Version Control Questions

[gjacob]

I am setting up our DataStage environment, and have some questions regarding Version Control. The PDF didn't seem to cover it.

1) Can you promote across servers (Development and Version Control on one server, Production on another)?

2) How do you have multiple job versions on the Production server? When I promote a job, it overwrites the job on the destination. I don't seem to have an option not to.

3) Is there a way to promote a version at runtime (via command-line)? That would allow our scheduler (Control-M) to run a specific version.

I would have thought #2 would be in the docs since they plug it as a feature.

Thanks!

[chulett]

Something in the way of answers for you...

1) Sure! Since this is a client-side tool, it doesn't care where all of your projects are... and they typically reside on different physical servers.

2) You don't.... and I'm not sure why you would want to. The only place where "multiple job versions" exists in inside the VERSION project. FWIW, they don't plug this as a feature that I recall. The feature you are probably thinking of is support for multiple VERSION projects which is new in this release.

3) No. Job Promotion isn't something that is typically scheduled, so again I'm a little lost as to exactly what you are trying to accomplish.

Perhaps more information or dialogue around your environment would help figure out how to work out some of these issues.

[gjacob]

Thanks, that's what I was thinking.

I was trying to mimic our current process (SAS scripts), where we have jobs that check out the program and run that version. This allows the option to run multiple versions (although very rare). To get around this, I will just create an Ad-Hoc project on the production server.

I do have one last question. When promoting a job, is there a way to grab all dependent components? I see there is a way to select those when checking-in a job.

[chulett]

When promoting a job, is there a way to grab all dependent components? I see there is a way to select those when checking-in a job.

Not that I am aware of... but then I don't believe you should really need it. As you mentioned, when importing you have the opportunity to bring in 'dependant objects' and they are then versioned together into the same "Batch". You would typically then select by Batch as well when you are setting up for Promotion.

Keep in mind the fact that VC doesn't check to see if something has changed and only bring it in if it has, you have to tell it which components have changed - and they are all brought in. If a dependant component hasn't changed, then there's no reason to import it again, nor to promote it along with the job(s) that use it as it already exists in your target Project. They come in together and they go out together...

Does that make sense?

[Ekkehard]

Additional questions regarding Version Control (DS 7.1, Win2k):
For the initialization I can grab all dependent components of a server job (a.e. routines); but I am missing the possibility to grab all the dependent components of a sequence that means all the called jobs. Or is there a way to configure this?

There is another interesting point:
I promoted a sequence with the "compile jobs" option. That was done without error and so the status of that sequence on my production server was "compiled". But this job sequence failed to run because of a not existing called job. The point is, that I am not able to compile this sequence anywhere because of that not existing job! I really wonder how does Version Control compile jobs (sequences)?

Ekkehard

[chulett]

For the initialization I can grab all dependent components of a server job (a.e. routines); but I am missing the possibility to grab all the dependent components of a sequence that means all the called jobs. Or is there a way to configure this?

Not that I am aware of. This seems like a good candidate for an enhancement request (or whatever they are officially called) to Ascential.

I promoted a sequence with the "compile jobs" option. That was done without error and so the status of that sequence on my production server was "compiled". But this job sequence failed to run because of a not existing called job. The point is, that I am not able to compile this sequence anywhere because of that not existing job! I really wonder how does Version Control compile jobs (sequences)?

I believe that all the Sequence needs to be able to be compiled is a job name in the Job Execution stage. It doesn't verify that the job actually exists at that point, that is saved until run-time. So while this seems a little strange, I don't think this can be classified as a bug. Perhaps the 'enhancement' discussed in your first point would help solve problems like this...

[ogmios]

DataStage Version Control is not version control in the classic sense. If you want to have "Version control" as Champ/SCCS/SourceSafe you have to roll your own for the moment.

Version Control is only to make your life easier moving jobs. Personally I don't even version control. Export/Import and Compile All plus is all need

And dependencies where never completely properly implemented in DataStage.

Ogmios