PI44947: EYUXC0023S MAXIMUM DATA CACHE LIMIT FOR MAS CACHE. CMAS TERMINATES OR WILL NOT START.

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• Your CMAS terminates after issuing the following messages and
  taking an SVC dump:
  .
  EYUXC0023S Maximum data cache limit has been reached for
             MAS cache.
  EYUXC0024S The CMAS is terminating due to a previous
             error with the Cache component.
  EYUXZ0910I EYU0XZPT Dump,applid,cmasname,xxxx,CMAS,
             SSC,0012345,TRAC,EYU0XCCL,mm/dd/yyyy,hh:mm:ss.    .
  .
  Subsequent attempts to restart the CMAS may also fail with the
  same messages. Examining the SVC dump, you find the MAS
  dataspaces may be filled with control blocks that have headers
  starting with:
  .
      >EYUXCEYURXCWF
  .
  At offset +x'80' into these control blocks is a scope and
  context.
  .
  Additional Symptom(s) Search Keyword(s): KIXREVSVR
  

Local fix

• Terminate CMAS, all MASes connected to it, and any API
  programs. This will allow the ESSS started task to delete and
  clean up so that orphaned control blocks are removed.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: All CICSPlex SM V3R2M0 Users                 *
  ****************************************************************
  * PROBLEM DESCRIPTION: When a CMAS or MAS terminates without   *
  *                      going through normal termination, for   *
  *                      example, if the region is cancelled or  *
  *                      EXEC CICS PERFORM IMMEDIATE is issued   *
  *                      against the region, storage in the MAS  *
  *                      data cache may be orphaned.             *
  *                                                              *
  *                      If this occurs enough times during the  *
  *                      life of the MAS data cache, the MAS     *
  *                      data cache storage limit may be         *
  *                      reached:                                *
  *                                                              *
  *                        EYUXC0023S Maximum data cache limit   *
  *                                   has reached for MAS cache. *
  *                                                              *
  *                      If the limit is reached while a CMAS is *
  *                      processing, the CMAS will issue the     *
  *                      following message and terminate:        *
  *                                                              *
  *                        EYUXC0024S The CMAS is terminating    *
  *                                   due to a previous error    *
  *                                   with the Cache component.  *
  *                                                              *
  *                      Restarting the CMAS will not correct    *
  *                      the situation.  Instead, the CMAS, all  *
  *                      MASes and batch API programs that were  *
  *                      connected to it must all be terminated  *
  *                      at the same time to cause the MAS data  *
  *                      cache to be deleted.                    *
  ****************************************************************
  * RECOMMENDATION: After applying the PTF that resolves this    *
  *                 APAR, all CMASes and MASes must be           *
  *                 restarted.  Note that the restarts do not    *
  *                 need to occur at the same time.              *
  ****************************************************************
  The MAS data cache is a shared data cache, which persists even
  when a CMAS terminates, as long as one MAS or batch API address
  space that was connected to CMAS when it terminated remains
  active until the CMAS is restarted.  This is called a CPSM CMAS
  warm start.  If no previously connected MAS or batch API address
  space is active when the CMAS is restarted, then the MAS data
  cache is freed, and is created new when the CMAS is restarted.
  This is called a CPSM CMAS cold start.  The setting of the CICS
  START SIT parameter has no affect on a CPSM CMAS cold or warm
  start.
  
  Four major storage areas are allocated in the MAS data cache:
  
  -  The CMAS-MAS communications area is allocated by method
     EYU0CLET (CLET) during CMAS initialization.
  
     If the CMAS terminates normally, the storage is released by
     CLET during CMAS termination.
  
  -  When a MAS connects to a CMAS, method EYU0NLGT (NLGT)
     calls method EYU0XCWF (XCWF) to allocate a block of storage
     mapped by EYURXCWH, to allow CPSM CICS task related user
     exits (TRUEs) to report certain events, for example, CICS
     resource installs and discards or transaction and system
     dump events, to the connected CMAS.
  
     If the MAS agent terminates normally, the storage is released
     by method EYU0NLRT (NLRT) during MAS agent termination.
  
