DSXchange

Hello -

This error can occur every few days or every few weeks, it's random. The design is as follows:

Oracle > Transformer > delete/create hashed file A
Oracle > Transformer > Reference hashed file A

When the error occurs, it always occurs on the Reference link. DataStage Tech support tells me the 1st thread, delete/create, will always run 1st followed by the Reference link and this should not occur. Another engineer suggested changing the create/delete to a clear. Another suggested creating the file in the project or default directory. I can do all this, but am not sure this will fix the problem. Today I read about MFILES 50 and T30FILE 200 configuration parameters in the uvconfig file not being large enough and thought perhaps this might also be related.

Any help is appreciated ... Thanks

Looks like a timing issue, with the first write to the hashed file happening during the delete and causing your error message. Can you redesign the job so that your reference link comes from the same hashed file stage object as you are writing to? That will ensure that you don't get timing issues.

The error does not happen on the write, it happens on the read.

ORACLE> Transformer2
/\ (Ref link always works)
ORACLE >Transformer1 > Hashed File A
\/ (Ref link sometimes fails)
ORACLE > Transformer3

The middle thread deletes/creates hashed file A, then the transformer2 and transformer3 reference it. The Ref link to transformer3 sometimes fails. When this job fails, I just reset it and rerun it and it works.

Just split your job up into 2 jobs and you won't have the timing issue. Your job control just coordinates the efforts, plus your design will be more modular to boot. :D