DSXchange

I did go through the Parallel Job developer guide but did not find any thing related to the format of data in data file. Look again. The section (I think it's in Chapter 2) is called something like "data types in DataStage" for example you learn that a Date requires four bytes and a times...

In that case don't pass NULL to the Num() function.

Looking more and more like the 2GB barrier. Merge stage works by creating a hashed file under the covers. As far as I'm aware, unless you have 64BIT_FILES set, there's no option for having that hashed file created with 64-bit addressing.

The error message is because the variable called k is not set in every path through the code, and has not been given an initial value. Before/after subroutines only ever have two arguments; this is not something that you need to check.

In a word, "management". You need to manage (co-ordinate) your QC environment.

As far as I am aware, there are no standards for this.

47shailesh wrote:#Is it possible to set the status of job from RunOk to Running and later reset the status again to RunOk again.

Not legally.

47shailesh wrote:#Or is it possible to disable the jobhandler for few jobs, such that no one can attach the job

Only by stopping the engine.

... only if they're in the job as parameters.

GetEnvironment() can reference any environment variable available to the process.

Overrides would be made against the rule set you're currently using, I guess USNAME.

To create a custom rule set the easiest way is to copy an existing one. Then modify the classification table, the dictionary and the pattern action language to suit your requirements.

When I view posts in DSXchange using Internet Explorer or Firefox the "unread" (orange) icon changes to "read" (blue).

But, when I use Safari as my browser, they remain "unread", which makes life difficult.

There is no mechanism for suspending, or "pausing", an already-running DataStage job.

You could, if you can make a good case, submit that as an enhancement request.

That design operates as two jobs, even though it's one. For a sufficiently small reference data set the hashed file write cache and read cache can be used, so that lookups are performed in memory, which should be faster than performing them to disk.

Yeah, link collector. But it performs the same function as the parallel Funnel stage that I clearly had in mind.

To be honest, I'm surprised that running a warning trigger into a Terminator does not do what you specify. Does your Terminator activity wait for all running jobs to finish?

Just curious, how can database NOT be specified in a Connector?

Without creating a custom rule set or a really comprehensive set of overrides, not really. You could look at MNS but that's mainly for addresses. It gets trickier, for example EVELYN is as likely to be a male name as a female name in GBNAME but is always female in USNAME rule set. And in the AUNAME ...