Prallel join creating rows as many as nodes

johnson_binkam · Post by **johnson_binkam** » Wed Jul 11, 2007 9:23 pm

We are converting server jobs to parallel. Coded a join stage with 2 inputs(oracle enterprise). query gives 100 rows and the join stage gives 800 rows(8 nodes).If we use ODBC stage inputs, it is giving 100 rows.

Is there any parameter or configuration can cause this problem?

ArndW · Post by **ArndW** » Wed Jul 11, 2007 9:32 pm

Are you doing any repartitioning? Does the job monitor show 800 rows output or have you actually checked the data to see if it is wrong. I've noticed some discrepancies in the monitor row counts when using several nodes, sometimes it showed total row counts per node.

johnson_binkam · Post by **johnson_binkam** » Fri Jul 13, 2007 8:34 am

No Re-partitioning. Out put is actually 800 rows.
Found the cause of the problem and we move on. One of the inputs were created by Ascential and not Oracle DDL, when we pre-create the table with Oracle DDL it is giving just 100 rows.

ray.wurlod · Post by **ray.wurlod** » Fri Jul 13, 2007 3:25 pm

If the problem is fixed please mark the thread as Resolved.

And post a little more information - what do you mean by "created by Ascential"? That company ceased to exist in 2005.