Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Mailing Lists -> Oracle-L -> Re: When do system events become effective?

Re: When do system events become effective?

From: Uwe Küchler <uwe_at_kuechler.org>
Date: Fri, 4 May 2007 17:56:51 +0200 (CEST)
Message-ID: <20118.193.24.34.77.1178294211.squirrel@kuechler.org>


<quote who="Niall Litchfield">
> A couple of people have pointed out that my old site does not render
> correctly in all browsers. I've therefore dug out the text below.

I wonder why blogger.com creates templates that won't render in Firefox? It's like going back to the Web stone age... But I digress...

Thanks Niall for pointing me to your site. Although I didn't make use of your system trigger solution, it gave me a hint for further research: I did not look for processes anymore but instead tried to find out which objects blocked the temp tablespace instead. It was a lot of LOB segments, and we run a configuration with connection pools (persisting sessions) that create one LOB after another.
Log story short: These LOB segments should have been freed with DBMS_LOB.freetemporary. After developers used this procedure, the problem was gone.

This, combined with your article, would explain why there was no trace file generated even though the errors appeared in the alert log. But it still stuns me that this issue is so difficult to diagnose.

Best regards and have a nice weekend,
Uwe

> On 5/3/07, Niall Litchfield <niall.litchfield_at_gmail.com> wrote:
>
>> Does the approach at
>>
> http://www.niall.litchfield.dial.pipex.com/2004/06/we-have-one-of-those-i
> rritating-issues
>> help? Note larry wolfson of this list pointed out that in 9i don't try
>> to monitor errors not caused by sql statements with this approach.
>>
>> cheers
>>
>> Niall
>>
>>
>> On 5/2/07, Uwe Küchler <uwe_at_kuechler.org> wrote:
>>
>>> Oracle 9.2.0.6, HP-UX 11.11
>>>
>>>
>>> Hello world,
>>> today I was trying to track down the SQL causing a full TEMP
>>> tablespace
> by
>>> tracing the 1652 event. I tried the following commands:
>>>
>>> - ALTER SYSTEM SET EVENTS '1652 trace name errorstack';
>>> - ALTER SYSTEM SET EVENTS = '1652 trace name errorstack';
>>> - ALTER SYSTEM SET EVENTS = '1652 trace name context forever, level
>>> 12';
>>> - oradebug event 1652 trace name errorstack

--
http://www.freelists.org/webpage/oracle-l
Received on Fri May 04 2007 - 10:56:51 CDT

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US