RE: RE: SGA Fragmentation - How to Monitor

  • From: "Mudhalvan Moovarkku" <moovarkku.mudhalvan@xxxxxxxxxxxxxxxx>
  • To: "Niall Litchfield" <niall.litchfield@xxxxxxxxx>, <Joel.Patterson@xxxxxxxxxxx>, <Oracle-L@xxxxxxxxxxxxx>
  • Date: Thu, 11 Mar 2010 10:31:01 +0900

Dear All,

            Thanks for everyone reply. I have checked the meta link
document and feel I can do some scripting which will help me to find out
the error before it come.

 

            By the way when I do analysis just few seconds before this
error ORA-04031 happened one trace file was created at udump folder
which has the following message and found the chunk is suddenly down
from 6000 to 15.

 

---- Snip Start -  udump trace file

            FREELIST CHUNK COUNT:6328  OBJECT SIZE:184

LATCH:10  TOTAL SPACE: 15447168

FREELIST CHUNK COUNT:5072  OBJECT SIZE:184

Permanent space allocated for KGL S handles

LATCH:0  TOTAL SPACE: 348480

FREELIST CHUNK COUNT:14  OBJECT SIZE:480

LATCH:1  TOTAL SPACE: 341760

FREELIST CHUNK COUNT:5  OBJECT SIZE:480

LATCH:2  TOTAL SPACE: 345600

---- Snip End - udump trace file

 

            Do you have any clue how to find why it is happened?

 

Regards

Mudhalvan M.M

________________________________

From: Niall Litchfield [mailto:niall.litchfield@xxxxxxxxx] 
Sent: Thursday, March 11, 2010 3:56 AM
To: Joel.Patterson@xxxxxxxxxxx
Cc: Jon.Crisler@xxxxxxx; karlarao@xxxxxxxxx; Mudhalvan Moovarkku;
Oracle-L@xxxxxxxxxxxxx
Subject: Re: RE: SGA Fragmentation - How to Monitor

 

Joel, this is about shared pool fragmentation not keeping blocks in the
SGA. Any PL/SQL intensive app is likely to benefit from this, assuming
they've packaged routines appropriately

        On Mar 10, 2010 6:47 PM, <Joel.Patterson@xxxxxxxxxxx> wrote:
        
        I really can't answer definitively one way or another.  If it
was good
        for you under 8 and 9, then of course you know for versions 8
and 9.
        
        The thinking has always been in the past (oracle 8 and 9), when
the
        subject came up, especially with tables, that the least recently
used
        algorithm employed by oracle would purge essentially keep the
hot tables
        anyway and so there wasn't any point in pinning them.
        
        Now I'm sure you have more detail than that, but thought I'd put
that
        out there for consideration for whatever it might be worth.
        
        Joel Patterson
        Database Administrator
        904 727-2546

        -----Original Message----- From: oracle-l-bounce@xxxxxxxxxxxxx 

        [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On Behalf Of Crisler, Jon
Sent: Wednesday, March 10, 2010 11:...

        Subject: RE: SGA Fragmentation - How to Monitor If this is an
Oracle Applications / Ebiz / ERP syst...

Other related posts: