Re: ora-04031

Hi George...

  You mention that the shared pool is "huge" and "most of it is unused"... Do you have any measurements of this? As Chris mentioned, what's the size of your largest free chunk? It doesn't matter if you have 2 GB of free space if the largest free chunk is only 1M and you need 1.5 M then you're in trouble... The classic swiss cheese memory issue... Here's a simple example...

Here is your free memory... X means it's free and Y means used... You start your instance and it's all free space... My imaginary example contains 36 chunks...

XXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXXX

Now... You start performing work... Memory chunks are allocated and deallocated as time goes by... You end up with a memory area that looks something like...

XXXAXXXAXXAXXXAXXXAXXXAXXAXAXXXAXXAX Now... I look at total space and see that 26 out of the 36 chunks are free... That means you're over 70% free space... But, can you allocated a contiguous chunk of XXXX? No...

So, you have to start looking at how big your chunks are and why you're so fragmented... Are you using bind variables? I'm guessing no between the fragmentation problem and the "huge" SGA statement... The one thing that throws me for a loop is the TOAD vs OAS thing... I'm guessing the experiment is impure... How did you perform it? Does the same thing happen in SQL*Plus? Do you have the same level of activity on the database? Is it repeatable? etc...

Tim

Michael Milligan wrote:

>The error means an object is trying to load in shared memory and can not
>find contiguous space to do so.
>
>Set your sga_max_size higher, so it allows you to dynamically alter sga size
>with the DB running. Also, set your values for shared_memory and
>shared_memory_reserved_space higher. The shared_memory_reserved_space should
>be 10% of your shared_memory space. This reserves space for large
>objects(not talking about LOBs here, just large packages, etc.) that would
>not ordinarily fit.
>
>HTH
>
>Mike
>
>-----Original Message-----
>From: Feighery Raymond [mailto:Raymond.Feighery_at_churchill.com]
>Sent: Thursday, March 11, 2004 4:53 AM
>To: 'oracle-l_at_freelists.org'
>Subject: RE: ora-04031
>
>
>I had a similar problem with 9iAS, MTS and the Large Pool. As Apache was
>indefinitely reusing each HTTP child server process for multiple requests
>the memory the child process required in the Large Pool constantly increased
>until ORA-4031.
>
>The solution was to set MaxRequestsPerChild to > 0 (e.g. 50). It defaults to
>0 (unlimited). Setting it to non-zero may incur a small performance hit.
>
>Here's a script I copied from somewhere (sorry unknown author) which may
>help diagnose your problem.
>
>It may also help to set an event e.g. event = "4031 trace name errorstack
>level 4" to generate a trace when the error occurs.
>
>By the way, what version of iAS, RDBMS and OS are you on?
>
>Ray
>
>
>prompt ***********************************
>prompt ***********************************
>prompt Diagnosing ORA-4031
>prompt ***********************************
>prompt ***********************************
>prompt THE SGA
>select * from v$sgastat;
>PROMPT SHARED_POOL_RESERVED_MIN_ALLOC
>select val.ksppstvl VALUE
>from x$ksppi nam,
>x$ksppsv val
>where nam.indx = val.indx
>and nam.ksppinm = '_shared_pool_reserved_min_alloc'
>order by 1
>/
>SELECT free_space, avg_free_size, used_space,
>avg_used_size, request_failures, last_failure_size
>FROM v$shared_pool_reserved;
>prompt The ORA-04031 is a result of lack of contiguous space in the shared
>pool
>prompt reserved space if:
>prompt REQUEST_FAILURES is > 0 and LAST_FAILURE_SIZE is >
>prompt SHARED_POOL_RESERVED_MIN_ALLOC.
>prompt To resolve this consider increasing SHARED_POOL_RESERVED_MIN_ALLOC to
