DSXchange

That's going to make your job difficult. I'd still pursue the resource angle and also involve your official support provider.

Sounds like a partitioning issue to me and somewhere in the stream you have it doing 'entire', so all rows go to each node.

Those errors logged during the reset are meaningless and can ignored.

:? That's not what I asked. Find an Aborted job, Reset the job from the Director and tell us if any 'From previous run...' messages get logged. Find out what changed last week, obviously something did. To me it sounds like a resource issue if you have the problem when "hundreds" of jobs ar...

There's no rules that I'm aware of, pick a number, any number... me, I'd probably change it from 60 to 600 and see if that helps.

And when you Reset the job?

You mean other than simply making its value larger?

Right, did say later that the job must Abort for the rollback to happen.

There must be other messages logged with this. If you manually Reset the aborted jobs, is there a 'From previous run...' message logged? If so, please post it.

Well, the question at hand was how to do it in a job so that's the question I answered. And not quite so straight forward if the tables are really in two databases, as was alleged here.

Yes, that is exactly what that routine does. How were you 'unable to give the routine'? It should work fine there or any other transformer derivation.

If the target is a database stage, there's generally no reason to sort the output data. Why do you think you need to do this? Or do you need to sort it before transforming it?

Please start the similar post.

You typically 'get the data in sequence' by sorting, would that not work for you here?

Put the third link through an Aggregator to get the total count, then run that single output record to your funnel.