DSXchange

Hi,

I am getting the following error in my SCD stage.

"SCD,2: Failure during execution of operator logic. SCD,2: Fatal Error: Index too large"

Kindly help me to solve this..

Not enough information. What logic are you using in your SCD stage? What is it about the third node (#2) that caused the failure to be there?
What data volumes are involved? How big (number of rows, size of rows) is the dimension table?

ray.wurlod wrote:Not enough information. What logic are you using in your SCD stage? What is it about the third node (#2) that caused the failure to be there?
What data volumes are involved? How big (number of row ...

Hi Ray,
Thanks for your response. I am using TYPE 1 logic.
Not only the third node, all the nodes return the same error. My source table is having 100 records and the lookup reference table is having 10 records.

You're going to have to involve your official support provider here. I suspect - but can't get close enough to the design to be sure - that the "index" here is a reference to a slot in the internal table where the dimension data are stored, and that it's gotten to be larger than the structure (that is, that you've found a bug). Obviously I can not prove this - it's only a guess. You need someone with more specialized diagnostic tools and who is immediately available.

I found out that this error message index too large with SCD stage is quite generic. For example I have two versions of client software installed, one with client side patch from support and another one without the patch. When using Oracle DBSequence for surrogate key generator inside SCD those clients generate different osh code for the same job.

The client version without the patch generated osh code related to SCD stage without remote_server, user and password parameters while patched version generated the code correctly.

When trying to run the job generated with the client without the patch I got:

handle_SCD,0: Failure during execution of operator logic.
handle_SCD,0: Fatal Error: Index too large
node_node1: Player 6 terminated unexpectedly.....

So this error message is at least not very describing in nature which might be related to the fact that SCD stage is quite new.

Another aspect is that there are some known bugs in SCD stage which require a correct patch from the support.

I am running DataStage 8.0.1 fixpack 1a on Linux Server with certain patches from support.

Are you using parameters for the surrogate key connection information? username / password... if yes then hardcode them and see if it works. It should work.

"index too large" is misleading. The real issue is 1) that the SCD stage can not access the oracle sequence object this is most likely because you are using job parameters and 2) that IBM doesn't believe in QAing their products. I had the same issue with the transformer stage and DB2 sequences.

If you are not using job parameters then check that your sequence works using a surrogate key generator stage. It will help flush out any connection issues.