DSXchange

Sorry, not April 1st over here quite yet.

I'm assuming that the problem only materializes when run from the Sequence, yes? Meaning, if you run the job standalone, it can make use of the parametrized password value without an error. Correct? If that's the case, then it sounds like you have a problem with how you defined the value to be passe...

I don't see a problem with option 1, nor a need for any 'looping'. A routine can be used to retrieve the values and called twice in the Sequence, once for each value. Or a UserVariables stage could pull both values, then either could pass them as job parameters to Job2. Neither difficult nor complic...

I never could, really. :( It happened with two jobs of almost identical design and once it started we never found a way to correct it, and "we" included both IBM and I. I ended up doing what I used to do before Sequence jobs were available - I hand-coded the equivalent looping process and ...

Interesting... let us know after you get the patch if that fixes it for you.

You really need to take that error back to your DBA and see if they can help, perhaps by changing some settings. They may also be able to recommend some "alter session" statements that might help, you can use those in the before-sql tab.

The first light can be shed by searching for that error here. For example, I got over 120 matches for "SIGSEGV" when I searched. You could also Google for SIGSEGV and find a wealth of information out there on the subject of "segmentation violations".

Are you saying that your organization does not have an DB2 licenses itself but has the need to access DB2 on 'remote' databases? In that case you may very well need to purchase DB2, I'd be surprised if the client itself could be obtained separately and without cost, but who knows.

Short answer - no. Only one repeating / key element. That's not to say you can't generate these as 'chunks' and then assemble the chunks into the complete output format that you need after that. The Best Practices document uploaded at Kim Duke's website details this technique.

Even the join-key in not correct reference value will be null right? Not necessarily. "Not correct" could mean any number of records are returned, including the cartesian product you seem to be getting or none / null output if no keys match. To me it sounds like at least one key is matchi...

If you are truly getting a "cartesian product" and that's not what you want, then it sounds like your join keys are not properly set up.

The short answer is "no". The long answer would involve creating your own checkpointing system or building in alternate entry points into the job flow rather than using the existing checkpoint functionality.

Why not check to ensure there's data to load before invoking this process? It doesn't seem like a "bug" per se to me, the stage utilizes a named pipe to feed sqlldr and one of the two of them doesn't like the fact that there's no data to process.

'Hide' as in ensure people can't even see that it exists? I don't believe so. Would restricting access to it not be a valid alternative?

(and he did)