Long ago, like MVT and the 1st MVS days, application programmers would allocate (IEFBR14, or just typos in JCL) datasets that were never opened. They lacked an EOF mark. These would use space and cause problems with various automated or manual clean-up efforts. And, residual data was accessible if you knew how, Creation of new GDSs also needed a model DSCB. Well before I started working here (41 years come August), the Sysprogs of the time wrote PGM=OPENCLSE. Which scanned the TIOT for =C'DD', opening and closing all DDNAMEs of the form DDxxxxxx. When used, this helped, but still such errors continued and part of periodic, manual DASD management was to search and destroy these wasters of space, and DASD wasn't cheap back then Sometime in the 90s, we implemented SMS management for the application data. Implementing a default DATACLAS ended the issues of no EOF (and the model DSCB need), and use of PGM=OPENCLSE has depreciated. I still use it for clearing a multivolume, non-SMS SMF dataset. My fading memory tells me that we also solved the issue of large allocations not RLSE'd then, but empirical evidence is that such allocations still occur on my system. However, they are not an issue because, as noted earlier, both DFHSM and FDR/ABR can routinely release such over allocated space.
As I look, I see that partial release is a function available via MGMTCLAS, times have changed, I may turn it on in my default MGMTCLAS and evaluate the impact. > -----Original Message----- > From: IBM Mainframe Discussion List <[email protected]> On > Behalf Of <David> <Mingee> > Sent: Wednesday, May 11, 2022 2:30 PM > To: [email protected] > Subject: Re: [EXTERNAL] Re: Potential Resource Savings Techniques > Great technique. What utility or JCL is being used, if the dataset is not > opened then RLSE does not happen. > ---------------------------------------------------------------------- For IBM-MAIN subscribe / signoff / archive access instructions, send email to [email protected] with the message: INFO IBM-MAIN
