DSXchange

Did you actually run the file with a 1-node configuration?

data is of type varchar(50) ex: in seqfile : "GATEWAY NEW BUSINESS 14.9" while viewing from unix : "GATEWAY NEW BUSINESS 14.9 (I used code mark so that the spaces display) What tool are you using to "view" the data? If you read the file in DataStage and output a LEN(In.Stri...

Turning on RCP doesn't change the behaviour that 2 columns with the same name exist but only one can be output.

What happens to the compile if you change the link name - does it remain the same? Also, is RCP enable for the link coming into the transform stage?

A couple of years back I did some performance testing between lookup filesets and dataset and, as a result of that testing, have not used lookup filesets since so I am not the one to answer these specfic detailed questions (unless you opted to try a normal dataset).

Open up the repository tree structure and drill down unter "Stages" -> Parallel -> Processing -> BASIC Transformer and drag it across to your PARALLEL JOB canvas.

This out of space error refers not to disk space, but to available memory for the lookup file set. Try to reduce the memory needed in the reference data: - remove unnecessary columns - declare all string as unterminated ones, removing the length definitions; since a value of "Hello" in a V...

Kris - that is a good analysis of the problem and makes sense for the example given, which presupposes that every new record has an effective "EndTS" of HIGH-DATE. If that is not the case then the logic gets complex beyond the scope of stage variables.

We have only seen it intermittently as well. Conjecture was that it occurs somehow when doing a "save as" or perhaps when cut-and-pasting the stage from one job to another.

I am a bit confused. Could you please post the actual warning message?

Hmmm... I don't see how you could still have the error if you did what Ray suggested, since you wouldn't be compiling the job on production.

Does the job run through if you change your commit size to 1 and/or if you run the job sequentially or with 1 node?

What sort of "logs" are created with the lookup removed from your job? Was the lookup to one of the databases? What is your actual question or problem?

p.s. The "FAQ" discussion is for frequently asked questions.

Because you can get multiple updates in one run, a transform stage with stage variables isn't sufficient. One input can generate 2 outputs, the next input might have an overlapping period on the 2 records and thus involve 2 update on the existing records (to terminate them), and two inserts for the ...

From your example it is unclear what you want to achieve. Is it that for each input record you want to make exactly 3 output records, one for 2008, 2009 and 2010 respectively?

If so, why not a transform stage with 3 outputs (one for each year) then a funnel stage?