PI85860: NEW FUNCTION: PROVIDE COLUMN DATA REPAIR EXIT FOR CLASSIC CDC

Obtain the fix for this APAR.

APAR status

• Closed as new function.

Error description

• The server provides configuration parameters to control what
  level of validation is done on native data and how numeric
  conversion errors are handled. These configuration settings
  control how badly formed numeric data is repaired when a badly
  formed native data value is mapped to an SQL numeric data column
   for replication or publishing.
  The server, by default, will repair numeric SQL columns when the
  badly formed native data cannot be successfully converted to a
  DECIMAL, SMALLINT or INTEGER column.  This is done to prevent
  failures in the event of badly formed native data.  The default
  repair values for the SQL numeric column is -9999?99 for the
  precision and scale of the SQL column.
  A column repair exit is required to provide the user with a
  mechanism to customize the repair value for a numeric SQL column
  that fails validation or cannot be converted to the SQL numeric
  data type.  Using the column repair exit, you can implement your
  own rules to determine what repair value should be provided when
  badly formed data is encountered for a specific Table, Column
  and or Subscription.
  

Local fix

• N/A
  

Problem summary

• ****************************************************************
  * USERS AFFECTED: InfoSphere Classic Change Data Capture for   *
  *                 z/OS                                         *
  ****************************************************************
  * PROBLEM DESCRIPTION: A column data repair exit for           *
  *                      Classic CDC is needed.                  *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  The server provides configuration parameters to control what
  level of validation is done on native data and how numeric
  conversion errors are handled.  These configuration settings
  control how badly formed numeric data is repaired when a
  badly formed native data value is mapped to an SQL numeric
  data column for replication or publishing. A column repair
  exit is required to provide the user with a mechanism to
  customize the repair value for a numeric SQL column that
  fails validation or cannot be converted to the SQL numeric
  data type.
  

Problem conclusion

Temporary fix

Comments

• Classic is providing a column repair exit to give the user
  a mechanism to customize the repair value for a numeric SQL
  column that fails validation or cannot be converted to the SQL
  numeric data type. Using the column repair exit, you can
  implement your own rules to determine what repair value should
  be provided when badly formed data is encountered for a
  specific Table, Column and/or Subscription.
  The column repair exit will be called after the server has
  determined that the native data being mapped to the SQL column
  fails validation or generates a numeric conversion error and
  needs to be repaired.
  Developing a Column Repair Exit
  The column repair exit runs in the Classic server address
  space. A copy of the exit is loaded by each subscription
  that is actively refreshing data, replicating or publishing
  in a Change Data Capture environment. The exit is called in
  key 8 problem program state. The exit is loaded from an
  APF-authorized library so that it can enter supervisory state
  and perform authorized functions, but that mode of operation
  is not recommended. The exit must reside in a PDSE load
  library.
  Classic provides a generic sample column repair exit that is
  written in COBOL and distributed as SCACSAMP(CECCREPX).
  This exit implements a basic rule to return a value of
  zero for any SQL column that has detected badly formed data
  that cannot be mapped to an SQL numeric column.
  The column repair exit is configured by setting the
  COLUMNREPAIREXIT global configuration parameter.
  You identify the name of the load module that you want called
  when badly formed data is detected that cannot be mapped to
  an SQL numeric column. The named exit load module is loaded
  at the time a subscription starts replicating, publishing or
  at refresh.  If replication is already active for a
  subscription, you need to stop and restart replication to
  enable the exit for a subscription.
  The exit is invoked in a language environment and can use any
  LE services. The exit is intended to be developed by using a
  high-level language such as COBOL, HLASM or C. The column
  repair exit must be compiled and linked as reentrant.
  When the exit is called, the exit has control of that service
  (TCB) until control is returned.  Time consuming calls should
  be minimized within the exit, as such calls will impact
  replication for all subscriptions.  Additionally, no special
  abend or ESTAE protection is in place, so if the exit
  abnormally terminates, default exception processing for the
  language environment is invoked.
  Column Repair Exit Processing
  When enabled, the column repair exit is called for the
  following initialization, termination and validation
  operations.
  Initialization:
  The column repair exit is called for initialization
  processing when a subscription is first refreshed or start
  mirroring operation.  During the initialization call the
  exit should verify the version of the exit parameter list to
  ensure it matches the version used by the server.  The exit
  can also use the initialization call to allocate working
  storage and return an address to this storage in the user
  double-word parameter that is passed on subsequent calls
  to the exit.   Any values stored in the user double-word
  parameter will be preserved for the subscription until
  termination processing is invoked when replication is
  stopped for the subscription
  The exit must communicate the version of the exit it is
  using by returning that value back to the server in the
  Column Repair Exit Parameter List Version parameter passed
  to the exit.
  Termination:
  The column repair exit is called for termination processing
  when replication is stopped for subscription.  During the
  termination call the exit must free any resources it
  previously allocated.
  Validation:
  The exit is called for validation processing after the server
  has determined that the native data being mapped to the SQL
  column fails validation or generates a numeric conversion
  error and needs to be repaired.   The exit will be called for
  each column that has badly formed data and needs to be
  repaired.   The exit will be called for after image processing
  and for before image processing when configured.  The exit
  will determine how to repair the badly formed data and
  communicate this back to the server.
  Parameters that are passed to the Column Repair exit
  1. Parameter list version
     2 bytes long
     Input parameter
     Description: Informs the exit what version of the exit
     parameter list it communicates with. This parameter is
     provided for future extension. The only valid value is 1.
  
