DSXchange

When you generate your surrogate key, use the partition number and partition count in the formula. Search for details. That way you will generate unique numbers. You ARE actually getting all the numbers, but alternate numbers (assuming round robin partitioning) are going to alternate nodes. If you l...

STARTED includes "finished", which also has type "Control". Pick out the timestamp from that and search for the next newest event, which has to be the start event for the same run. No events were auto-purged from your log, which is why the "finished" event was last. You mustn't leap to conclusions, ...

If it's a parallel job as you indicated, use DateToTimestamp() function, making sure to set the date format string to include microseconds.

That's a security issue. Consult with your network administrator.

Not really. What if there really had been a failure? The following would be preferable. Not only because it's documented (hint!). * Construct and execute SETFILE command. Cmd = 'SETFILE ' : hshdflpth : ' ' : 'hsdfile' :' OVERWRITING ' PERFORM Cmd * Record result of SETFILE command. Temp = @SYSTEM.RE...

Is the OP really still using version 4 of DataStage? If so, the variety of ODBC options may not be as great.

It's an internal subroutine, which means it is not exposed via the Routines branch of the Repository. There are others, such as DSExecute(), that are similarly implemented. As others have noted, they are documented, but can only be typed in. To use DSSetUserStatus from within a server job (Transform...

One or two "sandpit" projects too, I'll guess.

In Director log view, STARTED events show up with a type of Control. I attempted to be descriptive rather than prescriptive. That's why it was in quotes.

trobinson wrote:Have you taken a look at the Operators reference pdf for the changecapture Operator?

That is, the Orchestrate Operators manual. It doesn't ship with DataStage but can be had for the asking from the vendor (ask for the OEM 75 Manuals) or is handed out on pretty much every IBM DataStage class.

Not if your database is Red Brick, which supports efficient fact-to-fact joins, constrained on as many dimensions as you wish, via a multi-table index called a star index. It is worth adding that this technique - indeed the entire database now known as Red Brick - was originally designed by Ralph Ki...

I mean, how does a program not write an error message if it dies? If it's killed with SIGKILL it may not get the opportunity to. Further, all player processes redirect their stdout and stderr to their node's section leader process, which transmits anything to be logged back to the conductor process...

If you have multiple fact tables that refer to the calendar dimension I would suggest loading the calendar dimension into a Lookup File Set. That's because the "index" (hash table) that the Lookup stage would ordinarily have to build at run time is built and stored in the Lookup File Set when it is ...

Or use a Transformer stage and Field() functions to parse it?

As you've seen the Column Import stage is fairly particular about the input column you ask it to parse.

Who knows, maybe the OP will return and mark the post as resolved.