DSXchange

Put some echo commands in the top of the script so you can assess what values were passed via the command line arguments.

After you have scheduled the job, open a CMD shell and type the command AT. Inspect the queue. Is there one job reported there, or two? If only one, it's a problem with how Director is reading the queue; if two it's a problem with how Director is interacting with the queue. Once we know which it is,...

Try it without the spaces around the "=" sign. It does actually make a difference. And, since you're on Windows, use the set command.

That's the information I was trying to remember when I kept going on about system tables. Clearly it's important. But when one is in an internet cafe paying for connection time one does not really want to spend it searching.

QA? Wozzat den?

That's useful to know, thanks for posting it.

Obvious in hindsight but, then, what isn't?

Visio wouldn't help with that, either.

You don't need Visio. The job design itself shows the flow graphically. Choose the documentation icon from the Designer toolbar (second from the right) to produce job documentation. Similarly, the job sequence design shows the flow of control graphically. Build the documentation as above and edit th...

Restarting DataStage will clear all locks, but don't treat this as a magic cure-all solution. Keep it in reserve for (a) when you've tried everything else, and (b) when nothing is happening in DataStage.

Thank you for the testimonial.

All you need to do now is to post your question on the forum for questions on parallel jobs and you will be one step closer to getting an answer.

With those commands you can only export entire projects. Search the forum for exporting individual jobs. You can then script that with an items list.

The "?" in "OCLI/DR GERALD D?fAV" indicates the actual character that could not be mapped. You may like to check your source to verify that that character is properly encoded. You may require a hex editor for this task.

"Drop" causes any input row for which the lookup fails not to appear on the output of the Lookup stage. By discarding records in this fashion you are eliminating any chance of performing row count reconciliation. Prefer a rejects link from the Lookup stage, even if it appends to /dev/null. At least ...

Use a rejects link from the Sequential File stage to capture the header and trailer row (as raw strings) into a separate data stream. You can process them separately or just ignore them.

The column derivations become the SELECT clause of the query that is generated. That's why you can have DISTINCT, SUM(), expressions and so on in the derivations.