COGNOiSe.com - The IBM Cognos Community

Planning & Consolidation => COGNOS Planning => Topic started by: StuartS on 25 Feb 2009 07:52:41 AM

Title: Wierd Contributor Behaviour
Post by: StuartS on 25 Feb 2009 07:52:41 AM
Hello All

I am posting this in the hope someone has experienced something like this in a model.

One cube in a model in Contributor (8.1) is showing strange results in a cell.  An example to try to illustrate;

Cube A transfers a value to Cube B.  The Dlink says this is a 1 to 1 link.  I.e. cube A has a value of 2 and Cube B has value of 2. 

In my model a single elist (out of a total of 390ish) has cube A at 2, but is transferring the value 4 to cube B.  So ALL other elists are showing the correct result, just this one is wrong.

I have extracted the data from Contributor to Analyst.  The data shows Cube A as 2 and Cube B as 4.  But when I run the D-Link in analyst the value in Cube B goes from 4 to 2.  2 is correct.

Has anyone ever experienced this?  Does anyone think something may have gone awry in a reconcile, SQL creation by Cognos?

Heres hoping...

Stuart
Title: Re: Wierd Contributor Behaviour
Post by: sascha on 25 Feb 2009 01:59:02 PM
Does the link always double the data? What happens when entering 3 instead of 2? Will this get a 6?
Does this really happen by dLink or by Adminlink?
Title: Re: Wierd Contributor Behaviour
Post by: overflow.au on 25 Feb 2009 09:38:03 PM
The other question is whether there are any unpaired dimensions in the target cube which could be causing the duplication? 
Title: Re: Wierd Contributor Behaviour
Post by: StuartS on 26 Feb 2009 02:37:12 AM
Hello

Cube A, the 2, is a READ ONLY cell and so is copy/load/prepare & GTP/Reconcile.  I am waiting to get a copy of the database into a dev environment so I can test what will happen if this is changed.  I can only guess another reconcile of the data will fix the anomaly.

However, in the analyst model the D-link works as designed.  As mentioned if I extract the data from the Contributor model into analyst and then run the D-link the value in cube B is updated correctly.

The most bizarre thing here is that ALL of the other elists work ok so I know the model is sound, i.e. cube A passes to cube B correctly.  It is only this one elist where the anomaly is seen.


This model has not had a reconcile cancel, cut down model cancel, or any jobs error.


@overflow.au  I wish it was this problem, but sadly no.
@Sasha the D-Link does not do any transformation to the data.  It seems as if the value in the Contributor model only has doubled.  As mentioned in analyst it is OK.
Title: Re: Wierd Contributor Behaviour
Post by: StuartS on 26 Feb 2009 08:56:11 AM
Hello

I have now got a copy of the SQL DB in a development environment.

I forced a value in cube A, did a GTP and subsequent reconcile.

The problem fixed.  Cube B now shows 2.

Wierd.

Stuart