RE: Shutdown takes 20+ minutes

I don't think using abort = corruption potential.

If you have a cluster for high-availability, how can you afford to do a weekly bounce and cold backup (decidedly low-availability practices)? Why not leave it up and do online backups?

If you absolutely must have a cold backup, then the [checkpoint] - abort - restrict - immediate sequence is the right one for you to be using.

I realize you want immediate to work, but there are so many reasons that it can get stuck, it just doesn't seem to make sense to me to ever rely on it finishing in any kind of reasonable time.

Among the many classic causes of slowness with IMMEDIATE are cleaning up lots of temp segments and rolling back one or more large transactions. Maybe look in to the DROP_SEGMENTS event or loko for large uncommitted transactions.

Finally I believe there are events you can set in 8i to diagnose hanging IMMEDIATE. Why not just try a 10046 level 12 in your SYSDBA session when you are trying to shut down. Then you can tail the trace file in another window and see what wait events it is waiting on.

--
Jeremiah Wilton
http://www.speakeasy.net/~jwilton

On Wed, 22 Oct 2003, Tortorelli, Mary Jo wrote:



> We do shutdown immediate as the standard weekly cycle/cold-backup of

> this production db.  The cluster failover is part of the procedure

> and is a test - it's not a real failover.  We don't want to

> routinely shutdown abort and increase corruption potential if it's

> not necessary.

>

> We want a clean shutdown for the backup and I'm trying to figure out

> why it takes so long since it's messing up the fail-over test that

> follows.





> -----Original Message-----

> Sent: Tuesday, October 21, 2003 7:24 PM

> To: Multiple recipients of list ORACLE-L

>

> I bet you can get that 4 minutes down to 1 if you dispense with the

> down/up/down and just do checkpoint/abort/failover/startup.

>

> Why bother with "immediate" for a cluster failover?  Isn't it just a

> waste of time?

>

> On Tue, 21 Oct 2003, Tortorelli, Mary Jo wrote:

>

> > An 8.1.7.4 production database on HP-UX 11.0 running Apps 11.5.7 takes long to shutdown and is causing cluster failover testing to time out.

> >

> > Upon shutdown immediate it takes 1-3 minutes for the DATABASE DISMOUNTED, DATABASE CLOSED, and Archival stopped messages in the alert log but then it takes another 15+ minutes for the ORACLE database shutdown message to return to sqlplus.  During this time, a ps -ef shows no background or ghost database processes - only the sqlplus process.

> >

> > If a shutdown abort, startup restrict, shutdown is done after five minutes, the shutdown after the abort/startup takes 4 minutes total.

> >

> > Does anyone know what Oracle does between DATABASE CLOSED, ARCHIVE STOPPED and the ORACLE database shutdown message?   Has anyone else run into this?


-- 
Please see the official ORACLE-L FAQ: http://www.orafaq.net
-- 
Author: Jeremiah Wilton
  INET: jwilton_at_speakeasy.net

Fat City Network Services    -- 858-538-5051 http://www.fatcity.com
San Diego, California        -- Mailing list and web hosting services
---------------------------------------------------------------------
To REMOVE yourself from this mailing list, send an E-Mail message
to: ListGuru_at_fatcity.com (note EXACT spelling of 'ListGuru') and in
the message BODY, include a line containing: UNSUB ORACLE-L
(or the name of mailing list you want to be removed from).  You may
also send the HELP command for other information (like subscribing).