DSXchange

Better be some darn beefy hardware if you expect to be able to run 500 jobs in parallel on it.

10? That's not going to help. Basic formula is to take the size of the loaded DATA.30 and OVER.30 in bytes, add them together and then divide by 2048, assuming a page size of 1. That would give you a better minimum modulus to use.

You probably should have let people know that it is related to this problem that you are having.

I think one problem is the fact you are generating XML in your Filter command and then expect it to be read nicely by a Sequential File stage. This can be a little... problematic. Looks like an Error 11 on AIX is "Write error on direct file" which isn't a whole lot of help. Out of curiousity, you've...

You do, as the same column name needs to be mentioned twice in the query.

Good.

For anyone that stumbles across this thread later, can you explain what you ended up doing to correct your problem? That way someone with the same problem will know what worked to fix it.

As Ray would say: "It's a sharing place".

vijayindukuri wrote:... and what do i need to specify if i want the return value

If you are still not sure what to do, either reread my post or check the online help - the answer is there.

We know. How many triggers do you have defined on the Routine Activity stage? You can't use a single trigger for this. Read the Help, it explains what you need - either a specific failure trigger or a combination of 'Ok' and 'Otherwise' triggers - if not, then *it* will handle anything it perceives ...

Pretty sure you need an explicit 'failure' trigger to stop those messages, like an 'Otherwise' one.

It's not a failure in the code, it's simply the fact that the routine returns a non-zero value and you've asked the Sequence to handle these automatically for you.

Can you not use Custom / User Defined SQL to accomplish this?

You never "have" to use the DRS stage. From what I understand, it was created as part of the PeopleSoft deal. The goal was to allow your jobs to be "database independant" - a simple switch of a parameter and they go from talking to Oracle to any other supported database, for example. Of course, it i...

Rob, I think you are thinking of the ROWNUM psuedo column in Oracle, where it gets tricky to limit output based on that under some circumstances. What Arnd is saying (I do believe) is that, with specific sort criteria - an order by - then you will have a known 'last record' in the output dataset. An...

I doubt there are any problem characters in these fields, it's more about the fact that these kinds of large TEXT fields (or LONGs or CLOBs) aren't really appropriate for processing by an ETL tool. However, that's not to say it can't be done, I've done it for CLOBs in Oracle, for instance. First thi...

Simply concatenate them together 'after job' with the appropriate operating system command - in the correct order, of course.

UNIX would be 'cat', Windows - copy? type? something like that.