DSXchange

The other 'problem' with using ODBC via BCI on a UNIX server is the fact that the drivers that ship with the product are 'branded' and therefore are only for use by things like the ODBC stage. Once you start using them from the 'outside' via BCI it sets off a 30 day evaluation clock. After 30 days a...

Are you doing it from inside the Transformer? That's where you need to be right-clicking.

I have promoted one application through version control . and after some days I made some changes in application and promote , so in My version control I have two version , Now I need to clean all the entry and promote from base version . So pl advice me how to do ? I'm a little lost still. Do you ...

Which is pretty much exactly what I said... just with a different turn of the tongue.

But... at this point... isn't it 'too late' for corrections of this nature? These kind of things need to be done before writing the file out. Right now the OP can't even read the file in properly. First course of action (in my mind) would be to get whomever supplied you with this file to correct it ...

Those don't sound like 'ds queries', those wouldn't look any different from 'regular' queries run in the database. That looks more like someone is executing 'stored procedures' or whatever the equivalent term would be for SQL Server. Check the actual ODBC stages, both source and target and look in t...

The job has transaction zise = 0 but works as if trx = 1 (commit after each row) This makes no sense to me. The warning limit is 50 (the default) so the job aborts You didn't mention the fact that the job had actually aborted. Perhaps it's related to the fact that you marked this as a Parallel job,...

It simply means to have the packed fields unpacked. Not an 'internal' representation, but rather something 'external' or people readable.

Yah, noticed this when visiting ADN today - all of the forums were renamed with 'IBM WebSphere' stuck over the old 'Ascential' label.

No problem Shawn.

Yup, exactly! Trying to get it to work as he expects, and it should. From the original post: What I was expecting is that the job will abort because of duplicate keys and a rollback will be done... so no change will be done for the table. (I use transaction size 0 and array size 1000) As I mentioned...

I'd suggest leaving your commit size at zero so it works as you expect. The only thing you are missing is to set limits on your job so it aborts after X warnings. It is the fact that the job aborts that causes the rollback.

We keep our limit fairly low, typically 1.

I think that, if you search the forum, you'll find that the FIX option actually removes any record of the project.

There were several discussions in the recent past that could be found using those keywords from what I remember, they might help with a solution to your problem.

Welcome aboard. Sorry it couldn't be under better circumstances.

Please define "tried to recover"... what exactly did you do?

Busted.