On Fri, 11 Sep 2015 14:09:32 -0500, Roland Kinsman wrote: >Thanks for quick replies. > >I'll look into using ATTACHMVS instead of CALL, or find a way to add logoff >into the loop. That one looks very promising if I can pull it off. > Careful! the UNIX "address TSO" is radically different from the TMP "address TSO" Don't try LOGOFF in the latter. It will either do nothing or do something undesirable.
>I'm calling both programs using TSOEXEC CALL *(xxx). > I never understood TSOEXEC even though I've needed to use it at times. >We're already using REGION=0M. > >Not sure what language the programs are in--they are both utilities from >different vendors. The first utility is doing the open/close of the members. > >I'm getting the member names using LMMLIST. > >I am using DELSTACK in my loop, as I use a number of QUEUE statements. Not >sure if this has the same effect as DROP. > >I'm not actually processing the contents of the members. The first program in >the loop processes the PDS member, then the second program processes the >output of the first one. For the first program in the loop, one of the parms >is the name of the PDS, and the member name is on one of the control cards. >My REXX generates the control cards then calls the first program. Based on >the messages from the job, it appears that the storage problem is in the >second program. > Can you enumerate allocations using something such as BPXWDYN( 'info ...' ) or LISTALC STATUS SYSNAMES to see whether some are not being freed? Wrapping the content of the loop in PROCEDURE EXPOSE ... will automatically drop compound symbols not EXPOSEd. -- gil ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
