DSXchange

sshettar wrote:But now we are having hard time running the jobs which are calling these issues.

Rather than worry about unsetting their read-only-ness, perhaps you could start by explaining what exactly this means. What issues are you having?

Guess not.

Me, I'd want to build something similar and run some tests but I just can't spare the time right now. I couldn't even spare my spare time for this as I basically have none right now. [sigh]

Yes, and it should be documented in the 7.5 Installation and Upgrade Guide pdf document.

Let's put it this way - it only needs to open 1 connection per stage to do its work, and that's the behaviour I've seen with the stages I use. Now, if the DB2 stage does something stoopie... [shrug]

Um... No. One stage = one connection.

Perfectly fine if done properly. While you're here, learn from other's woes to cut down on the frustrations this path could bring.

Just the directory would suffice. Doing both wouldn't hurt.

Can you 'invoke' a command line O/S function via JAVA? Then just call dsjob directly without any scripting wrapper. Or check the Command Line Interface section of the Server Job Developer's Guide, the APIs that dsjob leverages are documented there.

It will open a connection per database stage in a job.

Sure. Dynamic hashed file gets 'corrupted', typically resulting in the lose of the .Type30 file. Result is it suddenly becomes a Type 19 (I believe) - one in which each record becomes a separate file. Delete the directory and it should 'rebuild' fine. Happens sometimes when using the 'Delete' option.

The error is pretty self-explanatory. You've told the Aggregator to expect rows in a certain order and it found a row 'out of sequence' - one that violated your alleged order. Correct it.

Have you configured DataStage for Oracle? It's documented in the Install and Upgrade Guide and requires the dsenv file be edited to contain specific entries for ORACLE_HOME and your 'Shared Library Path'.

Any job log will only show how many records were written to a hashed file, not what it ended up containing. As noted, a hashed file with a VOC pointer can be counted with normal sql or with EXECUTE COUNT RETURNING. Or build a brain dead job that sources from the hashed file to a sequential file but ...

There's no 'scheduler' that ships with DataStage. As noted, it's just an interface to cron on your UNIX server. If you are using DataStage to create cron entries for this, then just 'Add to schedule' four times once a year. Are you looking for something more... automated? Do you have an Enterprise s...

What documentation, exactly? The XML PACK 'guide' ships with DataStage and so should be on your PC in the 'Docs' directory. There is also a 'Best Practices' document that Kim Duke is hosting on his site for us.