DSXchange

---> StartLoop <--------------------- EndLoop | ^ | | V | DoStuff --------> Test -----------+ | >0 | = 0 V DoNothing

dsjob -jobinfo output includes a field called "controller". This is typically the parent sequence from which the job was invoked. Therefore, when your Ctrl-M script uses dsjob to retrieve information about status, also select the controller from that result.

Your BASIC Transformer stage does not have a reference input. However, your Lookup stage does. Investigate that.

I don't think you can do that in a multi-pass single match. I suspect you're going to need more than one match stage, maybe even in separate jobs. And then, of course, you will have the dilemma of combining the sets of records - you may need to introduce an artificial key to the source stream.

Are you running these in the parallel engine or the server engine? If parallel, are the configuration files the same (contain the same number of nodes)?

First you need to find out how your Chinese characters are encoded (for example GB2312 or BIG5), then how these are mapped to Unicode. Consult Unicode code tables to determine which code points represent Chinese characters (don't forget the CJK characters, which are shared between Chinese, Japanese ...

What additional data we will get into our logs, if we add this environmental variable to our job? One additional entry, containing the score. How this variable help us to performance tuning? It shows you the actual job that gets executed, which may be quite different from what is designed. With suf...

It IS correct. The even numbers are on the other node.

That's what I meant with "mark as XML". Too used to it, I guess.

Converting is not the same thing as renaming.

Export the entire project.

Every operating system version of DataStage is different. You install on AIX, then export from Solaris and import on AIX. There's no alternative.

Make sure you pick up other things like any hashed files, scripts and the like that have to be migrated.

Everything additional you do must necessarily degrade performance. Do only that which you must do.

Before-job subroutines and after-job subroutines in parallel jobs are server routines.

Routine activities in sequences invoke server routines.

Not until you post in the correct forum so that "some one" can know what version, what edition, what platform this problem is occurring on.