I decided to split this off the other thread.
During my performance testing with PDQ on I'm running into a computation delta from non-PDQ runs causing the entire process to bail when math is done wrong. I've found the likely issue as a bug that is fixed in FC8. Instead of going to that from FC7 I decided to go ahead and go to FC10.
Unfortunately, it isn't going well. I've actually never seen Informix upgrades fail, but this one is just dying hard. On two different instances that are radically different. Here's some of the stuff from the log.
Lots of this even though nothing can even connect to it at this point:
------------------------------------------------
13:39:09 listener-thread: err = -1834: oserr = 0: errstr = : System database rebuild in progress
13:39:09 listener-thread: err = -1834: oserr = 0: errstr = Sysmaster Rebuild Disabling Client Connections: System database rebuild
in progress
------------------------------------------------
This ends the log:
------------------------------------------------
13:39:16 Error building 'sysmaster' database.
13:39:16 See '/tmp/buildsmi.29605'.
13:39:16 Assert Failed: build_smi failed
13:39:16 IBM Informix Dynamic Server Version 14.10.FC10
13:39:16 Who: Session(102, informix@uatinf21, 0, 0x82c8e5e8)
Thread(150, build_smi, 82c5bae0, 12)
File: rspseudo.c Line: 11255
13:39:16 Results: Cannot build/check 'sysmaster' database
13:39:16 stack trace for pid 29616 written to /opt/informix/tmp/af.47eced4
13:39:16 See Also: /opt/informix/tmp/af.47eced4
13:39:21 See Also: /opt/core
13:39:21 Fatal error in ADM VP at mt_fn.c:15161
13:39:21 Unexpected virtual processor termination: pid = 29616, signal 11 received.
13:39:21 PANIC: Attempting to bring system down
------------------------------------------------
This is what's in that /tmp file mentioned:
------------------------------------------------
If sysmaster database exists, drop it.
1834: System database rebuild in progress
This file is the output of the dbaccess command in the shell script,
$INFORMIXDIR/etc/buildsmi.
This script is run when Informix Dynamic Server is first
initialized. It builds the database and tables which support the
System Monitoring Interface. If you were directed to see this file,
a problem probably occurred while building the database and/or its
tables and views. See below for more information regarding any problems.
Problems may have occurred if the server was shutdown after initialization
but before the entire database was created. If this was the case,
login as 'informix', rerun 'oninit', and wait for the message
"'sysmaster' database created successfully"
to appear in the message logfile before shutting down. Once it is
created successfully, it should never need to be recreated.
1834: System database rebuild in progress
Error in line 26
Near character position 1
1834: System database rebuild in progress
Error in line 28
Near character position 1
1834: System database rebuild in progress
Error in line 30
Near character position 1
1834: System database rebuild in progress
Error in line 31
Near character position 1
1834: System database rebuild in progress
Error in line 34
Near character position 5
1834: System database rebuild in progress
Error in line 42
Near character position 5
1834: System database rebuild in progress
Error in line 43
Near character position 5
1834: System database rebuild in progress
Error in line 44
Near character position 5
----------------------------------------------------------------------------
Error -1834 seems suspicious....I dont' see any DESCRIBE in these IBM scripts:
-1834 DESCRIBE INPUT cannot be used with static SQL statements.
The DESCRIBE INPUT statement can be used to provide information about dynamic
parameters in the WHERE clause, in subqueries, and in other syntactic contexts
where the old form of the DESCRIBE statement cannot provide information.
For more information about using the DESCRIBE INPUT statement, see the IBM
Informix Guide to SQL: Syntax.
I've re-built the directory twice....I don't think it is a install problem. Anyone ever seen this? From looking at the buildsmi script, I suspect it is failing to even drop the sysmaster database.
I also re-started both with FC7 successfully. I re-built sysmaster for both instances in FC7 and restarted the migration.....same error.
On a side note, I wonder if the sysmaster database is corrupt and that is what's making the PDQ go crazy.....
------------------------------
Jared Heath
------------------------------