MVSFORUMS.com

[davinski.bby]

I have a CICS program that calls a PROC which has the following steps:

[Terry_Heinze]

If CREATGDG does a lot of processing, one way of reducing contention would be to create a cataloged data set (instead of directly creating the +1 generation of the GDG), then copy that data set to the GDG base which would still be enqued for the duration of the copy, but might enque it for a shorter period of time than your current method.
_________________
....Terry

[davinski.bby]

thanks for the reply. If this would the case, would it also lead to contention if two users would access the program at the same time then two procs would create the cataloged dataset at same time?

[kolusu]

davinski.bby,

Change your GDG creation to a later step. Let the PGM CREATGDG create a temp sequential data set and FTP it and then copy the temp file to a GDG Version. By doing so you are avoiding the pitfall of FTP'ing the wrong file.

Hope this helps...

Cheers
_________________
Kolusu
www.linkedin.com/in/kolusu

[Terry_Heinze]

Either create a temporary data set as Kolusu suggests, or, unless you have a large number of users submitting the FTP job, each user could create a unique data set name because they would have unique jobs. I prefer cataloged over temp data sets because of recoverability.
_________________
....Terry

[CraigG]

Create the cataloged datasets with userid as one of the nodes of the dsn.

[davinski.bby]

Thanks guys for the help!

Kolusu, I did what you told me and it seems to be working now. I tried to run two simultaneous jobs and there were no contentions.

Can you explain why contentions happen when I am using GDG's and why it does not happen on flat files?

Thanks again.

[Terry_Heinze]

I don't have access to a mainframe, but my guess is that you still might have contention if the 2 jobs that created temp data sets are trying to add a +1 gen at the same time. It's just that the likelihood of contention is less. I think you just got lucky with "no contention". Try your 2 jobs with very large data sets and run the 2 jobs concurrently. I think you'll find that the 2nd job will get a "waiting for data sets" message and will wait for the 1st one to release the enque on the GDG base.
_________________
....Terry

[davinski.bby]

hi terry, you are also correct. i tried to execute simultaneous jobs and got the following messages in my logs:

[davinski.bby]

the jobs JOB00001 and JOB00002 both completed successfully. JOB00002 waited for JOB00001 to release the files. then it run to success. is this still possible if at least 50 jobs will run?

Thanks.

[kolusu]

davinski.bby,

I only see contention on the Flat files but not on GDG. I thought you used a Temp dataset to create the file from step 1 and then copy to output file.

The advantage of temp files is the OS decides the names and they are unique to each job.

Show us the JCL which you ran
_________________
Kolusu
www.linkedin.com/in/kolusu

[davinski.bby]

hi kolusu, the issue i am having with temp datasets is that the users who will be submitting the jobs have no access rights to create temp datasets. they are only defined to have access on certain datasets. i am receiving "Create access not granted" when creating temp datasets when using a TEST User ID.

[davinski.bby]

hi kolusu, i think i got what you mean. please disregard my last post. i will try the &&TEMP file and let you know if it works. thank you very much.

[davinski.bby]

hi kolusu, i am now having a different concern. if i use the temp dataset and the OS decides the name, i am unable to know what dataset to ftp since we use a standard ftp script:

[taltyman]

Use DD support in your job. See this manual http://publibz.boulder.ibm.com/cgi-bin/bookmgr_OS390/BOOKS/F1A1B950/4.2.4?SHELF=F1A1BK61&DT=20050708142126

And sample code from this manual.