OBIEE Blog

If you’re running OBIEE under 64-bit Solaris – this might be of use. I just want to add from myself – that it’s not a good idea to change repository in online mode.

Admintool Check-in changes: Internal Assertion Error Condition FALSE , file server/Utility/Generci/NQThreads/SUGThread.cpp, line 515 [ID 820803.1]

When checking in changes in Admintool, you are getting the error:

nQSError:28019 Near line 230: In the metadata expression … the following error occured nQSError 46036 Internal Assertion Error Condition FALSE , file server/Utility/Generci/NQThreads/SUGThread.cpp, line 515
Cause

This is a defect:
Bug 6652490: PSR:FUNC:ESSBASE: OBSERVED THE ERROR WHILE CHECK-IN OR SAVING RPD IN ONLINE MODE ?

Bug was logged for 10.1.3.3.2 with OBI server on Solaris 64 bits

Although the abstract mentions Essbase the bug is reproducible against any database.
But only when OBI is on 64 bits solaris platform.
It applies to all versions from OBI 10.1.3.3.1 and later.

Test case:
From Admin tool created the new rpd (i.e. physical layer) and Drag-Drop
the db to Business model & Presentation layer then tried to Check In / Save
form Admin tool in Online mode but I observed the below error form Admin tool
window as well as NQServer.log
Error Message:
[46036] Internal Assertion: Condition FALSE, file server/Utility/Generic/NQ
Threads/SUGThread.cpp, line 515.
2007-11-27 04:44:12
[46036] Internal Assertion: Condition FALSE, file
server/Utility/Generic/NQ
Threads/SUGThread.cpp, line 515

Off line mode able to check-in or saved sucessfully.

Also we only had the error on a Business Model that had been dragged and dropped from the physical.
Solution
The workaround is to modify the NQSConfig.INI file.
Change the “SERVER_THREAD_STACK_SIZE = 0;”
with “SERVER_THREAD_STACK_SIZE = 512 KB;”

and restart the OBI server

Increasing thread stack size (by setting SERVER_THREAD_STACK_SIZE to a bigger number) is the solution to this issue in all versions including 10.1.3.4.

Increasing thread stack size requires more memory from the server machine.
That is the only impact it should have. But increasing from 256k (default) to 512k is a minor change

This defect will be fixed in the next main release

Hello everyone.

I just realized that this post is inspired by Venkat.

I’m finally posting a solution to the problem I’ve identified a while ago related to a specific Pivot Table View. I think the final outcome should look something like this image:

To get this result – I’ve used the “New calculated column” menu on the pivot table (circled in red):

OBIEE pivot table menu

Then, what you really need to have in the Columns is the most granular level column (sounds redundant – in this case it’s a Date column). I used Excel to quickly generate formulas, such as this for Q1.  It’s really easy to generate these formulas below. I hope it’s pretty self-explaining – you add things up in a granular level – this way you can combine things – that you wouldn’t be able to combine otherwise:

‘1/1/2007′+’1/2/2007′+’1/3/2007′+’1/4/2007′+’1/5/2007′+’1/6/2007′+’1/7/2007′+’1/8/2007′+’1/9/2007′+’1/10/2007′+’1/11/2007′+’1/12/2007′+’1/13/2007′+’1/14/2007′+’1/15/2007′+’1/16/2007′+’1/17/2007′+’1/18/2007′+’1/19/2007′+’1/20/2007′+’1/21/2007′+’1/22/2007′+’1/23/2007′+’1/24/2007′+’1/25/2007′+’1/26/2007′+’1/27/2007′+’1/28/2007′+’1/29/2007′+’1/30/2007′+’1/31/2007′+’2/1/2007′+’2/2/2007′+’2/3/2007′+’2/4/2007′+’2/5/2007′+’2/6/2007′+’2/7/2007′+’2/8/2007′+’2/9/2007′+’2/10/2007′+’2/11/2007′+’2/12/2007′+’2/13/2007′+’2/14/2007′+’2/15/2007′+’2/16/2007′+’2/17/2007′+’2/18/2007′+’2/19/2007′+’2/20/2007′+’2/21/2007′+’2/22/2007′+’2/23/2007′+’2/24/2007′+’2/25/2007′+’2/26/2007′+’2/27/2007′+’2/28/2007′+’3/1/2007′+’3/2/2007′+’3/3/2007′+’3/4/2007′+’3/5/2007′+’3/6/2007′+’3/7/2007′+’3/8/2007′+’3/9/2007′+’3/10/2007′+’3/11/2007′+’3/12/2007′+’3/13/2007′+’3/14/2007′+’3/15/2007′+’3/16/2007′+’3/17/2007′+’3/18/2007′+’3/19/2007′+’3/20/2007′+’3/21/2007′+’3/22/2007′+’3/23/2007′+’3/24/2007′+’3/25/2007′+’3/26/2007′+’3/27/2007′+’3/28/2007′+’3/29/2007′+’3/30/2007′+’3/31/2007′