  2. Column repair exit parameter list version
     2 bytes
     Output parameter
     Descr: The exit is expected to inform the caller what
     version of the exit parameter list it communicates with.
     The only valid value is 1.
  
  3. User double words
     8 bytes
     Input/Output
     Descr:  The user double word is initialized to binary
     zeroes before the initialization call. The exit can use
     this area for any purpose. The value that the exit puts
     in this area is returned to the exit for each subsequent
     call for the given subscription until the replication is
     stopped for the subscription.
  
  4. Function code
     4 bytes
     Input
     Descr: One of the following functions for which the exit
     is being invoked:
     CREX-FUNC-INIT, CREX-FUNC-REPAIR, CREX-FUNC-TERM
  
  
  5. User Name
     32 bytes
     Input
     Descr: Left-justified, space padded
     User Name associated with operation.
     Will be spaces in Change Data Capture
  
  6. Subscription Name
     64 bytes
     Input
     Descr: Left-justified, space padded
     Subscription Name associated with operation.
     Will be spaces in non-Change Data Capture operations
  
  7. Source System ID
     8 bytes
     Input
     Descr: Left-justified, space padded
     8 byte unique identifier for subscription.
     Will be spaces in non-Change Data Capture operations
  
  8. Database Type
     8 bytes
     Input
     Descr: Left-justified, space padded
     Database Type associated with operation.
     Values: $IMS for IMS data source, $VSAM for VSAM data source
  
  9. Table Name
     20 bytes
     Input
     Descr: Left-justified, space padded
     Table or View Name associated with operation.
  
  10. Schema Name
     8 bytes
     Input
     Descr: Left-justified, space padded
     Schema Name for the table associated with operation.
  
  11. Column Name
     32 bytes
     Input
     Descr: Left-justified, space padded
     Column Name that failed validation or conversion.
  
  12. Native Data Type
     1 bytes
     Input
     Descr: The Native Datatype that failed validation or
     conversion.   Valid values are:
     C   Mixed mode character data. Fixed Length
     P   Packed decimal data
     D   Floating point data.
     F   32-bit signed binary value where the sign is in the
         high order bit.
     H   16-bit signed binary value where the sign is in the
         high order bit.
     V   Variable mixed-mode character data, data is preceded
         by a 16-bit signed binary number that identifies the
         actual length of the data.
     B   Fixed length binary data.
  
