DSXchange

So you can view the data on your PC correctly. What is your PC character and how are you copying the file to your UNIX system? Is it a binary FTP? If so, use your PC Character set definition on the UNIX machine.

1. What system is the data on? Windows, UNIX? What Character set is defined on the system where you can view the data?

2. If you use a binary transfer from your system, then you need to use the same character set in DataStage as on your system.

I am afraid that is covered in the documentation. You need to create a system DSN ODBC connection to your excel sheet (this is not part of DataStage but of Windows). Once you have a functioning ODBC connection you can bind that into your job in DataStage and write to the Excel sheet as if it were a ...

Have you checked the "dsenv" file, it is set there or it calls a DB2 script which might set it.

The only way that you can generate in true .xls format is if you use an ODBC system DSN to write to Excel and not use a flat-file format.

Windows tries to use MS-EXCEL to open up .xls files by default. I think if you make a comma-separted file with a .xls suffix that might work.

just tested it on my workstation and it didn't work.

You cannot sort in the join stage, unless you edit in the input links and do it there, but it is preferable to do so in an explicit sort stage since you can opt to not sort certain columns there.

You are probably using a .xls extension. Change that to .txt and your windows will be happy.

API Stage1 & API Stage2
Since you are re-partitioning your sorted data it is no longer sorted, so you need to make that a "real" sort on your join criteria.

How about sorting the data on your key criteria, adding a cluster key change column (1 for first duplicate, 0 for subsequent ones), then sort again but add the change column, then a transform stage using a stage variable to store previous record key data. If change column = 1 and previous record is ...

Actually, more like:

There is a perfectly good "join" stage in PX which can do this for you.

The OSH looks normal. Can you add $APT_STARTUP_STATUS, APT_PM_PLAYER_TIMING, APT_PM_SHOW_PIDS to your job? With the PIDs you can use "truss" to see which process might be doing more than just issuing calls to sleep?

What sort of tables are they? Are the numbers closer together if you compute the number of bytes/sec instead of rows/sec?

At Version 8 you can configure DataStage to use a specific user, check your web IS admin tool to see what your settings are. The log files are stored in the project directory, they are named RT_LOGnnn where nnn is number.