  -  When CPSM MTRAN monitoring becomes active in a MAS, method
     EYU0MMCL (MMCL) calls XCWF to allocate a EYURXCWH block used
     by MAS TRUEs to collect CICS monitoring and statistics data,
     which is used to build the MLOCTRAN and MREMTRAN resource
     records for the MAS.
  
     When MTRAN monitoring ends in a MAS, either due to a PERIODEF
     ending or if the CMAS or MAS terminates, and the retention
     period for MTRAN elapses, then the storage is released by
     either method EYU0MCTK (MCTK), method EYU0MMTM (MMTM), or
     method EYU0MPCT (MPCT) in the CMAS, or method EYU0NMMC (NMMC)
     in the MAS, calling method EYU0XCBR (XCBR).  If the MTRAN
     monitoring termination is due to the CMAS or MAS terminating,
     the storage is only released if that termination is a normal
     termination.
  
  -  When CPSM HISTORY recording is active for the MAS, method
     MMCL calls XCWF to allocate a EYURXCWH block used by MAS
     TRUEs to collect CICS monitoring and statistics data, which
     is used to build the HTASK resource records for the MAS.
  
     When HISTORY monitoring ends in a MAS, either due to a
     PERIODEF ending or if the CMAS or MAS terminates, and the
     retention period for MTRAN elapses, then the storage is
     released by either MMTM or MPCT in the CMAS, or method
     EYU0NHCT (NHCT) in the MAS, calling XCBR.  If the HISTORY
     monitoring termination is due to the CMAS or MAS terminating,
     the storage is only released if that termination is a normal
     termination.
  
  The MAS data cache storage areas described above are only
  anchored in CICS storage in the CMAS and/or the MAS.  If an
  abnormal termination occurs in either a CMAS or MAS, these
  anchors are lost.  If a CPSM CMAS warm start occurs, then the
  storage is orphaned in the MAS data cache until a CPSM CMAS cold
  start next occurs.  If this continually occurs before a CPSM
  CMAS cold start occurs, then the MAS data cache limit can be
  reached, and the errors documented above will occur.
  

Problem conclusion

• To allow the CMAS-MAS comms storage area and the EYURXCWH
  storage areas to be recoverable when a CMAS or MAS is terminated
  without going through normal termination, the following changes
  have been made:
  
  -  A new control block, called the MAS data cache warm start
     block (EYUBXCNB), has been created in the MAS data cache.
     This block will contain information about the CMAS-MAS comms
     storage allocation and the EYURXCWH allocations, that will
     allow these areas to be recoverable in the case of an
     abnormal termination of a CMAS or MAS.
  
  -  A new WORKINI function has been created for XCWF.  This
     function will be called at CMAS initialization by method
     EYU0XLBV (XLBV) and at MAS agent initialization by method
     EYU0NLRT (NLRT).  The function of the calls depends upon
     whether called in a CMAS or MAS, and whether the CMAS start
     is a CPSM CMAS cold or warm start:
  
     -  when called in a CMAS for CPSM CMAS cold start, XCWF will
        allocate the EYUBXCNB in the MAS data cache, and anchor it
        in the MAS data cache definition table (EYURXCDT).  XCWF
        will also allocate CPSM cache list structures for each of
        the four storage areas, which will be used to record
        storage allocations for the four areas.  The tokens for
        the cache lists will be stored in the EYUBXCNB.
  
     -  when called at CMAS initialization for either a CPSM CMAS
        cold or warm start or at MAS agent initialization, XCWF
        will retrieve the pointer to the EYUBXCNB from the MAS
        data cache EYURXCDT, and update the Cache component major
        object environment block (MOEB - EYURXCMO) so that
        subsequent calls to XCWF will have access to the cache
        lists whose tokens are stored in the EYUBXCNB.
  
     Note that the first time a CMAS is restarted with the PTF
     that resolves this APAR, it will create the EYUBXCNB whether
     or not it is a CPSM CMAS cold or warm start.
  
