DSXchange

In one of our project , we dont use server job. Why not? You have it. If you own a set of screwdrivers, do you elect not to use some of them on some ideological basis? If you really must use a parallel job - and I again state that there is no good reason to here - then you can construct an appropri...

Welcome aboard. :D

Please verify that you want to perform this task using DataStage TX (transformation extender), which we have assumed since you have posted in the DataStage TX forum.

Otherwise please post the question in the Parallel forum.

Probably the easiest is to override the ?F&F pattern. But, if you want to generate two records from one, then you will need to add an extra rule to your NAME rule set.

Show us what you want to do and the Modify stage specification that you used to try to achieve it. You need the lookup table as part of the specification. For example: lookup_ustring_from_int32[{defaultstring="unknown"}(1=Cust1;2=Cust2;3=Cust3)](cust_numb)

Check that both of the child jobs are compiled. Use Status view in Director.

Did you make sure that the Oracle 32-bit libraries are referenced ahead of the Oracle 64-bit libraries in the shared library search path used by DataStage processes - that is, the LD_LIBRARY_PATH (or SHLIB_PATH or LIBPATH) environment variable set in the dsenv script?

You can use a UV stage to extract the required records from the DS_JOBS table (the short descriptions ARE stored there, as well as in DS_JOBOBJECTS). The advantage of using DS_JOBS is that the relevant column names all exist in the metadata. Importing them into DataStage, however, requires a "trick"...

I disagree. An ETL developer need have no interaction with a data warehouse at all. It is also possible to create and populate a data warehouse without using any ETL (for example by using an ELT approach).

Yes, simply invoke it as a function. However, you will need a DEFFUN declaration to inform the compiler that the name followed by a left parenthesis is that of a function rather than that of a dimensioned array.

Why not use a server job?

Clearly your impression does not hold for parallel jobs. (It is correct, as far as I am aware, for server jobs.)

Did the export include the job executables and job source code? If not, then you will need to recompile the job designs in the new project. It's probably a good idea to do so in any case.

You may find it easier to search-and-replace the ReadOnly property in the export file. Change it from 1 to 0.

DataStage projects are supposed to be totally self-contained. There is no mechanism for creating global Transforms.

There is a mechanism for creating global Routines, but this mechanism will cease to be valid in the next ("Hawk") release.

Resizing the dictionary portion of the hashed file for 64-bit addressing is a complete waste of time. Yet this is what you've done. The file dictionary, in which metadata is stored, will never get anywhere near 2GB. You have to resize the data portion of the hashed file. It is the data portion that ...