DSXchange

All,

We have developed Parallel Jobs and we have used Hash Partion in the Jobs and for DB2 enterprise stage we have set "DB2" as partion method

Jobs are running fine with single node and when we are testing the same with 4 node we have problems

When we have changes the partion method as "SAME" on DB2 Enterprise the Jobs are working fine in 4 node .

Can anybody explain the reasons for this behavior and what are the pros & cons of using "DB2" or "SAME" as partion method

Thanks
Pani

How Many 'nodes' are there in the DB2 database?
(check the db2nodes.cfg file)

Hi Pani,

Eventhough you set the partitioning method as DB2 for DB2 stage, with single node the data is actually not partitioned. That is why you dont get any problems.

But when you run on multinode the data is partitioned across the nodes and that creates the problems

BTW what do you mean problems. Are they warnings/errors. Pls explain clearly.

And finally if you set the partitioning to SAME it will use the partitioning method use in the stage prior to the DB2 stage.

I hope I have answered all your questions.

Regards
--Rich

Thanks for replies

We are trying run with 4 nodes and Jobs are getting aborted if we use "DB2" as partion.

Error message is

main_program: Fatal Error: Virtual dataset; output of "fltChangeData"; input to "db2udbTEST_DB2ENT": dataset does not have a partition method

Thanks
Pani