IBM Support

PM61982: LOCKS NOT BEING HELD AFTER DYNAMIC BACKOUT FAILURE

A fix is available

Subscribe

You can track all active APARs for this component.

 

APAR status

  • Closed as program error.

Error description

  • Following situation. Database is PHIDAM with an SI based on some
    part of the root segment. The SI is defined in RECON as
    non-recoverable. A BMP is started on IMN2 which deletes some
    root segments (leads to deletes in the SI) and then is halted
    by a WTOR. Multiple log swicthes are now forced with SWI OLDS
    until the message 'SLDS required  for  backout' is issued. Now
    the BMP is abended using STO REG ABDUMP.
    The  first of the deleted root segments is backouted but the
    attempt to backout the corresponding update to the SI fails
    because the necessary records are not available on the SLDS
    (non-recoverable !).
    Dynamic backout fails and the database now has status STOPPED,
    BACKOUT and  Backout Needed is on in RECON. Any attempt to
    start a further BMP against this partition on IMN2 abends U3303
    because of the status STOPPED, BACKOUT. Any attempt to start
    the partition fails   becuase it causes a retry of the dynamic
    backout and that, of course, fails for the same reason as
    originally.
    So far so good. However, if a read BMP is now started on IMN1
    using PROCOPT=A it is possible to read all root segments from
    the beginning to the end of the partition ! This should not be
    possible since IRLM  locks should still exist for some blocks
    as a result of the failed Dynamic Backout.
    

Local fix

Problem summary

  • ****************************************************************
    * USERS AFFECTED: All IMS V11 users who data share and update  *
    *                 High Availability Large Databases, HALDBs.   *
    ****************************************************************
    * PROBLEM DESCRIPTION: After a backout failure on one IMS, a   *
    *                      partition can still be accessed with a  *
    *                      PROCOPT=A PCB on a sharing IMS. This    *
    *                      should not be allowed.                  *
    ****************************************************************
    * RECOMMENDATION: INSTALL CORRECTIVE SERVICE FOR APAR/PTF      *
    ****************************************************************
    User has a PHIDAM database that has a dependent segment under a
    root that is indexed via a PSINDEX database.  The PSINDEX
    database is defined as non-recoverable in the RECON.  The
    databases are globally shared; e.g. data shared.
    Two IMS systems are sharing the databases when a BMP is run
    on one IMS that deletes several root segments which result
    in updates to the PSINDEX. The BMP does not terminate and
    goes into a wait state. Multiple log switches occur. A display
    of the OLDS shows:
    DFS000I      SLDS REQUIRED FOR BACKOUT - RGN 00001, SLDSREAD ON.
    
    The BMP is then terminated with a /STOP REGION 1 ABDUMP command.
    During ABDUMP processing the following messages are received:
    DFS987I ERROR READING SLDS DURING BACKOUT WITHIN PSB=PSBNAME
    DUE TO NON-RECOVERABLE DB IMSID
    DFS979I BATCH MESSAGE PROGRAM PSBNAME WAS ACTIVE AT TIME OF
    FAILURE IMSID
    DFS975W  NON-RECOVERABLE DBD=PSINDEX  WITHIN PSB=PSBNAME STOPPED
    DUE TO UNKNOWN BACKOUT STATUS IMSID
    DFS981I  DBD=PARTNAM1 WITHIN PSB=PSBNAME STOPPED DUE TO BACKOUT
    FAILURE IMSID
    DFS981I  DBD=PARTNAM2 WITHIN PSB=PSBNAME STOPPED DUE TO BACKOUT
    FAILURE IMSID
    DFS980I 16:45:17 BACKOUT PROCESSING HAS ENDED FOR PSBNAME IMSID
    
    On this IMS a display of the related partitions show a status
    of: STOPPED, BACKOUT, ALLOCS
    
    Any attempt to access the stopped partitions, on this IMS,
    results in the following messages and an ABENDU3303.
    DFS3303I PSB PSBNAME  PCB DBDNAME DBD PARTNAME PARTUNAV  JOBNAME
    JOBNAME  RGN 00001 IMSID
    DFS0832I ABEND U3303 REASON CODE 4020 PARTITION PARTNAME IMSID
    
    A /START DB for the stopped partitions fails because dynamic
    backout is part of the command processing and it fails for the
    same reasons as previously described.
    
