This discussion is archived
5 Replies Latest reply: Nov 21, 2012 12:43 PM by Srini VEERAVALLI RSS

Error 38015

Ric79 Newbie
Currently Being Moderated
Hi all,
I have a modeling problem.

Suppose that there i a fact: RELATIONSHIP using twice the dimension NODE.
The dimension NODE is a factless fact merging other dimension: role, product, sw, hd, version, area, etc.
                    +--- SW
                    |
RELATIONSHIP ===== NODE --- ROLE
                    |
                    +--- PRODUCT
When I try to link the fact twice with the dimension NODE, OBIEE return error 38015 (have multiple joins. Delete new foreign key object if it is a duplicate of existing foreign key).
So I should create an alias for the NODE (NODE_ORG, NODE_DST) and for all other dimensions related
                  +--- SW_ORG                      +--- SW_DST
                  |                                |          
 ROLE_ORG --- NODE_ORG  ----- RELATIONSHIP ----- NODE_DST --- ROLE_DST
                  |                                |           
                  +--- PRODUCT_ORG                 +--- PRODUCT_DST
In my case there are several dimensions and NODE is shared with other facts. So I must duplicate everything connected to NODE.

Is there an elegant solution for bypassing the problem in OBIEE?

Riccardo
  • 1. Re: Error 38015
    Srini VEERAVALLI Guru
    Currently Being Moderated
    If a dimension NODE is joined with diferent facts like F1,F2 etc no need to go with alias for NODE.
    If the NODE wants to join with same fact F1 for different join conditions then you have to go alias for NODE.

    As per the consistence check rpd will allows a single join between the physical layer objects, so as per the best practices we go for alias tables for each leaving original tables a side.

    hope this helps
  • 2. Re: Error 38015
    Ric79 Newbie
    Currently Being Moderated
    Thanks!

    let'us suppose that NODE is a conformed dimension shared between 10 facts.
    let's suppose that from RELATIONSHIP fact I want to drill down to NODE ORIGIN or NODE DESTINATION and than to the 10 facts..

    I need to create and alias for all 10 fatcs for splitting the case of role ORIGIN from role DESTINATION...Is it correct?

    Is there a better design approach avoiding this problem?
  • 3. Re: Error 38015
    Srini VEERAVALLI Guru
    Currently Being Moderated
    NODE as confirmed dimension shared among 10 facts: Assuming these 10 facts sitting in same subject area, you suppose to use Content tab to set NODE levels so that BI server pick right source(fact) based on columns select.

    If helps pls mark
  • 4. Re: Error 38015
    Ric79 Newbie
    Currently Being Moderated
    sorry my question was different

    IF NODE is conformed among 10 facts and, cause my original question, I need to create two aliases of NODES ... this means that i need to create aliases for all 10 facts for ensuring a right drill-down or rollup

    right?

    Edited by: Ric79 on Nov 21, 2012 9:38 PM
  • 5. Re: Error 38015
    Srini VEERAVALLI Guru
    Currently Being Moderated
    Yes,To drill or roll up based on NODE ORIGIN or NODE DESTINATION

Legend

  • Correct Answers - 10 points
  • Helpful Answers - 5 points