DSXchange

-99 is "general Repostiory error" which means that DataStage can not diagnose it. Chances are that the pipe has filled as some point - the writer to the pipe running faster than the reader from the pipe, thus causing a "broken pipe" condition. Do the individual job logs contain any more useful infor...

... with the "=" character quoted, of course!

See what I mean? A specification makes it clearer what your intent is. A full specification would help you (though not us) even more. Is the database Oracle or DB2? If so, you can do this using a sparse lookup in combination with user-defined SQL. Or you might pass the ten key values to a stored pro...

PPK might be price per kilogram or a style of pistol. SKU typically means stock keeping unit. You don't really need to know what they mean, Maveric.

Probably all the extra work that has to be done to verify the identity of the domain user that is mapped to the DataStage user. Are your servers on separate machines? Are they or the network busy?

Yes there is: APT_TSORT_STRESS_BLOCKSIZE
Minimum is 1MB.

I think what you're seeking to do is a join, best accomplished with a Join stage. But I'm not sure, which I why I suggested that a specification is the first thing you need - I don't think you're sure either.

You need to declare them as Key on the DB2 link. "Key" in this context means "lookup key" rather than "primary key".

Build stage probably won't help all that much. What you need first is a specification. You can generate the computed comparison value in an upstream Transformer stage, before attempting the lookup/join/merge. Is the comparison against one column in every row, or against more than one column? In the ...

Substrings() is not the preferred mechanism in server jobs, as it includes code for dealing with multi-valued data. The square bracket notation is the preferred method in server jobs, as it is somewhat more efficient. Beware, in parallel jobs, that the first ("start") argument for substring is one-b...

Import the table definition from the sequential file. Fiddle about on the Format tab until Preview shows the correct data. Then tidy up column names, data types, descriptions on the Define tab. In the Sequential File stage, set "First Line is Column Names" property correctly. On the Format tab click...

Arabic script isn't complex.

Thank you for alerting us to a potential bug.
What did your support provider advise?

You're out of luck. All you can do is to import the Sequential File definition.
RCP, after all, makes columns appear "by magic". You do not have control of metadata, and can perforce not undertake usage or lineage analyses.

Use Continue on the first, make the second lookup conditional on one of the returned columns from the first being NULL, and use Drop on the second. Or use Continue on the second and filter out the NULLs downstream if you need to count them.