DSXchange

I think you have made a good finding by enhancing the sample queries. The developerWorks forums and articles are outside of what IBM Support is in control of, much like DSXchange... Those are not part of the product documentation, but like they said, they are another resource. If you find an incorre...

Shared locks are sometimes called "read locks" and exclusive locks are sometimes called "write locks". However, because locks on Unix are advisory, this isn't enforced. - Wikipedia "file locking" topic The "advisory" part was news to me, assuming the informat...

Cool. I had simul-edited my post to ask a few more questions...

Interesting find. That all seems very odd and confusing. Is there any chance the isx file was being appended?

What version are you on? Were you running the export on the Client tier only? Is IBM aware of the issue?

Interesting. There is actually some documentation about that setting too.

http://www-01.ibm.com/support/knowledge ... ig_bg.html

This feature was first introduced in DataStage 9.1.0. You can set the write method to "Generate Multiple Files" and use a key column, akin to what you labelled as "bursting." You can optionally append the key value to the file name. http://www-01.ibm.com/support/knowledgecenter/S...

Also check Fix Central to make sure you have all the latest fix packs and rollup patches applied.

Again, how does your compile time compare to a client that is "right next door" (on the network) to your DataStage server?

What version are you on?

How does your compile time compare to a client that is "right next door" (on the network) to your DataStage server?

That is disturbing. Is your client by any chance very, very remote from the server, perhaps using an old style dial up connection?

Have you checked on Fix Central or with Support to see if there is any patch available?

Have you tried use the Index function?

http://www-01.ibm.com/support/knowledge ... tions.html

My "database 101" understanding is that surrogate keys are supposed to be system generated, not carry any meaning, and not be visible to the user. It sounds like perhaps someone may be trying to attach meaning to particular SK values, which does not make much sense. You should be free to s...

I'm guessing this is an academic question but would be curious if there were a real world reason behind it. Parallel jobs are unaware of Server jobs' internal date format. A parallel concept in Parallel jobs can be found in the JulianDayFromDate() function. It returns an integer based on the input d...

I wonder if a NULL value appearing in the column being checksum'ed could cause such an error.

It is possible that your job actually executes under some other user account (you could add some sort of whoami command to verify) and/or likely that the run time execution environment has different settings that your command line session. Start doing some comparisons there.