Trigger Job from CA7 based on GDG availability

parab.nandan · Posted: Sat Sep 12, 2009 11:29 pm

I want to trigger a job if the generation of gdg file is available on the system.
For example: I have an ADSO application which creates a gdg generation 1 by submitting the job. Once a gdg generation 1 is created my CA7 job looking for that generation should kick off, how can I define a job in CA7 with such trigger?

Also this trigger job should be capable of handling multiple files which might get created at the same time or in matter of seconds from the previous file.

Is it okay to reference the gdg base in the triggered job?

Thanks
Nandan

dick scherrer · Posted: Mon Sep 14, 2009 5:00 am

Hello and welceom to the forum,

Depending on the specifics of what you want to do, you can probably accomplish the scheduling requirement via ca7. . .

Jobs may be triggered by the creation of a dataset.

A gdg base may be referenced in a submitted/triggered job.

If there may be many generations created in a very short time span, there may be a timing issue - if the job run too quickly, it may miss some of the data that should have been input (indeed, some generatoins might not yet be created).

Rhetorical question - When processed as input, are all generations to be deleted?

You need to think thru all of the timing issues and consider what the complete requirements are. Once you get to the specifics rather than generalaties, you might implement the needed process(es).

parab.nandan · Posted: Mon Sep 14, 2009 5:39 am

Yes all the generations would be deleted once processed as input. The question is what if the generation dataset arrives while the previous job is running?

I have no control over the timing at which the file will arrive as they can arrive at any time. But I need to ensure none of the files are missed and they need to processed at whatever time they come.

Do we have the complete solution here? Hope you have all the conditions covered.

parab.nandan · Posted: Mon Sep 14, 2009 5:42 am

So if I understand this correctly we need to mention to gdg base as dataset trigger for the job in CA7, correct?

dick scherrer · Posted: Tue Sep 15, 2009 2:24 am

Hello,

Are all of the datasets created on the same system that will process them?

What business problem is there if there is a delay between the creation and the processing of the data? In the previous posts, you mention that there might be multiple files to process, but later mention that "they need to processed at whatever time they come". It might be one or the other, but both cannot occur

If you explain the functionality to be supported, others have already implemented solutions and may provide a more usable reply.

parab.nandan · Posted: Tue Sep 15, 2009 3:45 am

Yes all the datasets will be created on the same system. Let me explain you the sequence of steps here.

1. I have an ADSO application where we have multiple users accessing an application ABC. Once we press an enter on this application it submits a batch job on the mainframe through internal reader which creates a gdg version.
2. As on today I have a scheduled job PEDASN which runs every 15 mins which looks for the arrival of this gdg version, but to prevent it from abending due to empty version I create an empty version every time this job runs. The last job run step is to delete all the versions of the gdg.

I want to get rid of 15 mins job schedule as part of business initiative and run this job only when the gdg version gets created.

As of today ADSO application is access by multiple users across americas so in a given minute or two there can 10 or 15 or 2 etc files getting created.

I hope the problem statement is clear now.

dick scherrer · Posted: Wed Sep 16, 2009 1:23 am

Hello,

parab.nandan · Posted: Wed Sep 16, 2009 3:30 am

We are looking to cut down on job runs to save on CPU time and are looking for the job to run only when it has something to process - this is business initiative.

dick scherrer · Posted: Wed Sep 16, 2009 3:46 am

Hello,

If the goal is to reduce cpu time, i suspect you'll need to look elsewhere. . .

Suggest you run this a few times with no input and notice that almost zero cpu time was needed.

If the cpu time used is not nearly zero for a "no input" run, that would surely be something to look into. This would mean that the code is doing some large amount of unnecessary processing.