DSXchange

Search will be your friend.

How big are the two files that are inputs to the Merge stage? You may have run out of memory trying to accomplish the join.

Did I not specify "trimmed of field mark characters"?

The Admininstrator client does not execute natively in the operating system shell. But you can open one.

No, you can do it with an expression. For example (and this is not the only way):

Your Search capability certainly can.

How? Mind reading? How about you tell us precisely how you set up the properties of your Merge stage, then we might have some idea of what you want to do and how you went about doing it.

I don't know. I can't see vital things like your reference key expression and whether the value(s) being supplied to it are null or not. YOU need to do the detective work. Only you have access to all the details of your job design. That it fails consistently in the same place strongly suggests that ...

Are you trying to do a lookup using a null-valued reference key?

DSD.BCIGetByKey is the internal routine that retrieves a row from an ODBC or UniVerse stage on a reference input link.

You did not mention doing any lookups in your original post in this thread.

Only the obvious one - that you're doing something in step 7 that DataStage doesn't like.

When you reset the job in Director after it fails, what extra information appears in the "from previous run..." event in the log?

You did not mention sparse lookups in your original question. Sparse lookups do have the restriction of one reference input link. A Lookup stage can support any (reasonable) number of reference input links performing normal lookups, but only one lookup if it is performing a sparse lookup. A normal l...

There is such a document, a thoroughly comprehensive one. It is available for the asking from your support provider. It is NOT available from DSXchange so please don't ask.

The reason you don't know which stage is your own fault; you are allowing the operators to combine (to execute in the same process). For diagnosis at least you do not want this to occur. This particular message results from the way in which decimal data are stored internally, and the first half-byte...

If the data are sorted on the grouping keys then on the column for which you need the maximum value per group, you could also use a Remove Duplicates stage.

Mainframe jobs actually have fewer options than server jobs. If I were you I'd track down one of the DataStage experts in IBM Australia for a chat.