DSXchange

You enable it the same way that you debug the dsrpcd - by directing its outputs (stdout and stderr) to somewhere other than /dev/null. You can control the amount of warning via the debug-level switch (most verbose is -d9). There is a noticeable impact in DataStage clients' response times, and startu...

No. The Sort stage does not read from disk. It requires an input link.
You can, however, specify the UNIX sort command as a filter in a Sequential File stage.

It could be one of those horrible bit-wise storage formats, where a particular five bits contains the day number (1-31), another four bits contains the month number (1-12), and so on for year, hour, minute, second and, perhaps, millisecond. You really do need to get the doco, because there's no guar...

I think by now you are convinced that it's an Oracle problem, not a problem with accessing flat files.

Hi Craig, Thanks for the tip, we have now solved our problem: in the dsenv template the NLS_LANG parameter was not defined, nor did the DS manual mention it. A bunch of other parameters were mentioned, and I had set them all to our NLS settings. But adding the NLS_LANG parameter to the dsenv file d...

It is not a configuration file - rather there is an NLS database. You configure it from the Administrator client - NLS settings system wide for character maps to be built and loaded into shared memory, and NLS per-project settings.

Do you have NLS (national language support) enabled? Without NLS the accented characters should be able to be handled satisfactorily. I assume that the database is storing this column as NVarChar of some kind. Do you have your Oracle language environment variables (such as NLS_LANG) correctly set?

For only 10 rows to be processed you may as well perform lookups directly against the target table, unless you are constructing a design that is staging data (which is not apparent from your design). Or, since you know these 10 rows, just preload their key values into the hashed file. For so few row...

Your text specifies "reading data from sortstage to sequential stage" but your design shows something different.
Does the text file contain an empty line at the end? (I vaguely recall that -100 is the "end of data" code.)

Is the Oracle client propertly set up and configured on the machine whose fastname is mentioned in the configuration file for node1? In particular is there a tnsnames.ora file there that refers to the correct Oracle instance(s)?

Once disconnected no record remains of a connected client. You can script your own, or you can enable logging in the dsrpcd process.

As with all variable settings, the most recent has effect. For DataStage processes the setting made in dsenv should be the most recent, unless overridden within the job design itself. The second message of a job run in the log contains all environment variable settings - did you check there to see w...

The error message DOES give you more information. "Abnormal termination of stage <Job Name>..< Transformer Name> detected". (emphasis mine) Look for a problem with the Transformer stage, rather than jumping to the conclusion that the Aggregator stage is at fault. In another thread on exactly this pr...

What makes you sure that this is the culprit? Could it be some other Char, Varcher or Numeric column that lacks any precision setting, or has an invalid precision setting, that might be the cause? Certainly the cobination VarChar and 1500 is perfectly valid.

You did not check well enough. You (that is, the user ID under which DataStage connects to Oracle) lacks appropriate permission to sys.dba_extents. You need to be able to SELECT from this system table. Arrange with your Oracle DBA to get such privilege granted.