DSXchange

What's the actual warning message?

The Hash aggregation method is appropriate when there aren't too many distinct grouping values - up to, say, 1000 per partition. This method does not require sorted input, so you could lose the Sort stage with no detriment. If you are sorting for some other reason, change the aggregation method to S...

Hi Ray , We could force to run the lookup stage run sequentially , which would avoid any issues due to parallelism . let me know your comments on that Then you may as well use a server job, which has much lower overheads than a parallel job. But you'd STILL be doing the lookup for every row process...

A cheaper solution is to have a CSV or TXT dump of the worksheet sent from the original Excel workbook (on Windows) to the UNIX machine, rather than the workbook (XLS file) itself. You could write a Windows command (in Visual Basic, for example) to invoke Excel and dump the worksheet(s) then FTP or ...

When you use the -jobstatus option, the exit status of dsjob is the exit status of the job that was run. You need to know that for the efficacy of Craig's solution to be explained.

In the Administrator client, select your project from the Projects tab. There you will find an NLS button where you can select the project default map (which, of course, can be overridden at the job level). You ought not, but may, need to restart DataStage services for this change to be picked up.

It is impossible to help you without knowing exactly what error message you are getting. Otherwise your question is like "my car won't start - what's wrong with it?".

Triggers don't fire per-commit, they fire per-row. The syntax of the CREATE TRIGGER statement tells you that.

The COMMIT only determines when the insert actually occurs, in your case after all rows have been sent to the database server.

Metadata are - deliberately - incompletely described in the Repository. Further, import of the table definitions from the Repository tables is deliberately blocked. You would need to create synonyms and import the definitions from those. You may like to review [urlhttp://www.dsxchange.com/viewtopic....

videsh77 wrote:I am not using sort / hash method in the aggregator.

Yes you are. These are the only choices.

Without knowing which method you're using, it is fatuous to offer any suggestions.

If you don't require any specific maps, why did you install DataStage with NLS? It goes faster without.

The problem with using a stage variable, or even a lookup, is that you are getting information on each processing node. This may turn out to be tricky. Better would be to use a separate job, executing sequentially (or a server job), that extracts the max value once from Oracle and stores it somewher...

Sorry about that. My notes were a little jumbled! BMC Patrol is used by the operations people to monitor the systems, not for scheduling. Active Batch, as you say, is the scheduling tool. Currently.

The command output includes line terminators. These are translated by DataStage into field marks (@FM) so that what you get is a dynamic array containing one field per line of output. The angle brackets notation with embedded "1" returns field #1. You could also have specified Field(ExecCommand....

OK, maybe if I were being paid by the hour...