Phantom errors and Hashed Files

[tracy]

I've got a job that runs fine in our production environment and obviously worked fine in our development environment (before we moved it to production).

Now, when we go back to dev to run the job for the first time in a couple of months, we get the following errors:

ADPDS_IMPACT_FINTAX..IMPACT_FINTAX_SUBPROJ_ID.in_fintax_subproject_id: ds_uvput() - Write failed for record id '<Missing or NULL>'

DataStage Job 1054 Phantom 5544
Program "DSD.StageRun": Line 626, Attempted WRITE with NULL record ID
DataStage Phantom Finished

From the errors, it seems like there are NULLs that are causing the problem, but I've checked the data and it isn't true.

The job is pretty simple:
Text File --> Transformer --> Oracle Table.

Additionally, there's a "lookup" that takes data from an Oracle table to a Hashed File and then gets passed to the above Transformer stage.

When I run the job in Designer so that I can see the pretty colors, I see 100 records get passed from the Oracle table to the Hashed File, but 0 records go from the Hashed File to the Transformer Stage.

I really don't understand what's going on. It happened once before to me when I was developing and my solution was just to start over from scratch. In this case, it's a lot of jobs that it's happening to, so starting from scratch isn't an option.

Has anybody seen this before?

[chulett]

From the errors, it seems like there are NULLs that are causing the problem, but I've checked the data and it isn't true.

Yes, it *is* true or the error wouldn't have been logged.

When I run the job in Designer so that I can see the pretty colors, I see 100 records get passed from the Oracle table to the Hashed File, but 0 records go from the Hashed File to the Transformer Stage.

Which, if I follow your directions of who went where, means all of your lookups failed. When that happens, the result of the lookup is NULL and unless you are checking and blocking that, I would wager that is the source of your error.