DSXchange

Split the job. Simple as that. Aborting the job or even stopping the job before its true end doesnt always comeout with good results, IMHO. Thats why i advised to finish the job by limiting the execution to one row.

Ouch , typo. Thats what i meant, 7.5x2

Try to run the view at the command line and collect its contents in a file. Then do a wc -l on the file to check the number of records. If it is what you are getting via DataStage then the sql for creating the view is too huge to accomodate that amount of records in the database memory. For that you...

First of all you need a unix id, that can access the DataStage server. Your unix admin can take care of that. Then you need to add that to the DataStage group. Even that only the admin can do that.

I think even the row merger can take care of your problem. Look into the row merger stage.

Or you can specifiy the limit as 1 row in the director. You can also specify it in command line.

Double check all the derivations, stage variables and constraints in the transformer. Something in there is causing the problem.
Have you encountered similar problems with any other job?

At this point, Kenneth Bland is one post away from 4k. I, personally, have gotton so much support and encouragement from him. Thanks for helping us carve our path in better understanding ETL concepts, DataStage tips and tricks and in becoming better ETL developers. Congratulations on accomplishing t...

Are you getting this warning for all the records? I mean is your director's log filled with warnings? You need to get the date or timestamp in the correct format. What is your source? Is your source DB2 as well?

Give us some details of your sequence, like how many jobs are being called in parallel, how many stages do those jobs have that are called in parallel?

Even with indexing, you cannot get a enormous boost in the performance if the target stage is set to Insert and update or update and insert. For 12M records, it would be a better to go with Ken's recommendation. A logged activity for this size of data set will be slow.

Welcome Aboard
If you are on DS 7.1 or less then you may not be able to run PX jobs on a windows platform. This was possible with version 7.2 as balajisr noted. You might be able to build and compile it but not execute it.
Regards,

If you can get a list of all the files present in the folder you want to manipulate then you can read it via a sequential file and then use the field function to extract the date and the terminal id. Otherwise you can use write a small script to read in the files and use the power of awk to get the ...

The SQLServer Load stage is not available for unix platform, thats why this workaround of building the flat file, ftp'ing it over to the windows server and then invoking the bulk loader.