PK63836: SHORTAGE IN REAL STORAGE MAY OCCUR WHEN MONITORING DISTRIBUTED DB2 V9 WORKLOADS

Obtain the fix for this APAR.

APAR status

• Closed as program error.

Error description

• When monitoring distributed workloads in DB2 V9, unnecessary
  storage may be used. The problem most often occurs with
  Real Storage and appears to be a result of 24G out of 32G
  being used as Shared Storage.
  This problem may exhibit itself in a CONSOLE DUMP with the
  following SYSTRACE entries:
  018B 00000000  PGM    011 470C7001 96EC8C60  00040011 0000024F
                                                        7D350001
  018B 00000000  PGM    011 470C7001 96EC8C60  00040011 0000024F
                                                        7D351001
  018B 00000000  PGM    011 470C7001 96EC8C60  00040011 0000024F
                                                        7D352001
  ... with ASID(X'018B') being the DB2 Distributed ADDRESS SPACE
  xxxxDIST.  Note the Above The BAR addresses starting with
  0000024F 7D350001 and so forth.  The dirty high half of a
  register (non-zero) is used in error causing reference to
  Above The Bar storage (Shared in this case) as Query Monitor
  scans storage.  There are hundreds of these entries as QM
  repeatedly increments the address, causing a shortage in
  REAL STORAGE as these areas are referenced.
   The PSW from SYSTRACE points to code after an eyecatcher of
  CQMTTEXT, a Query Monitor CSECT.
  Additional Symptoms:  AUX Storage Shortage or excessive
                        growth in AUX Storage usage.  The slots
                        used may indicate higher than normal use
                        of AUX in CQMT and/or DB2 regions.
  Excessive paging with IRA200E AUXILIARY STORAGE SHORTAGE
  or IRA204E nn% AUXILIARY STORAGE ALLOCATED TO uuu
  MSGIRA200E MSGIRA204E
  Additional Verification Steps:    IP RSMDATA HVSHRDATA ALL
                        will show which job(s) have which range(s)
                        of High Virtual Shared storage.  Note that
                        this by itself does not indicate how much
                        of that High Shared Virtual is currently
                        backed by REAL FRAMES.
   z/OS V1.9 & V1.10 --->  IP L 10?+490?    points to the RCE
                           RCE+248 (8bytes) = # of high virt
                                              shared pages
                                              allocated
                           RCE+260 (8bytes) = # of high virt
                                              shared pages
                                              backed in REAL
                           RCE+268 (8bytes) = # of AUX slots
                                              used for high
                                              virt shared
                           RCE+4   (4bytes) = # of REAL frames
                                              available to z/OS
  If the ratio of  (RCE+260):(RCE+4)  is inordinately high,
  often seen to be over 50%, then that is the percentage of
  the Real Frames in the system which is currently backing
  only the High Virt Shared area.   This occurs because CQMT
  touches so many of the high virt shared pages, thus requiring
  z/OS to back them with Real Frames.
  

Local fix

• Do not monitor distributed workloads on DB2 V9.
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: Users of DB2 Query Monitor.                  *
  ****************************************************************
  * PROBLEM DESCRIPTION: When monitoring distributed workloads i *
  *                      storage may be used. The problem most o *
  *                      Real Storage and appears to be a        *
  *                      result of 24G out of 32G being used as  *
  *                      Shared Storage. This problem may        *
  *                      exhibit itself in a CONSOLE DUMP with   *
  *                      the following SYSTRACE entries:         *
  *                      018B 00000000  PGM    011               *
  *                       470C7001 96EC8C60  00040011 0000024F   *
  *                       7D350001                               *
  *                      018B 00000000  PGM    011               *
  *                       470C7001 96EC8C60  00040011 0000024F   *
  *                       7D351001                               *
  *                      018B 00000000  PGM    011               *
  *                       470C7001 96EC8C60  00040011 0000024F   *
  *                       7D352001                               *
  *                      ... with ASID(X'018B') being the DB2    *
  *                      Distributed ADDRESS SPACE xxxxDIST.     *
  *                      Note the Above The BAR addresses        *
  *                      starting with 0000024F 7D350001 and so  *
  *                      forth.  The dirty high half of a        *
  *                      register (non-zero) is used in error    *
  *                      causing reference to Above The Bar      *
  *                      storage (Shared in this case) as Query  *
  *                      Monitor scans storage.  There are       *
  *                      hundreds of these entries as QM         *
  *                      repeatedly increments the address,      *
  *                      causing a shortage in REAL STORAGE as   *
  *                      these areas are referenced. The PSW     *
  *                      from SYSTRACE points to code after an   *
  *                      eyecatcher of CQMTTEXT, a Query         *
  *                      Monitor CSECT.                          *
  ****************************************************************
  * RECOMMENDATION: APPLY the PTF.                               *
  ****************************************************************
  Code has been changed to correct this condition.
  

Problem conclusion

• APPLY the PTF.
  

Temporary fix

• *********
  * HIPER *
  *********
  

Comments

APAR Information

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

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

  UK35628 UK35627

Modules/Macros

•    CQMTTEXT
  

Fix information

• Fixed component name

  DB2 QUERY MONIT

• Fixed component ID

  5655E6701

Applicable component levels

• R220 PSY UK35627

     UP08/04/21 P F804 Ž

• R221 PSY UK35628

     UP08/04/21 P F804 Ž

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.