DSXchange

Somewhere around 5.0 preload switched to paging, so you're not guaranteed a full snapshot of the hash file in memory. I spoke at length with tech support regarding the new caching logic back then because there were GTARS opened regarding the incorrect handling of multiple updates to the same row in ...

Check your umask setting and verify all users importing/creating jobs are in the equivalent permission groups. If person A imports a job and B tries to run it, the permissions may not allow it.

Server jobs are very gentle with mismatched data types. Your expression is invalid: If lo.custid Matches ("1970" Or "1971") Then in_data.value Else 0 can be: If lo.custid = "1970" Or lo.custid = "1971" Then in_data.value Else 0 or If lo.custid = 1970 Or lo.custid = 1971 Then in_data.value Else 0 Th...

Server jobs are very gentle with mismatched data types. Your expression is invalid: If lo.custid Matches ("1970" Or "1971") Then in_data.value Else 0 can be: If lo.custid = "1970" Or lo.custid = "1971" Then in_data.value Else 0 or If lo.custid = 1970 Or lo.custid = 1971 Then in_data.value Else 0 The...

Passive stage - files, tables, ftp of a file, etc. Physical objects that are either sources to a operation or targets of an operation. Active stage - an operation that manipulates the data, ie transformers which derive outputs from input and reference sources or aggregator/sort/collect/partition sta...

Transforms: single lines of code that can be used in derivations and constraints. During compilation, the source code is substituted into the job design information and then compiled in. Changing the Transform requires recompiling all jobs that use it. Routines: subroutines that are CALLed from a jo...

Are you trying to view from DS Manager or from a Designer job? You should try from a Designer job and use the get SQL info button.

snassimr wrote:what happend when I set the limit on limits tab during running the job ?

That works, but from an un-attended execution you should really consider custom job control.

Use the graphical metaphor and create a link out of a transformer stage. Use your functions or whatever to constrain the output. Your initial idea is not a good choice, avoid independent file i/o from functions in a transformer stage. Let the tool do the work it was designed to do. You'll be happier...

A job cannot control itself in any way. Job Sequencer logic is not available for changing. You could cut and paste it into a Batch job, but then you lose the graphical maintenance.

You can achieve the same effect with a series of hash files and passes thru the data. But it would not be the elegant solution. If a UNION is what you desire, that is a database action best suited to be done in a database. a UNION ALL can be achieved by simply concatenating like structured text file...

Practical is easy, it's how much degradation you're willing to sustain reading all of that data from the hash, or potentially parking it in the hash. You also have to deal with the large record size, which means you're going to want to hand tune a static hash file. Still looking for that CLOB soluti...

If your data has any chance of containing a lot of characters in the CLOB column, you're better off not sending it thru DataStage. Think of it like sending an email with an attached file. Your email may be small, but the attachment can clog the network if it's of sufficient size. The same holds true...

Your problem is simple:

Your request for data takes too long to get out of the database and the database can't preserve the snapshot of the data you wanted anymore.

You need to get the data out faster, or get it out in smaller sets.

When you run a select statement, Oracle copies everything that you are going to select into temporary space. If you take too long to go thru all of that data, Oracle ends your session because it can't hold all of that data any longer. Your issue is typical of people who attempt to take a lot of data...