DSXchange

Yup... hence the big "IF".

A 'backup' copy of each table, yes, that exists during the load of that table only. After the load completes - one way or the other - you're back to only one table.

IF you can get the grants / permissions, you really should check out the 'swap' that I mentioned. WAY faster than your dump before and reload after.

"Something completely different"? I'd be curious to see what that something is. And I just want to reiterate what Ray stated earlier: while we can provide general help you really should be having this conversation with your Oracle DBA - someone who is familiar with how everything is setup ...

Hmmm... there's no coming back from a truncate. You'll need an alternate approach in that case, perhaps something like a table swap? Rename the original table, create a new empty version and load... on failure swap the old table back.

That would be the tool as long as you have it in your version. It's not available until 8.5 from what I recall...

Raise an error.

Rather than wait for Ray to wander by again and reply, why not just... give it a try? That way you'll have something to talk about when he shows up.

Welcome aboard. First I'm curious... what have you tried so far? Do you have any thoughts on how to approach this we can discuss?

Problem is you've said you've set the SP stage to 'transform' which means you are calling it for every row through the job. If that's the case, when should the commit be executed? Row by row? Once at the end of the job?

It's not looking for "service name" but rather the name of the TNS entry itself. While that may typically match the service name it can be any name you like... including BILLY_BOB should you so desire. :wink: However, there can be other parts in play depending on things like your "sql...

By default: $ORACLE_HOME/network/admin/tnsnames.ora

Correct, that's a parallel only feature. Here you need to see the SQL it generates for an equi-join and then modify that to do the between. That way you'll know what it is looking for in the way of bind variables - positional or numbered, I don't recall off the top of my head. One thing to know as a...

You use CFF if your file is in EBCDIC format. That's certainly one reason to use the stage but hardly the only one. As the name implies it is meant for Complex flat files, regardless of EBCDIC v. ASCII. Typically that means COBOL format files from a mainframe in EBCDIC with packed fields but it's c...

FWIW, that wasn't something I ever saw.