  13. Native Data Signed
     1 bytes
     Input
     Descr: Value indicating if Native numeric data is signed:
     Y - data is signed
     N - data is unsigned
  
  14. Native Data Length
     4 bytes
     Input
     Descr: The length of the Native Data
  
  15. Native Data Address
     4 bytes
     Input
     Descr: The address of the Native Data that failed
     validation or conversion
     The exit cannot modify the contents of this area
  
  16. SQL Data Type
     4 bytes
     Input
     Descr: The SQL data type for the data pointed to in SQL
     Data Address. Possible Values:
     SQL-TYPE-SMALLINT, SQL-TYPE-INTEGER, SQL-TYPE-DECIMAL
  
  17. SQL Data Length
     4 bytes
     Input
     Descr: The length of the SQL data type for the data
     pointed to in the  SQL Data Address.
  
  18. SQL Data Precision
     4 bytes
     Input
     Descr: The precision of the SQL data type for the data
     pointed to in SQL Data Address.
  
  19. SQL Data Scale
     4 bytes
     Input
     Descr: The Scale of the SQL data type for the data pointed
     to in the SQL Data Address.
  
  20. SQL Data Address
     4 bytes
     Input/Output
     Descr: The address of the SQL Data that failed validation
     or conversion.  The value will contain the server default
     repair value of -999's.
     The exit can modify the contents of this area to change
     the repair value for the operation.
  
  21. SQL Data Null Indicator
     1 bytes
     Input
     Descr: Value indicating if the SQL Data supports NULL
     Y - data supports NULL
     N - data cannot be NULL
  
  22. SQL Action
     4 bytes
     Input
     Descr: Data Server action associated with the operation:
     CREX-ACT-NONE
     CREX-ACT-DELETE
     CREX-ACT-INSERT
     CREX-ACT-UPDATE
     CREX-ACT-QUERY
     CREX-ACT-AFTR-DELETE
     CREX-ACT-BFOR-DELETE
     CREX-ACT-AFTR-INSERT
     CREX-ACT-BFOR-INSERT
     CREX-ACT-AFTR-UPDATE
     CREX-ACT-AFTR-UPDATE
  
  23. Return code
     4 bytes
     Output
     Descr: One of the following values:
     CREX-DEFAULT :
     CREX-SUCCESS: User Server default behavior
     CREX-REPAIRED: Exit provided new repair value in buffer
          pointed to by SQL Data Address
     CREX-FAILED: Exit requests that the operation is terminated
          with a failure
  
  24. Reason code
     4 bytes
     Output
     Descr: The only valid reason code is: CREX-REASON-0
  
  25. Diagnostic string
     80 bytes
     Output
     Descr: Always set to low values on entry to the apply exit.
     The exit can use this area to return a null terminated
     diagnostic string.   The maximum length of the message is
     79 bytes allowing for the 1 byte null string terminator.
  .
  The IBM Knowledge Center will be updated with more details
  about this update for Version 11.3.
  See the link below:
  http://www-01.ibm.com/support/docview.wss?uid=swg27039728&wv=1
  

