Where are my omvs address spaces?

I was running a Java program in batch, and it started an OMVS address space to run the Java. When Java stopped, I could not find the OMVS output – because I was looking in the wrong place!

The program I was trying to run was RSEAPI. When this starts it creates other jobs with jobnames like RSEAPI6.

Display the jobs

In SDSF it had

JOBNAME  StepName ProcStep JobID    Owner  
RSEAPI6  STEP1             STC06719 STCRSE 
RSEAPI1  STEP1             STC06722 STCRSE 
RSEAPI   RSEAPI   RSEAPI   STC06728 STCRSE 

The jobid of the RSEAPIn jobs are lower than the value for RSEAPI, this is because the address spaces were reused.

Shutdown or cancel

I cancelled RSEAPI6 and the other jobs stopped as well.

If you look in the spool for RSEAPI* it only showed job RSEAPI

Where are the other jobs?

There are system address spaces BPXAS. If your program issues a spawn or fork, it runs the work one of these address spaces. When the work request finishes, the address space stays running and becomes available for other work.

If you were hoping to find End of Step SMF statistics displayed (such as CPU and IO counts), these will be displayed when the BPXAS address space shuts down, and the figures are for all work which ran in that address space.

Purging the BPXAS job output

If you display the BPXAS jobs, it shows it is PROTected. This stops the casual end user from purging it. You have to add PROT to the command, for example $PS6723,PROT

Interacting with these address spaces

I tried to set CEE run time options to display the run time storage options, and to set heap size etc. I could not find how to do this.

3 thoughts on “Where are my omvs address spaces?

  1. Earlier this year, I was attempting to run an interactive debugger in both an assembler program and a C program running under USS. I also learned then that USS places a suffix on the job name. I was submitting my program via JCL, EXEC PGM=BPXBATCH, and was unable to trace that job name. Then, while reviewing SMF records later, I saw my jobname with a numeric suffix on it. I tried trapping the jobname by specifying that numeric suffix, and was still unable to trap it. I played around with running my program from the omvs command line, both from the foreground and using the & character to run it in the background. No go. I tried it under the shell and a subshell. No go. I called the vendor of the debugging tool and they could not figure it out either. Surely there is a way to predict where the process will run, but I don’t know what that is.

    Like

  2. I am so glad to reach to your blog, so much of Zos information. You are truly a IBM Champ.
    Can you advise me on ‘how to capture dump in OMVS address space ? ‘

    Like

Leave a reply to Colin Paice Cancel reply