    The above is all correct processing. The problem is a PROCOPT=A
    BMP is allowed to access a stopped partition on the sharing IMS
    when the partition should not be available. All locks were
    released at the end of the STOP REGION ABDUMP processing and
    database (partition) integrity should be protected by flagging
    the partition as stopped and in error.
    A display of the stopped partitions on the sharing IMS do not
    show STOPPED, only ALLOCS.
    A RECON listing shows BACKOUT NEEDED =ON for the partitions that
    are marked as recoverable. The non-recoverable PSINDEX shows
    BACKOUT NEEDED =OFF.
    
    Missing notifies are the reason the recoverable partitions are
    not flagged as stopped, DDIRNOSC, and in error, DDIRDBIE.
    
    DFSRBOI0 is called by DFSFXC50 during /STOP REGION ABDUMP
    processing to do dynamic backout.  Part of the backout failure
    processing is to notify all sharing subsystems that the
    databases, partitions in this case, are in error and should be
    stopped.  The sending of the notifies were skipped because flag
    DDIRGLCK was not on in the recoverable partition DDIRs.
    DDIRGLCK should be set on during authorization of the partitions
    but code in DFSDBAU0 after label ALOOP2 checked for partitioned
    databases and bypassed the code that set the global locking
    required flag, DDIRGLCK.
    
    Additional Keywords: MSGDFS3303I STACMD CMDSTA MSGDFS0832I
                         MSGDFS000I MSGDFS987I MSGDFS979I MSGDFS975W
                         MSGDFS981I MSGDFS980I STOCMD CMDSTO REG
    

Problem conclusion

  • GEN:
    POSTREQ PM73440
    KEYWORDS:
    
    *** END IMS KEYWORDS ***
    Code has been added to the partition authorization routine,
    OLICCALL, in DFSDBAU0 after label DSGSDONE to set DDIRGLCK
    when the lock scope flags indicate global locking.
    

Temporary fix

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

Comments

  • REPINNED RP12/09/21 (ATXT) TO ADD POSTREQ PM73440 INFO.
     **** PE12/09/21 PTF IN ERROR. SEE APAR PM73440 FOR DESCRIPTION
    ×**** PE12/09/21 FIX IN ERROR. SEE APAR PM73440  FOR DESCRIPTION
    

APAR Information

  • APAR number

    PM61982

  • Reported component name

    IMS V11

  • Reported component ID

    5635A0200

  • Reported release

    100

  • Status

    CLOSED PER

  • PE

    NoPE

  • HIPER

    YesHIPER

  • Special Attention

    NoSpecatt / Xsystem

  • Submitted date

    2012-04-06

  • Closed date

    2012-06-26

  • Last modified date

    2012-10-12

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

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

    PM64598 UK79837

Modules/Macros

  • DFSDBAU0
    

Fix information

  • Fixed component name

    IMS V11

  • Fixed component ID

    5635A0200

Applicable component levels

  • R100 PSY UK79837

       UP12/07/03 P F207 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.

[{"Business Unit":{"code":"BU054","label":"Systems w\/TPS"},"Product":{"code":"SG19M","label":"APARs - z\/OS environment"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"100","Edition":"","Line of Business":{"code":"","label":""}},{"Business Unit":{"code":"BU048","label":"IBM Software"},"Product":{"code":"SSCVRBJ","label":"System Services"},"Component":"","ARM Category":[],"Platform":[{"code":"PF025","label":"Platform Independent"}],"Version":"100","Edition":"","Line of Business":{"code":"","label":""}}]

Document Information

Modified date:
12 October 2012