DSXchange

Your APT_COMPILEOPT should be "-O -c" and APT_LINKOPT should be "-G". Can you execute the compile from the command line? I also see that the error refers to "munch" instead of directly to the compiler.

Please post the "from previous run" entry.

Welcome to the trials and tribulations of BASIC transform stages in PX jobs. We've had similar issues and are working towards getting rid of this stage as it occasionally has problems that require manual fixing. What has probably happened is that your process has gotten hung up. Check with "ps ...

1. These are essentially "T"d output file from the main background (PHantom) process 2. No, the director does not use these files, except when a "reset" is done in which case the file contents are copied to the log entry titled "from previous run" 3. Yes, they are tempo...

If you disable operator combination, which stage is the one generating the message - the input or output one?

More likely that your default date formats are different and that your job is using those defaults.

This probably means that while you have used MS-1256 within DataStage, the Oracle Database or output table has another character set defined which cannot hold some character that is available in MS-1256. What are your Oracle NLS settings (check in Oracle, not DataStage)

It is difficult, if not impossible, to help with the data given. Your job reads a dataset located at "home/dsadm/Ascential/DataStage/DataSets/eng/Ord" which does not exist. That would lead one to assume that something went wrong with Job 1; or that Job 1 wrote to a dataset, but one located...

...At the same time, we enabled "Server Jobs" on this installation (using IBM tech engineer)... There is no way to disable server jobs, so I am not sure what this step actually involved. The same question applies to the "Oci9 package" installation - does this mean you installed ...

Reset the job and look at the "from previous run..." entry.

I've had this error before. It happens when another userid is used to access DataStage clients than the one who used the the first time and when permissions on the PC are set to be restrictive. Login with administrator rights to the PC and set permissions on the XML file.

Unless I am missing something from the description, this looks like a straightforward lookup stage application.

If the codes come from a flat file, perhaps you could arrange a pseudo-join 1:1 in your job to assign those values to your data.

Why not use @OUTROWNUM combined with @PARTITIONNUM to get a unique value (plus setting the transform to run in parallel)?

That approach was often used in earlier versions of DataStage, where the parameters could be changed from within the invoked job. This is no longer possible, and attempt to change the value of an input parameter results in the job aborting.