DSXchange

I've seen similar behaviour from OCI stages with regards to reject handling, the posts are here somewhere and can be found if you are curious. Yes, it's all because of the array size. For Oracle, bumping the array size up still leaves the log with the correct errors but what is sent down the reject ...

Confusing. Assuming your User Defined SQL is in the Target stage, what in the heck is the source stage doing? That method is fine for something like TOAD but not appropriate for a DataStage job. Put the sourcing query in your first OCI stage. Let the target OCI stage do both the inserts and the crea...

Even with Write Caching turned on, the principle of 'destructive overwrite' based on the defined Key fields still applies, so nothing will be 'destroyed'. Existing data will be replaced just the same as if you hadn't checked that option. From what I've seen, write caching is good for writing into em...

Truncate. Or drop and recreate. Or ask your friendly neighborhood DBA to do it for you. Only thing is, if there are 'many referential integrities', you may not be able to do either... unless they are disabled first. How were you planning on resolving those issues once the table is emptied - will you...

I just did exactly what I described, it lets me use the job in our 'KBA' Job Control architecture easily. You can certainly do something similar in a before or after job subroutine, but the job has to do something - it can't just do the before/after stuff - so if there isn't an appropriate job to ha...

tonyczarnik wrote:Can I build a DS job that does not have input or output, but just executes a unix script?

Sure. You don't need to have anything on the canvas, but I'd suggest a nice note explaining what you are doing. Then put the code to execute the script into the Job Control tab of the Job Properties.

There's different ways to acomplish this... you could write to three different files and then concatenate them together after job. Note it's not the 'one stage' that is the problem here, it would be the fact that they all point to the same physical file. Or write a single record and concatenate the ...

Horrorshow!

Private, phooey - it should be a public pet peeve! While that kind of crap is "fine" for text messaging your Milkbar droogies, please leave it there. It has no place in any sort of professional environment - especially one where English may not be the native tongue of a great many of the participants.

Hey Arnd - two minutes. You give good ASAPmy friend!

You could also investigate the WaitForFile stage, which actually doesn't have to wait at all. You can use it as a file existance checker, for example, and branch accordingly.

koolnitz wrote:Also, if you could (A) tell how does this daemon work and (B) does it remove the unwanted (or better unused) RT files?

A. See link above.
B. Nope.

Search is your friend!

Here is a long post which discusses it amongst other interesting things. Also known to contain the rare Double Full Wurlod.

they aren't really invisible, just too small to see. Interesting distinction. Reminds me of a new feature we just saw from a 'partner' who supplies flat files to us. Suddenly, processing job blowed up due to a format change they inadverdantly introduced post COBOL upgrade. We assumed they had switc...

The other typical 'invalid characters' that on occasion I still insist on using are a trailing semi-colon (TOAD likey, DS not so much) or you've mistyped a job parameter in there somewhere and it's not getting translated. Oracle then complains about the '#' as an invalid character.