DSXchange

The file doenot have straight f0rward columns and rows. Its a file like a normal end report with scattered data. You've got your work cut out for you then. Not being in .xls or a flavor of .csv or .prn format means you would need to read each record as a single large varchar field using the Sequent...

You'd need to be on a 7.5.x version of DataStage to have those stages. If you don't have them, you are stuck with writing your own 'batch' job control and controling the looping yourself, as already noted.

Try using a single stage for your two Oracle links. This is a valid solution for a Server job and wondering if it will solve the same transactional issues in a PX job as well.

If not, you should get your DBA involved and have them trace what is going on during the run of the job.

For recent versions of Oracle, check out the DBMS_LOB package. I've found it helps with handling data of this nature more than using a large (or Long) Varchar datatype by itself does.

I've seen this in Server as well, but don't recall the gory details of exactly when it happens. Most of the time a link rename handles all occurances of it in the Transformer but occassionally it doesn't. It seems like constraints are the biggest place where a link rename doesn't happen. The compile...

guggi wrote:How to supply job parameter here?

You can't - use the Parameter in a column's derivation and send that to the query using the appropriate marker.

I'll be really surprised if you can make this work if this is all in one 'link'. Why not use two if that's the case?

incognico wrote:I can't use the funtion date() in a ODBC stage for my selection in my clause where.

That's true - you got the 'with DataStage' answer that I thought you were looking for. Nothing changes in your SQL query just because you are running it in a DataStage job, everything still works the same there.

Known 'bug' in certain versions and platforms. Upgrading to the lastest version - 7.5.1A or 7.5.2 - should correct it for you.

Otherwise the work-around is to write a Routine to do the same thing via ExecSH... or ExecDOS on Windows, I suppose.

And REPLICATE does what, exactly? If you are padding the output field with zeroes, use the FMT function for that. There are various date related functions in DataStage, check Date() in the online help as one example. From what I recall, it returns the system date in internal format, which means you ...

No problem... told ya that was the most important step.

You need the Windows Service Pack 2 (SP2) patch.

Lack of quotes, actually. Of course, you could just try it and see... worst case you'll get a different error.

vardhan354 wrote:i tried with this dsjob -lognewest projectname jobname it is giving a message newest id = 50, statuscode = 0.

That's what it is supposed to return.

Sometimes an uninstall doesn't uninstall enough. As long as your machine does not host the Server install as well, try: 1) Uninstall the DataStage client 2) Remove any leftover Ascential directories 3) Remove any Ascential Registry entries using RegEdit 4) Reboot 5) Reinstall the client Seem like #3...

Assuming you just have access to Server jobs, you have the XE or Extended Edition of the product. You would need to upgrade to EE or the Enterprise Edition to have access to Parallel jobs as well. Many monies would need to exchange hands.