DSXchange

Convert it to a null or an 'in-band' null value if the target doesn't support nulls. Failing that, they get rejected.

kumar444 wrote:please see the loader logfile:

/opt/IBM/InformationServer/Server/Scratch/ora.19206.704488.0.log

for details.

Did you?

No problem... one last nitpick, however. :wink: Best practice is a common prefix on stage variables so they are easily recognized when referenced elsewhere, much like you are all doing for job parameters and the like, yes? I use "sv" so would do something more like this: svCurrentVal=> Inp...

Bad assumption. All you know is that records with the same hashing keys will end up on the same node but you have no knowledge of which will go where.

While it may work, you really should be using meaningful stage variable names so your code is 'self documenting'. SV1, SV2 and SV3 aren't especially helpful in figuring out what they are accomplishing.

I don't think there's any automagic way, Support might be able to provide one for you, though. Worst case a script using dsjob could iterate through a list of job names and validate them one by one. You could also use dsjob to generate the list of jobs.

This FAQ post should help illustrate what we're talking about:

viewtopic.php?t=107844

You did note that Ray said common group, not common user yes? If you can't go the common user route then your second best approach is the common group, ensuring that everyone running DataStage jobs has that same group as their primary group. Between that and the umask you should be fine.

As noted, if I had something to add on topic, I would have. And an exact search of the over 300,000 posts here for "Creating 2 instances instead of one" only finds your thread. So you're either the first or that's not quite the right message wording. Regardless of what happens here, don't ...

Yah, thought I recalled from my RTI days that the 'Input' stage isn't a requirement, so made a leap of faith to hoping that a Sequence could in fact be exposed. Ah well. Nice workaround, though.

FP1 = Fix Pack 1, as in... patches.

If anyone did, I assume they would post. If you received a patch from IBM that didn't help, can we assume you reported that back to IBM? And that they are continuing to work on your issue?

And the how would depend on the nature of the hashed file, pathed verses account-based. The former can simply be copied by whatever mechanism you choose as long as (for dynamic hashed files) all of the files in each director (including the hidden file) make the trip together. Account-based hashed fi...

A job is a job is a job, be it a Server, PX or Sequence job so I would think this should be feasible as long as it is Multi-Instance. Can't say I've ever tried it myself, however... can they be "ISD enabled" or whatever it is called nowadays?

Don't forget, we're talking first/last 2 entries per group here.