.Diagnostics.Utilities.ErrorStrings&EvtID=rsModelingError&ProdName=Microsoft%20SQL%20Server%20Reporting%20Services&ProdVer=9.00.1399.00

------------------------------
Program Location:

at Microsoft.SqlServer.Management.UI.RSClient.RSClientConnection.GenerateModel(String DataSource, String Model, String Parent, Property[] Properties)
at Microsoft.SqlServer.Management.UI.RSWrapper.NodeDataSource.GenerateModel(String name, String folderpath, String description)
at Microsoft.SqlServer.Management.UI.RSUserInterface.DataSourcePropertiesGenerateModel.OnRunNow(Object sender)
at Microsoft.SqlServer.Management.SqlMgmt.PanelExecutionHandler.Run(RunType runType, Object sender)
at Microsoft.SqlServer.Management.SqlMgmt.ViewSwitcherControlsManager.RunNow(RunType runType, Object sender)

I've searched every forum and every page from microsoft but couldn't find a single reference about this problem. I don't know what to do and I really need to get this model working fine for report builder

Can you help me

Andre

Answer this question

Semantic Model

tamberg

Can you be a little more specific on what you mean by "relating the month to the first day of the month in the Time table."

I believe I have this exact problem in generating a model, and I can't seem to identify a work around.
IIM

Hi Dave,

This case was created some months ago, even before SP1's release I guess, so I can't guarantee you that the same behaviour and the same solution are appliable.

Anyway, my problem came when I was trying to relate two different fact tables. One had the "real" data, refreshed everyday and the other one had the "budget" data, refreshed every month.

So, what I tried to do was to relate the "real" fact table to the attribute "day" of the time dimension and the "budget" fact table to the attribute "month" of the time dimension.

What happened was that although everything was alright in the cube, when I tried to create a model I got the problem listed above.

After knowing what was causing the problem, my work around was simple. Instead of relating the "budget" table to the attribute "month", I've related it to the attribute "day", the first day of the month (Example: all the January data was inserted as "01-01-2006" instead of "01-2006").

In the model, the user who is going to analyse the comparison "real"-"budget" will get the same results, except if he tries to analyse at "day" level. In that case, he will get all the January budget on the first day of the month.

In our case, the customer was confortable about this because they would not allow the end-users to drill down to the day level, but we managed to propose a solution, wich was to aggregate all measures from the "budget" fact table as "last non empty". That will get the same January budget at the month level and at the day level.

Hope it helps.

Best regards

Andre Santana
gl3nn86
Each entity in a model must specify one or more attributes as its "identifying" attributes. Automatic model generation is supposed to ensure this always occurs so you don't get errors like this, but it appears you've found a case we missed. Do you have a table in your database with no columns (at least none visible to the logged-in user) Do you have a table with no primary key defined
Fipil

Relationships in AS that are defined at the non-leaf level are not picked up by Report Builder. This probably contributed to the problem listed above. Tables without primary keys will also not be picked up in model generation though should not throw an error.
ReneMiner

Bob,

I have now just hit this same error... using SQL Management Studio, connecting to Reporting Services and attempting to 'Generate Model...' and have a report model built off of my cube data source.

The error text is much like that posted above...

Is there a better answer than the one posted thus far on this topic

Thanks!
James E. Davey

Hello Rob

First of all, thanks for your reply.

I've checked my cube and there were two tables with no primary key defined (the fact tables). I've created a auto-incremental field in each one and placed them as the primary key.

So, my cube doesn't have tables with no visible rows and doesn't have tables with no primary key defined, but the error still occurs, exactly the same that happened before.

Do you have any other ideia about this

Thanks

Andre
simonl

I have the solution.

After the opening of the case in microsoft, this problem was analysed by the support teams and I got an answer today.

The problem seems to be one relationship i've built between a fact table and a dimension table.

In the current release of SSRS and SSAS, a relationship from one entity to another is implicitly to the key of the destination entity.

I was using a Time dimension (the dimension key was "day") and I was building a relationship based on the attribute "month".

So, the relationship was ignored and my fact table had no relationships and therefore, no identifyers, and that was the cause of the problem.

The solution for my case was relating the month to the first day of the month in the Time table.

I was told Microsoft plans to lift such limitation on the next releases.

Hope this helps for anyone with similar problems

Andre
mageets

Thanks Andre and Carolyn. That definately explains the error. I have a data in the cube that is not at the leaf level.

For a whole host of reasons, I can't implement the relationship change Andre suggested. Short of removing the offending data, is there any way to have Report Builder resolve this issue Possibly ignore the data

In this case, for a variety of reasons, the data is implemented as a separate "cube" within the MSAS database. Suppressing / ignoring the cube would be fine in the model generation. I'm wondering if it's possible to generate a model, where certain cubes are ignored or suppressed. I've tried to create a model using a user account with limited access to the database, but it appears that the user generating the model must have full (Administration) access to the cube.

Suggestions.
Semantic Model

Answer this question