ODBC Stage Error Regarding Oracle Table Partition

[djwagner]

Hello,

I'm running a parallel job on IBM WebSphere DataStage Enterprise Edition 8.1.0.4987 against a partitioned table on Oracle.

The job runs fine when the destination (output) ODBC database stage is set to an Execution Mode of "Sequential". But when it is set to "Auto (Parallel)", the job aborts with the following error: "ORA-14400: inserted partition key does not map to any partition".

Am I understanding correctly that this means that the partition key that Datastage chooses for its own partitioning MUST match the partition key that is set up for Oracle's table partitioning? But what I don't understand is why they have to match, because aren't they two totally separate types of partitioning? (Datastage partitioning is its own internal processing across multiple CPUs compared to Oracle's partitioning of the table at a storage level)? Why do the keys have to match?

And if they DO in fact have to match, then how can I determine what partition keys Datastage is choosing when it is set to "Auto"? (I can set the Oracle table partition keys to match, if necessary)


The job also runs fine when the destination (output) database stage is the native Oracle database stage. Unfortunately, we are receiving intermittent "ORA-21561: OID generation failed" which is why we are attempting to use the ODBC database stage instead.



Thanks,
David Wagner

[chulett]

This sounds like one of those times when you would not want to go with Auto partitioning and instead would want to explicitly control it. Rather than consider changing how the Oracle table is partitioned, change the DataStage job to match the table.

And then see if that indeed is your issue.