DSXchange

Almost certainly, and it happened to occur on the second node in your configuration file.

Disable operator combination (set APT_DISABLE_COMBINATION) and re-run to determine in precisely which stage the error is being generated.

Of course you can use a hashed file. Seven million is not so many rows. You may even be able to fit it in a standard hashed file (up to 2GB of data). Be selective about what columns you load into the hashed file and what rows you load into the hashed file - only load the ones you need. If you still ...

You must use identical directory pathname in both Hashed File stages, and you must use the identical hashed file name in both Hashed File stages, irrespective of the link names.

Correct terminology is "hashed" file, not "hash" file.

This is not a DataStage issue - it's the time required by Informix to effect the GROUP BY. Until all the rows are processed (so that all the groups can be determined) Informix can not deliver any rows. Yet the clock is still running, so rows/sec appears low. It is not a useful metric. What else is h...

Maybe the fact that you need a small contribution to get full technical information could be used to justify the expense to the company?

If this is a user requirement - or a PHB requirement - just yell in their face that a primary key is, by definition, NOT NULL.

Real dimensional models in the business world seem to contain between 4 and 15 dimensions. Models with only 2 or 3 dimensions are rare, and often lend the suspicion that more dimensions should be added to the design, if only the designer would stop and think. Models with 20 or more dimensions, for ...

That is not an innate capacity of the job sequence, in which "failed" means only "aborted". You could design a job sequence that explicitly tested the exit status of J2, and decided whether to re-run it. This could be done with a Nested Condition activity. J2 would need to be marked "do not checkpoi...

Find out from whom your company purchased DataStage. Most (all?) resellers have a contractual obligation to provide first level support.

Not only is a job sequence a special case of a server job, the message logging mechanism is different. In server jobs, every process in the job looks after logging its own messages; in parallel jobs all messages are logged by the conductor process. So there can be a single point of interception.

Push back on insane requirements.

There is NO WAY to create a Data Set other than using a Data Set stage.

Hey , Use the below command : when to save ur file in Unix do this : :%s/ctrlv+m// means colon then % ,then s then / ,then control v and controll m ,then double slash enjoy Grammatical Note Ur was a city in ancient Babylon. Please use correct English pronouns such as "your" here. Would you use "ur"...

Could the rejects link from the first CFF stage be directed into a second CFF stage (or even a Column Import stage) to handle the other record type in the same job?

Welcome aboard. :D You can use Fmt(InLink.TheString, "25T") The "T" format is "text" - it breaks on whole words. The Fmt() function inserts a "text mark" (@TM) at each division - you can convert this to anything you require, for example a newline, using Convert() for single character or Ereplace() f...

Did you write the value using a UniVerse stage, or using a Hashed File stage? Writing using a Hashed File stage and reading using a UniVerse stage can cause the scale to be interpreted as a "scaling factor". UniVerse tables treat Numeric and Decimal identically, but for some reason always throw this...