Conference galvia::decintact

436.0. "Failure with intact$get by RFA" by KERNEL::BIRKINSHAW (Press Enter to Exit) Wed Apr 09 1997 19:45

Engineering may already be aware of this problem but it has
now been officially logged and I'm dealing with it. I've 
reproduced the trouble statement below. Could anyone give
me any ideas what to try.

thanks
Simon


VMS 6.2 DECintact  V2.1-1 Vax 6440 Midland Bank

This TSWFRXM1 is an offline program and is a server to a queue.  

The queue item in this queue is inserted by another offline program called
TSWFRQM. TSWFRQM creates the queue item using intact$create_queue_item() and it
will put in the RFA and the key of a record that TSWFRXM1 is supposed to read
into the queue item.

TSWFRXM1 will be woken up upon insertion of the queue item, it will 
1. use intact$start_transaction() to start the transaction,
2. use intact$remove_queue_item() with INTACT$M_QUE_FIFO as the removal option, 
3. use the RFA in the queue item to fetch the record by calling intact$get with
   INTACT$M_RB_RFA and INTACT$M_RB_LCK as parameter
4. if this fail, it will use the key to fetch the record by calling intact$get
   with INTACT$M_RB_KEY and INTACT$M_RB_LCK as parameter.  It will then call
   intact$decommit_transaction().
5. if step 3 is successful, it will do some processing, put some records in 
   some other files, queue the queue item into another queue
6. use intact$commit_transaction()
7. hibernate


However, the rfa in the queue item must have been garbled because the status
returned from the intact$get in step 3 is "invalid record's file address
(RFA)".  Also the key must have been set to 0, because we found that TSWFRXM1
was hibernating at the time the problem occurred and it was locking the record
in the file which have 0 as the key.  It was also not processing other queue
items that were queued in the queue.  The INTACT-I-FMG_BAKERRCOD error in the
enclosed log is produced by INTACT.  Later SYSTEM-F-ABORT error was reported
when program intact$version:forcex was used to abort TSWFRXM1.

Logfile Information......

$! -------------------------------------------------------------------------
$!      Vaxcluster Common System Login procedure
$! -------------------------------------------------------------------------
$       set noverify
This is SYCP51
It is   22:40 on 21-MAR 
Task :   TSWFRXM1 PID : 212016ac Function :         start at 03-21-97 22:40:03 
%INTACT-I-FMG_BAKERRCOD, Backout of operation failed with error status of 02E383
7A
%INTACT-F-INT_QBACKOUTFAIL, backout of queue operation failed - suspending
%SYSTEM-F-ABORT, abort
  MIDLPROD     job terminated at 22-MAR-1997 16:08:18.41
    Accounting information:
  Buffered I/O count:             164         Peak working set size:    5698
  Direct I/O count:             39230         Peak page file size:     60892
  Page faults:                   8266         Mounted volumes:             0
  Charged CPU time:           0 00:04:24.60   Elapsed time:     0 17:28:17.06

    This problem is being worked. No further information is available yet.
    
    Joe.

Joe,

I've had Peter Boyes from Midland asking for an update on this call.
Any news yet?

cheers
Simon