Re: Shutdown of one instance takes about 1hr

On Jan 31, 5:08 am, Alexander Skwar <alexan..._at_skwar.name> wrote:
> Hello!
>
> Each night, we shutdown the Oracle 9.2.0.4.0 databases on our
> Solaris 9 host. I now noticed, that it takes about 1 hour for
> a certain instance to shutdown:
>
> SQL*Plus: Release 9.2.0.4.0 - Production on Wed Jan 31 02:55:56 2007
>
> Copyright (c) 1982, 2002, Oracle Corporation. All rights reserved.
>
> SQL> Connected.
> SQL> ORA-01013: user requested cancel of current operation
> SQL> Disconnected from Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production
> JServer Release 9.2.0.4.0 - Production
> Database "RACEREF1" shut down.
>
> SQL*Plus: Release 9.2.0.4.0 - Production on Wed Jan 31 03:57:44 2007
>
> To shutdown, I use the Oracle supplied dbshut script, which simply
> does a "shutdown" after having connected with "connect / as sysdba".
> Obviously, I don't run those commands manually, but have a cron
> job setup, which runs dbshut and 15 seconds after it finishes, it
> runs dbshut again. Actually, it does:
>
> su - oracle -c "ORACLE_HOME=$ORACLE_HOME $ORACLE_HOME/bin/lsnrctl stop
> ORACLE_HOME=$ORACLE_HOME $ORACLE_HOME/bin/dbshut"
>
> Anyway, when it shuts down the databases for the 2nd time, it again
> takes about 1 hour to do the shutdown.
>
> But is the shutdown actually successful? I wonder because of the
> message
>
> ORA-01013: user requested cancel of current operation
>
> Further, when I startup the databases again, I see this in the logs:
>
> SQL> Disconnected from Oracle9i Enterprise Edition Release 9.2.0.4.0 - Production
> JServer Release 9.2.0.4.0 - Production
>
> Database "RACEP1" warm started.
> Database "RACEREF1" already started.
>
> Database "RACEREF1" possibly left running when system went down (system crash?).
> Notify Database Administrator.
>
> SQL*Plus: Release 9.2.0.4.0 - Production on Wed Jan 31 05:00:58 2007
>
> The system did not go down. At noon today, I manually shutdown the
> database, and that was very fast (a few seconds). It also came up just
> fine (ie. no message).
>
> What's happening there?

Maybe the 1013 is because the second shutdown happens before the first one finishes, and the hour is how long some timeout takes for the first one to die.

Maybe you need some script like this:

echo "Killing list:"

   ps -ef|grep oracleRACEREP|grep -v grep | awk '{print "kill -9 "$2} ' >> $$kill_oracle_attaches

   ps -ef|grep oracleRACEREP|grep -v grep echo "[End of list]"
if [ -f $$kill_oracle_attaches ]
then chmod +x $$kill_oracle_attaches

     ./$$kill_oracle_attaches
fi
rm $$*

You can put in more grep -v's if you have stuff you know you don't want to kill, and redirect the output to a log to see what you've been killing. Maybe there is some agent or transaction processor running or something.

Not sure what the deal with RACEREF1 is, look in its alert log and see if it is even being told to shut down.

jg

--
@home.com is bogus.
http://www.oracle-base.com/blog/2007/02/02/middle-tier-diet/