DSXchange

This looks an awful lot like an interview question and we're not in the habit of answering questions of that nature here.

Why don't you take a stab at answering the question? How many different ways have you seen so far?

So... by 'crontab' I'm guessing you don't mean the Director client and if so, why not? You need a wrapper script for dsjob to set up the environment properly, it generally won't run well all on its own.

adityavarma wrote:My Mistake, I have placed it in the server forum, it should be in parallel forum, dont know how to move it now.

You wait for the Moderator to stumble by.

Me too!

So... three posts, all of them interview questions?

Any 'performance implications' should be for the positive if they are leveraged properly. There isn't in this case, either way, it just doesn't show a good understanding of how to leverage them. Typical use would be for derivations that you need to leverage multiple times in the stage - derive it on...

Typically, create three files and then cat them together in the proper order after job. That or build the proper output in three streams and funnel them together at the end.

Search.

Already addressed that issue. Type it in yourself, do not copy/paste from the post that suggested it... as that "dash" and your dash are different characters.

Only if you know it will always end with warnings, now a clean run without warnings will not trigger the following step. Perhaps both conditions should be used?

Assuming 'Tax_Rate_area' is the correct field, check for spaces or other vaules that would cause the check to fail. Perhaps a trim is in order?

Again, not responding to the original issue but the later pile-on from dfkettle. And we really don't know what they mean by "this error" other than (my assumption) what is in the subject.

Copy/paste characterset issue. Replace the "-" with one typed on your keyboard. And there's no need for a stage variable to do this.

Why not just stick with the -lstages option of dsjob? You'll need to get the list of jobs first and then loop through that, I do believe, but it would be very easy to accomplish.

There's no relationship between the number of hash keys and the number of partitions.