DSXchange

Please clarify the rule you are using. For example, is the the year of the Wednesday, is it the year containing the larger number of days from the overlap week, is it something else? Doing this will probably guide your logic in the Transformer stage.

Another way is to compile the job in trace mode, which then gives the ability to force all stages (operators) to execute in sequential mode. When you're done, re-compile not in trace mode.

It wasn't an upgrade - it was a new install.

Right click in the background area of the Monitor's display and choose "Show Instances" from the menu.

Hard coding for each particular N is your only option in parallel job.

You can simply put it into the Administrator client manually. In the User-Defined environment variables section is the usual location.

Check the "restartable" check box in the sequence compiler options (in job properties) and re-compile the sequence. Are you comfortable with what that will and won't give you?

No, you don't need 100 extracts. For 95 tables each with different metadata, 95 jobs should do it.

Not until you get version 8.0 or later.

You must have a commit interval of 1 row/transaction (that is, every row sent to the target table must be committed immediately). You must also specify an array size of 1 and disable row buffering. An alternative would be to use a hashed file instead of the target table. In this case you would need ...

Never was. Was a Primate.

How are your data partitioned? How many nodes in your configuration file?

But one thing i cannot understand is by default "waiting for job" is directed to std error by dsjob command. thats why we are redirecting the erro to output and then to log. Why the dsjob command sending the "waiting for job" to std error.Why it is not direclty sending the "...

Provided that the record structure is identical, you could use one job with the table/file name a job parameter.