DSXchange

And your command line utility appears to be SQLCMD. I'm sure your DBA could help you with syntax questions.

Not sure why in one and not the other but it seems your quotes are being stripped. Sometimes the easiest solution is to put the quotes in the query itself rather than the parameter:

MEASURE = Paid Cont Comm

and Measure = '#MEASURE#'

Yes. I'm pretty sure that as long as there is a VOC record for the hashed file (and there is one automatically when it is created in an "Account") then sql statements like that are valid.

There is other (Retrieve?) syntax that is valid as well, I'll leave that for Ray or Kim to add.

Keep in mind the fact that hashed files do a "destructive overwrite" based on the key(s) defined and "last one in wins". In other words, just because you wrote X records to the hashed file doesn't mean there will be X in there when you select back out from it... or count them.

... or you could upgrade to 8.5 where they've implemented "transformer looping".

Define for us what exactly "validate file" means.

... and/or your Teradata DBA.

Does any database allow a bulk load to a view? Other than RedBrick.

What is the characterset of your Oracle database? Is it in fact using a multi-byte one? This is typically exacerbated by using "byte semantics" in Oracle, meaning you have a VARCHAR2(80) column that is defined as 80 BYTES when it really should be 80 CHARACTERS.

suresh_dsx wrote:we observed loss of data at sum of the column not at the row level.

This implies an issue with however you are doing the "sum" of the file data... can you explain how you are doing that?

You can't in this particular scenario. Use two targets for this, one for the insert and one for the update. Yes, that does mean you need to determine which is which rather than relying on the crutch of the database doing that for you but that should be a very simple lookup.

Also, after the job Aborts, please Reset the job from the Director and post the contents of any "From previous run..." log message that gets added from the reset.

Best way to check that the target writes are not the issue, put an @FALSE constraint in the last transformer so nothing goes to the target.

I don't recall which way is which and no longer have any ability to test the theory, but I seem to recall DataStage can handle this rather easily. Meaning, one of the two slashes works in both environments and it knows that it needs to "flip it over" for the "other" environment. ...

Best to start by posting in the correct forum so I don't have to move it - the FAQ Discussion forum is for discussing topics that could become FAQ postings, not general support. With that out of the way, curious why you stated "without using insert statement"? You want to put data in a dat...

That's true, I always forget you can change the default, mostly because you really (really) shouldn't do it.