Calculated item OBIEE

The same logic works for Q2,3,4. For individual dates, it would be the same idea:

And for months, it’s the same logic:

To sum up  – by using the “calculated item” feature of pivot table view – you can create reports with mixed granularity on them. The pros are that you’ll be able to support legacy views (in some instances it’s absolutely critical to mimic existing view). The cons are that it involves some manual entry / data manipulation and that it’s not very flexible (I couldn’t find out how to use variables in this example).

Last thing – it seems as first column gets added on the left and the consequent ones to the right – keep this in mind while designing this view – since I’ve noticed you can’t really manipulate with the locations of the calculated columns.

A. “Ago function returning wrong results in answer when comparing this year versus last year” This actually happened to me once – and i thought it was an incorrectly configured time dimension. You can experience this error when your “Time Series functions gives incorrect results when leap years are involved”. Oracle support suggested the following action plan :

1) Leave the logical column definition as earlier i.e. Ago(Measure, Year, 1)
2) Change the filter condition in the request to filter on year and months instead of date range e.g.
year = 2009 and month in (1,2,3,4)

B. “Value Suppression doesn’t work for Level Based Measures”. Someone had a requirement where they wanted to use level-based measures for reports accessing several fact tables in the future. To realize a prototype before starting with the necessary reports they tried to implement an easy example to test the ‘value suppression’ functionality

Unfortunately they are not able to build a nice looking report with four simple columns showing the year, # of orders (per year), quarter and # orders. Their problem is that the level based measure ‘# of order (per year)’ is shown line by line. They wanted to suppress these values and only show the value once per year. They tried to use the ‘value suppression functionality’ without success. They were able to Work around this issues by using a Pivot Table.

C. This issue in pivot table is very common –

OBIEE says “governor exceeded error”. The pivot tables and the charts might give these errors:

Error Generating Chart
An exception occurred in the cube.
Error Details
Error Codes: MOJ5XVNL:QBVC92JY
Governor limit exceeded in cube generation (Maximum data records exceeded.)

and for pivot table

View Display Error
Governor limit exceeded in cube generation (Maximum data records exceeded.)
Error Details
Error Codes: QBVC92JY

Theses messages mean that users are trying to return more rows to the pivot table then the governor will allow. One needs to adjust the following parameters until the report runs successfully.

Adjust the parameters below in instanceconfig.xml (add those tags if you don’t have them – since those values are by default) and then restart Presentation Server (sawserver)

< CubeMaxRecords >40000< /CubeMaxRecords >
< CubeMaxPopulatedCells >150000< /CubeMaxPopulatedCells >

Someone asked the Oracle support if Siebel Analytics could use 2 LDAP servers for authentication. “We currently use one of the LDAP server for authentication and it works fine. We have some users from another LDAP server and want to add it to our authentication.”

Response was:
Siebel Analytics can only authenticate against on LDAP server at a time. If you define multiple LDAP servers, the 2nd LDAP server is not being looked at unless the connection to the 1st failed.

You define multiple LDAP server by putting the LDAP server names in a concatenated string and use space as delimiter in the Host Name entry, i.e. “ldap1.siebel.com ldap2.siebel.com:9872 ldap3.siebel.com” in AN78x release

In the latest Oracle Business Intelligence Enterprise Edition, you can define multiple LDAP server definitions in the Manage > Security > LDAP Server.

But Authentication tries the first LDAP server, if it fails then it uses the second LDAP server defined in the repository

“Product change request 12-HOAZ87 has been logged to ask for authentication against multiple LDAP servers simultaneously.”

I’ve been working on helping to configure system subject area to automate configuring Delivery Profiles. Unfortunately, Oracle’s documentation is severely lacking – and there’re only a few tutorials online. Even though they’re good – something isn’t working out yet – there’s an error in the Sessions log – that says something like nqsError “unable to locate table SET”. The only SET which is there looks like “Set Loglevel=2″. I’m trying to recreate all steps from the beginning.

On the other hand, I strongly believe that a partial in-place drill-down should be a feature in the OBIEE. Even though clicking on a column in BI Answers partially solves the issue – it doesn’t do in in such a way it would be useful. I’ve seen OBIEE being used for transactional type of reporting where users want to go deeper into detail level without loosing focus on other items.