> If a JOB (NOT STC) is cancelled, it appears that its LSQA storage is not > cleaned up (I will use ELSQA for above the line). Over a period of days > with a JOB here and there being cancelled (and by chance in the same > INIT), I noticed that the amount of PVT available to a JOB was > diminished (again, EPVT for above the line). > > Any one else seen this kind of behavior?
It happens a lot. Task and step owned storage (including LSQA) is freed when the owning task (or job step task) is freed. The other LSQA choices are all "address space" level storage (subpools 205/215, 225/235 and 255) which are never freed automatically. So if an authorized program obtains storage from any of these subpools and does not free it, then you will have a "memory leak" and/or fragmentation of the high private area. The other situation (batch jobs only) is where an authorized program such as an exit runs "between" steps and obtains LSQA storage from the then-current task, or job step and does not explicitly free it. Both situations would typically be considered a programming error, although there may be cases where you would want to leave control blocks hanging around across steps. So it's probably a bug and if you're seeing ACEEs in the orphaned storage you ought to suspect somebody's exit is asking questions of RACF without the proper environmental setup/teardown. I could have sworn there was a new keyword something like STOPREGIONLOSS, to cause initiators to periodically recycle instead of giving S822 abends, but now I can't find anything about it, so maybe I was just dreaming it. CC ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [EMAIL PROTECTED] with the message: GET IBM-MAIN INFO Search the archives at http://bama.ua.edu/archives/ibm-main.html
