DSXchange

I just recently worked with DataStage support on an issue I was having much like this. What it turned out to be was a job level setting. IBM advised me at that in the Job Parameters option to look on the performance tab. Once there they had me un-check the 'use poject defaults' and select the 'enabl...

Something I have done that is a bit cheesy is to call a unix script, and pass the variables into it. And then in the unix script, execute the java program.

You can also source the dsenv in the actual script you are executing in your before-routine.

I did some more testing and it is a deadlock issue. It appears to be an issue that is a combination of how parallel jobs work in combination with bitmap indexes in oracle. I have no 'real' proof, but I believe that since my data is unique(according to its primary unique key) but many insert threads ...

Thanks for the reply! It got me thinking about something and I found a solution. I'm not sure it's the best, but it works. I set the oracle commit level environment variable to '1'. It processed all 500k+ records in 17 minutes.

I have a very simple PX job: Oracle ---> Transformer ---> Oracle In the left oracle stage I am selecting a subset of a table, and in the right oracle stage I am inserting those records. The volume of data is not that large (approx. 500,000), but the job runs forever and never completes. I also get n...