Page 1 of 1

Figure out the call chain of programs on zOS.

Posted: Thu Feb 20, 2014 12:08 am
by Anuj Dhawan
Hi,

Does anyone here know how do a search through the call chain to find out the root program's ID? I need this for Natural Adabas programs.

I think it'd help if I try o trace it suing debugging/monitoring or profiling tools - like TMON but at this shop they don't seem to turn on TMON for batch jobs/programs...

I read about READY TRACE but manual uses that the phrase "has no effect on your program". This "actually" would mean that the statement is tolerated, but treated as a comment. And that's for COBOL and not for Natural.

What SMF would capture this behavior?

Thanks for stopping by,

Re: Figure out the call chain of programs on zOS.

Posted: Thu Feb 20, 2014 4:08 am
by William Collins
That's one question in the title and a different one in the body.

Is Natural "LE Compliant"?

Does Natural use the standard /360 calling conventions? If not, what does it do?

Can you call other languages from Natural?

What is it exactly that you want to see? The call chain would normallyl be considered to be, "the program that called this one, the program that called that, the program that called that".

Is your conglomeratin of Natural programs statically linked, or are all the calls dynamic?

Re: Figure out the call chain of programs on zOS.

Posted: Thu Feb 20, 2014 5:14 am
by Robert Sample
I'm not sure I would want to attempt to use SMF to work out the program calling sequence -- while it probably can be done, I doubt it would be easy.

IPCS would be the tool I would expect to use. Capture a system dump when the program fails and process it through IPCS. I forget the precise command (that's what manuals are for) but one of the IPCS verbs gives you the call program sequence. You may need to use special IPCS verbs for Natural / ADABAS.

Re: Figure out the call chain of programs on zOS.

Posted: Thu Feb 20, 2014 5:29 am
by enrico-sorichetti
I'm not sure I would want to attempt to use SMF to work out the program calling sequence -- while it probably can be done, I doubt it would be easy.
unless I am completely wrong SMF will just report the LOAD caused by the first CALL,

and later on in the program the CALL SEQUENCE/CHAIN might be completely different

Re: Figure out the call chain of programs on zOS.

Posted: Thu Feb 20, 2014 10:41 am
by Anuj Dhawan
Thanks William.
William Collins wrote:That's one question in the title and a different one in the body.
I realized that I made a different question, sorry about that... :oops:

Please ignore the strike through text - "Does anyone here know how do a search through the call chain to find out the root program's ID?" and read it as "Does anyone here know how do a search through the call chain to find out the all the programs involed in chain?"
Is Natural "LE Compliant"?
Does Natural use the standard /360 calling conventions? If not, what does it do?
I think yes, but I'm not very sure on this. They have written a Natural code which in turn CALLs the other program by passing the program name as parameter. I've little experience with Natural but I think what they are doing in their system is as per standard from Software AG.
Can you call other languages from Natural?
Yes.
What is it exactly that you want to see? The call chain would normallyl be considered to be, "the program that called this one, the program that called that, the program that called that".
Yes, this is what I want to see. Though the final goal is to have get some "IT performance" clues out of it. So in other words, we'd need to know which program might have caused the "CPU hog". They don't have a profiling tool, so it;s tough to predict that - one way is pick, high CPU consuming program, get all the sub-programs involved and work with them manually; at least to start with.
Is your conglomeratin of Natural programs statically linked, or are all the calls dynamic?
I need to check on this.

Re: Figure out the call chain of programs on zOS.

Posted: Thu Feb 20, 2014 10:49 am
by Anuj Dhawan
Robert Sample wrote:IPCS would be the tool I would expect to use. Capture a system dump when the program fails and process it through IPCS. I forget the precise command (that's what manuals are for) but one of the IPCS verbs gives you the call program sequence. You may need to use special IPCS verbs for Natural / ADABAS.
Thanks Robert.

I'll explore the possibility to use IPCS, however, I'm not that good in playing with "dumps"...let me see if they've IPCS available.

Re: Figure out the call chain of programs on zOS.

Posted: Thu Feb 27, 2014 1:23 am
by dick scherrer
You might open an issue with Natural / ADABAS.