z/OS DFSMS Using Data Sets
Previous topic | Next topic | Contents | Contact z/OS | Library | PDF


Restrictions and Guidelines for Shared Resources

z/OS DFSMS Using Data Sets
SC23-6855-00

Restrictions for using the LSR and GSR options are:
  • Empty data sets cannot be processed (that is, loaded).
  • Multiple LSR pools in an address space are obtained by using the SHRPOOL parameter of the BLDVRP macro to identify each LSR pool.
  • Control interval access cannot be used (ACB MACRF=CNV and ACB MACRF=ICI).
  • Control blocks in common (CBIC) cannot be used.
  • User buffering is not allowed (ACB MACRF=UBF).
  • Writes for data sets with SHAREOPTIONS 4 cannot be deferred (ACB MACRF=DFR).
  • Request parameter lists for MRKBFR, SCHBFR, and WRTBFR cannot be chained (the NXTRPL parameter of the RPL macro is ignored).
  • For sequential access, positioning at the beginning of a data set must be explicit: with a POINT macro or a direct GET macro with RPL OPTCD=NSP.
  • Temporary CLOSE and ENDREQ do not cause buffers to be written if MACRF=DFR was specified in the associated ACB.
  • Address spaces that use Hiperspace buffering (LSR only) should be made nonswappable. Otherwise, the expanded storage (and, therefore, the Hiperspace buffers) will be discarded when the address space is swapped out.
  • With GSR, an ACB and all related RPLs, EXLSTs, data areas, and ECBs must be stored in the common area of virtual storage with protection key 0 - 7; all VSAM requests related to the global resource pool may be issued only by a program in supervisor state with protection key 0 - 7 (the same as the resource pool).
  • Checkpoints cannot be taken for data sets whose resources are shared in a global resource pool. When a program in an address space that opened a data set whose ACB has MACRF=GSR issues the CHKPT macro, 8 is returned in register 15. If a program in another address space issues the CHKPT macro, the checkpoint is taken, but only for data sets that are not using the global resource pool.

    Checkpoint/restart can be used with data sets whose resources are shared in a local resource pool, but the restart program does not reposition for processing at the point where the checkpoint occurred—processing is restarted at a data set's highest used RBA. See z/OS DFSMSdfp Checkpoint/Restart for information about restarting the processing of VSAM data.

  • If a physical I/O error is found while writing a control interval to the direct access device, the buffer remains in the resource pool. The write-required flag (BUFCMW) and associated mod bits (BUFCMDBT) are turned off, and the BUFC is flagged in error (BUFCER2=ON). The buffer is not replaced in the pool, and buffer writing is not attempted. To release this buffer for reuse, a WRTBFR macro with TYPE=DS can be issued or the data set can be closed (CLOSE issues the WRTBFR macro).
  • When you use the BLDVRP macro to build a shared resource pool, some of the VSAM control blocks are placed in a system subpool and others in subpool 0. When a task ends, the system frees subpool 0 unless it is shared with another task. The system does not free the system subpool until the job step ends. Then, if another task attempts to use the resource pool, an abend might occur when VSAM attempts to access the freed control blocks. This problem does not occur if the two tasks share subpool 0. Code in the ATTACH macro the SZERO=YES parameter, or the SHSPL or SHSPV parameters. SZERO=YES is the default.
  • GSR is not permitted for compressed data sets.

Go to the previous page Go to the next page




Copyright IBM Corporation 1990, 2014