  -  Since CLET does not have access to the Cache component MOEB,
     it will call method EYU0XCCI (XCCI) during CMAS
     initialization to retrieve the pointer to the EYUBXCNB, and
     will store the cache list token it needs from there in its
     working storage, which persists for the life of the CMAS.
  
  The general processing for the MAS data cache storage EYUBXCWH
  structures is as follows:
  
  -  When XCWF is called for existing function WORKFMT to
     allocate a storage area for a MAS, it will check the
     appropriate cache list to see if there is an existing storage
     allocation for that block for that MAS.
  
     -   If so, it will free the previous area and remove its
         allocation information from the cache list.  This is safe
         to do because if XCWF is called to allocate a storage
         instance for a MAS, any previous storage allocation for
         that MAS can no longer be in use.
  
     XCWF will then allocate the storage area and record its
     allocation in the cache list.
  
     Note that the reason a previous allocation is freed and then
     re-allocated is because the size of the area can change from
     one MAS execution to another.
  
  -  A new storage release function has been coded in XCWF,
     WORKREL.  All current callers of XCBR to free the EYUBXCWH
     storage areas, MCTK, MMTM, MPCT, NHCT and NMMC, have been
     updated to call XCWF for this new function instead of XCBR.
     When called for this function, XCWF will remove the cache
     list element documenting the existing storage allocation
     before freeing the storage area.
  
  -  If XCWF encounters an error when attempting to free a
     previous storage allocation, it will issue an exception trace
     and new message EYUXC0025E.  The text of this new message
     will be similar to the following:
  
       EYUXC0025E  Recovery failed for MAS data cache storage area
                   allocated for <masname> in CICSplex <plexname>.
  
  The general processing for the CMAS-MAS comms storage area has
  some differences.  First of all, it is all performed by CLET.
  Secondly, unlike the EYURXCWH areas, an existing allocation
  cannot be guaranteed to not still be in use.  Its processing is
  as follows:
  
  -  When CLET is called at CMAS initialization, it will check if
     there are at least two previous storage allocations.
  
     -  If there are, CLET will determine when the second
        allocation was made.  If made more than an hour ago, the
        first allocation is freed, and its cache list element is
        removed.  If there is a third previous allocation, and it
        was started more than an hour ago, then the second
        allocation is freed and its cache list element is removed.
        This processing will continue until a previous allocation
        that was started less than an hour ago is found, or until
        only one previous allocation remains.
  
     CLET will then allocate the storage area and record its
     allocation in the cache list.
  
  -  When the CMAS terminates normally, CLET will free the cache
     list element for the current allocation, and then will free
     the allocation.  If the CMAS has been active for at least an
     hour, then CLET will free any remaining previous allocations
     and remove their allocation elements from the cache list.
  
  -  If CLET encounters an error when attempting to free a
     previous storage allocation, it will issue an exception trace
     and new message EYUCL0135E.  The text of this new message
     will be similar to the following:
  
       EYUCL0132E  Recovery failed for MAS data cache storage area
                   allocated for this CMAS.
  

Temporary fix

•             *********
              * HIPER *
              *********
  FIX AVAILABLE BY PTF ONLY
  

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

  PI41563

• APAR is sysrouted TO one or more of the following:

  UI33495

Modules/Macros

• EYUMCCLC EYUMCCLE EYUMCCLK EYUMCXCC EYUMCXCE EYUMCXCK EYU0CLET
  EYU0MCTK EYU0MMCL EYU0MMTM EYU0MPCT EYU0NHCT EYU0NLGT EYU0NLRT
  EYU0NMMC EYU0TSQP EYU0TSSC EYU0XCWF EYU0XLBV EYU9NAPU EYU9NAP3
  EYU9NAP4 EYU9TSPU EYU9TSP3 EYU9TSP4 EYU9XCPU EYU9XCP3 EYU9XCP4
  EYU9XCP6
  

Fix information

• Fixed component name

  CICSTS V3 Z/OS

• Fixed component ID

  5655M1500

Applicable component levels

• R50M PSY UI33495

     UP15/12/05 P F512 ½

Fix is available

• Select the PTF appropriate for your component level. You will be required to sign in. Distribution on physical media is not available in all countries.