IT03513: TIMING PROBLEM WITH READAHEAD AND SQLEXEC THREADS CAN LEAVE COMMITTED READ LAST COMMITTED READERS WAITING ON LOCKS

APAR status

• Closed as unreproducible in next release.

Error description

• When using isolation level committed read last committed, a
  thread should not wait indefinitely on a lock, it should get the
  locked row from the logical logs.  However, with this timing
  issue, the attempt to try to do the read ahead request to get
  the last committed row from the logical logs can run into
  issues, which then causes the code to just have the sqlexec
  thread act as if it had done a set lock mode to wait, and just
  then wait on the lock until it's released.
  
  The stack of the committed read last committed isolation reader
  would look something like this:
  
  mt_yield
  lkwait
  lkmanagex
  krow
  read_record_lc
  rsseqscan_read
  fmseqscan_read
  readseq_batch
  gettupl
  scan_next
  next_row
  get_first_row_from_producer
  group_open
  prepselect
  open_cursor
  sql_open
  sq_open
  sqmain
  
  onstat -u output would show user waiting on a lock (also not
  lock timeout is -1 which would be as if the session had done set
  lock mode to wait):
  
  Userthreads
  address          flags   sessid   user     tty      wait
  tout locks nreads   nwrites
  66611448         L--PR-- 251      user1  -        6f464600
  -1   1     0        0
  
  onstat -g ses output would show committed read last committed
  isolation level but lock mode here is reported as not wait:
  
  Sess       SQL            Current            Iso Lock       SQL
  ISAM F.E.
  Id         Stmt type      Database           Lvl Mode       ERR
  ERR  Vers  Explain
  251        SELECT         clear              LC  Not Wait   0
  0    9.24  Off
  
  Current statement name : slctcur
  
  Current SQL statement (3) :
    select count(*) from t1 where c4 = 40 and c5 is null
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  * users using last committed read isolation                    *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  * See Error Description                                        *
  ****************************************************************
  * RECOMMENDATION:                                              *
  * Upgrade to IDS 12.10.xC4 or above.                           *
  ****************************************************************
  

Problem conclusion

• Fixed in IDS 12.10.xC4.  idsdb00499637 fixed the issue
  

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

• RA10 PSN

     UP

• RA10 PSY

     UP

• RB10 PSN

     UP

• RB10 PSY

     UP

• RB50 PSN

     UP

• RB50 PSY

     UP

• RB70 PSN

     UP

• RB70 PSY

     UP

• RC10 PSN

     UP

• RC10 PSY

     UP