DSXchange

What UNIX are you on? If you post that and someone here in DSXChange is on the same OS they can check to see if the specified library file is present on their system.

Please give explicit examples and make sure that you don't have the most common cause of this type of problem - embedded leading and trailing spaces.

Good point. If the source is a single node and running just one process, then ordering will be maintained. But if the source is one node with multiple readers (think along the lines of a fixed width sequential input file) then ordering might indeed not be in ascending order; and if one is repartitio...

Sure you can create your own library file, as the OP proved. But don't expect it to contain library-like information :P tehavele - I am not on a UNIX system to check, but try the following from the command line (after deleting your dummy library) cd /opt/IBM/InformationServer find . -name libqabwvcd...

If you go to your source stage, double-click on the number at the left of the column or right-mouse click and "edit row" and then see if that column might have a default value set.

Since the utility calls orchadmin to get the details, the error would seem to lie in the utility and I'd classify that as a bug.

Yes. If you are using default UNIX sequential file terminators, putting a CHAR(10) in your job will add a line feed.

The :1 and :2 refer to column numbers - I can't see how those could have changed between versions.

Add a peek output stage in addition to your Oracle stage. Identify a row that isn't working correctly, and see what the contents are in the Peek stage. Is the string too large? Nulls in other columns? Key constraints in the Database?

The STAGECOM.ARR values are not documented and, as Ray has pointed out, there is no official way to do this. Manipulating the STAGECOM.ARR takes some knowledge of the internals of DataStage. I would go the path that Sainath has recommended and use explicit calls from your transform stage to the send...

The difference is that you have quite a few more options with the explicit sort stage as opposed to the implicit sort stage; they call up the same sort by default.

orchadmin ll {dataset} from the command line will show you the Kilobytes per segment, the number of records per segment is not obtainable. Since segments cannot be individually processed the record count is a moot point.

What is your job design? When I put a PEEK in my job and view the schema I don't see this arrangement.

1. Login to unix shell and check that "/opt/IBM/InformationServer/Server/DSEngine/QAS/lib/libqabwvcd.sl" exists and has rx access to DataStage users. If it is not there, try to find it. (i.e. from /opt/IBM try "find . -name libqabwvcd.sl -print") and add that directory to your LD...

What is your output stage type which is triggering this error?