APAR Information

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

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

  UI49806 UI49807 UI49808 UI49809 UI49810 UI49811 UI49812 UI49813
  UI49814 UI49815

Modules/Macros

•    CACCHICT CACENGCT CACJPNCT CACZHOCT CECAPI2
  CECAVSM1 CECCABHD CECCADAB CECCADAI CECCADV  CECCAIMC CECCAIMM
  CECCAIMO CECCAIMS CECCAOPS CECCAPEX CECCAPIC CECCASHL CECCASSI
  CECCAVMM CECCBCFG CECCBCP1 CECCBNEW CECCBTIN CECCBUFF CECCCATL
  CECCCATU CECCCDCP CECCCHDL CECCCHSH CECCCNFG CECCCNTL CECCCOMM
  CECCCONN CECCCONV CECCCREX CECCCSIM CECCCSVS CECCDBDL CECCDCFG
  CECCDD05 CECCDESC CECCDIAG CECCDISP CECCDOPS CECCDRAI CECCDSSI
  CECCDUMP CECCEFSM CECCEIF  CECCENDE CECCENGN CECCESSI CECCFAHD
  CECCFASH CECCFGUT CECCFILE CECCFLDP CECCGET  CECCHNDL CECCIDMD
  CECCIDRA CECCIIMS CECCILE  CECCIMSA CECCIMSI CECCINFO CECCINIT
  CECCIOPS CECCIRES CECCIRRS CECCITCA CECCITCI CECCITIC CECCITMA
  CECCITMI CECCITMS CECCITSC CECCITSS CECCIWLM CECCLDAM CECCLEX
  CECCLRFU CECCLRIC CECCLRID CECCLRIE CECCLRIH CECCLRII CECCLRIM
  CECCLRIO CECCLRIQ CECCLRIR CECCLRIS CECCLRIT CECCLRSM CECCLRSS
  CECCLRVC CECCLRVD CECCLRVE CECCLRVH CECCLRVI CECCLRVL CECCLRVQ
  CECCLRVR CECCLRVS CECCLRVT CECCMADA CECCMAIN CECCMAMD CECCMAMT
  CECCMAM1 CECCMAM2 CECCMAM3 CECCMAM4 CECCMAPF CECCMAUT CECCMCFG
  CECCMDB2 CECCMDCM CECCMEMO CECCMIDM CECCMIMS CECCMISC CECCMSEQ
  CECCMSGS CECCMSHL CECCMTO  CECCMVSM CECCNFG1 CECCNLOG CECCNLS
  CECCONV1 CECCONV2 CECCOPFU CECCOPXM CECCPAAD CECCPACM CECCPADB
  CECCPAEV CECCPAPF CECCPASC CECCPASS CECCPAS2 CECCPATG CECCPAUT
  CECCPFIX CECCPFS  CECCPLOG CECCPSHL CECCPUMN CECCQGEN CECCQNOD
  CECCQOPS CECCQPIN CECCQSSI CECCREPC CECCREPM CECCREPU CECCREPX
  CECCREP2 CECCREP3 CECCRES  CECCROBJ CECCROC  CECCROLE CECCRSLV
  CECCRTNS CECCRUNT CECCSHLB CECCSHLQ CECCSHLR CECCSHL2 CECCSHR
  CECCSHSL CECCSKA  CECCSLS  CECCSMOS CECCSOPS CECCSPDA CECCSPDI
  CECCSPDR CECCSPNT CECCSPSM CECCSPXL CECCSP62 CECCSQER CECCSRVC
  CECCSRVR CECCSSUB CECCSTMT CECCSTSK CECCSVC  CECCTBRI CECCTCBM
  CECCTCMN CECCTDIS CECCTDM  CECCTFLT CECCTLOG CECCTRAN CECCTSHL
  CECCTSPI CECCTSSI CECCTSUB CECCT2AS CECCT2CD CECCT2TS CECCUOPS
  CECCUOR  CECCUSER CECCUSH1 CECCUTIL CECCVBRI CECCVMAP CECCVSAM
  CECCVSHL CECCVSMS CECCVSSI CECCVUTL CECCWQUE CECCWRTR CECCWSHL
  CECCXBLD CECCXCOM CECCXEXT CECCXLE1 CECCXOPS CECCXUTL CECCYFMT
  CECCZSSI CECC2CAF CECC2QSI CECC2SSI CECDUMP1 CECENCR  CECFILE1
  CECMCFG1 CECMISC1 CECOPER1 CECREPC1 CECREPL2 CECREPL3 CECREPL4
  CECSHLL2 CECSHLL3 CECSHLL4 CECSHLL5 CECSPRS1 CECSQL1  CECTCBM1
  CECTRAN1 CECUTIL1 CECVHS   HABIB30J
  

Fix information

• Fixed component name

  CLASSIC FED & E

• Fixed component ID

  5697I8200

Applicable component levels

• RB3A PSY UI49807

     UP17/08/29 P F708

• RB3G PSY UI49808

     UP17/08/30 P F708

• RB3H PSY UI49809

     UP17/08/30 P F708

• RB3J PSY UI49810

     UP17/08/30 P F708

• RB30 PSY UI49806

     UP17/08/29 P F708

• RB31 PSY UI49811

     UP17/08/29 P F708

• RB35 PSY UI49812

     UP17/08/29 P F708

• RB36 PSY UI49813

     UP17/08/29 P F708

• RB37 PSY UI49814

     UP17/08/29 P F708

• RB39 PSY UI49815

     UP17/08/29 P F708

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.