IT27616: QUERIES PERFORMING INDEX SELF JOIN MAY NOT YIELD FREQUENTLY ENOUGH LEADING TO BLOCKED CHECKPOINT REQUESTS

APAR status

• Closed as program error.

Error description

• Certain types of SQL query that use an index self join on large
  tables where the index pages are held in the buffer pool may not
  yield frequently enough during their execution. This could
  result in a blocked checkpoint request if another thread is in
  critical section and waiting on the result of an I/O request and
  that result needs to be processed by the kaio thread running on
  the same CPU VP as the non-yielding query.
  
  A stack of the running thread may be similar to:
  
  btfind
  btsearch
  kposition
  rsstart
  fmstart
  nextlkeyrg
  startsubq
  scan_next
  getrow
  fetchrow
  exfetch
  sql_nfetch
  sq_nfetch
  sqmain
  
  As the thread is in a running state the top few functions may or
  may not be present. The key indicator for this problem is the
  function nextlkeyrg which may perform a loop operation.
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * Users of IDS prior to 12.10.xC13.                            *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * Queries performing index self join may not yield frequently  *
  * enough leading to blocked checkpoint requests.               *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  

Problem conclusion

• Fixed in IDS 12.10.xC13.
  

Temporary fix

Comments

APAR Information

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

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

Fix information

• Fixed component name

  INFORMIX SERVER

• Fixed component ID

  5725A3900

Applicable component levels