DSXchange

SFTP allows shelling out on the local host and executing commands. Thus, if your sftp is "push" then it will work, if you "pull" the files then you will need to do something else, perhaps "rsh" or "ssh" to execute the delete command.

Try to put a dummy copy stage between the reject link from the lookup and the input link to the BASIC stage, this transforms the link type to a "normal" one.

Yes, you can declare multiple record types in a CFF stage. I am not at a DataStage client now, but you can declare various 01-level type records in the "Record" tab of the CFF stage.

What happened is that the job is trying to fit the long timestamp value string into a Varchar(8) and it doesn't fit, thus the error message. You have gotten your columns mixed up - either incorrect order or incorrect number of columns.

If the error is not reproduceable if you run with a single-node configuration, or repartition to your database key column, or commit each row then you might hazard a guess that you have a database deadlock issue at the heart of the problem.

If you turn on APT_DUMP_SCORE and look at the generated scores do you see any repartitioning being performed?

I would turn on echoing commands and output in your shell script. There are various means to do this, search Google for "debugging shell script {YourUNIXFlavor}" to see which options are open to you on your platform.

Does you BASIC transform stage in a PX job have one input and one output? No reference inputs are allowed in the BASIC transform stage, unlike the normal server transform stage.

Because before-job and after-job routines are not called from transform stages. PX Transform stages cannot call Server functions. You can use a BASIC transform stage and use Server functions in Parallel jobs; but this is generally not recommended unless it is absolutely unavoidable.

From the command line execute "dsjob -ljobs MYPROJECT". This is documented in the Programmer's Guide (page 4 at Version 8.1)

It is not so simple as you think. You can quickly put together a "hello world" type of buildop by creating a "Build" type stage in the designer and having the job auto-read and auto-write links and then put your logic in the "per-record" section. Compile the customer st...

You need to realize that a floating point number cannot represent each possible real number; internally the number is stored as a mantissa and exponent value which approaches the actual number but may not be equal to it. In this case the "0.375" and "2.803" values have been conve...

The analogy used for the original Orchestrate product is an orchestra with a conductor, section leaders (i.e. the first wind or string) and players. In music, the "score" is the book of notes that each instrument player uses. Thus the "score" in a PX job is the script which is ex...

Yes, re-running might help to diagnose the problem. Are you running on a 1-node configuration file or more? Are you inserting or updating (or both)?

I'm not sure what "to nearest zero" implies, but can you not use a type conversion function such as "DecimalToDecimal" to achieve your desired result?