>lower
>prompt the number of objects being cached into the shared pool reserved
>space and
>prompt increase SHARED_POOL_RESERVED_SIZE and SHARED_POOL_SIZE to increase
>the
>prompt available memory in the shared pool reserved space.
>prompt The ORA-04031 is a result of lack of contiguous space in the library
>cache if:
>prompt REQUEST_FAILURES is > 0 and LAST_FAILURE_SIZE is <
>prompt SHARED_POOL_RESERVED_MIN_ALLOC
>prompt or
>prompt REQUEST_FAILURES is 0 and LAST_FAILURE_SIZE is <
>SHARED_POOL_RESERVED_MIN_ALLOC
>
>
>-----Original Message-----
>From: George Leonard [mailto:leonarge_at_vodacom.co.za]
>Sent: Thursday, March 11, 2004 10:23 AM
>To: oracle-l_at_freelists.org
>Subject: RE: ora-04031
>
>
>Hi there
>
>It seems this did not help. (problem was not on the application server
>infrastructure database)
>
>And as for the production database the shared pool is huge and most of it is
>unused and free.
>
>I am looking at scripts to see if I can figure out what the problem is.
>
>
>
>George
>
>________________________________________________
>
>George Leonard
>Oracle Database Administrator
>Dimension Data (Pty) Ltd
>(Reg. No. 1987/006597/07)
>Tel: (+27 11) 575 0573
>Fax: (+27 11) 576 0573
>E-mail:george.leonard_at_za.didata.com
>Web: http://www.didata.co.za
>
>You Have The Obligation to Inform One Honestly of the risk, And As a Person
>You Are Committed to Educate Yourself to the Total Risk In Any Activity!
>Once Informed & Totally Aware of the Risk,
>Every Fool Has the Right to Kill or Injure Themselves as They See Fit!
>
>
>-----Original Message-----
>From: oracle-l-bounce_at_freelists.org [mailto:oracle-l-bounce_at_freelists.org]
>On Behalf Of George Leonard
>Sent: 11 March 2004 10:54 AM
>To: oracle-l_at_freelists.org
>Subject: RE: ora-04031
>
>Hi there
>
>We found something interesting, it normally happens on a specific set of
>queries, big ones. What we found if execute via the front end (Oracle
>Application server involved) we get this error, if executed via toad
>directly on the back end we have no problems.
>
>What we are thinking is that the error is actually produced by the
>application servers infrastructure database. We are altering the shared
>pool and buffer cache parameters for this and restarting the environment
>to see if this resolves our problems.
>
>George
>
>________________________________________________
>
>George Leonard
>Oracle Database Administrator
>Dimension Data (Pty) Ltd
>(Reg. No. 1987/006597/07)
>Tel: (+27 11) 575 0573
>Fax: (+27 11) 576 0573
>E-mail:george.leonard_at_za.didata.com
>Web: http://www.didata.co.za
>
>You Have The Obligation to Inform One Honestly of the risk, And As a
>Person
>You Are Committed to Educate Yourself to the Total Risk In Any Activity!
>Once Informed & Totally Aware of the Risk,
>Every Fool Has the Right to Kill or Injure Themselves as They See Fit!
>
>
>
>-----Original Message-----
>From: oracle-l-bounce_at_freelists.org
>[mailto:oracle-l-bounce_at_freelists.org] On Behalf Of
>chris_at_thedunscombes.f2s.com
>Sent: 11 March 2004 10:47 AM
>To: oracle-l_at_freelists.org
>Subject: Re: ora-04031
>
>George,
>
>Does this always happen even after a shutdown and restart of the
>database?
>It might help to look at the size of the free chunks in the SGA. Steve
>Adam's
>ixora site (www.ixora.com.au) has some excellent scripts for getting the
>size
>of each of the free chunks.
>
>HTH
>
>Chris
>
>
>Quoting George Leonard :
>
>
>
>>Hi all, weird thing, the developers is executing pretty big queries
>>
>>
>via
>
>
>>java. Sometimes on some of the queries they get this error, if they
>>
>>
>take
>
>
>>the SQL and execute it via sqlplus immediate it works no errors.
>>
>>java.sql.SQLException: ORA-04031: unable to allocate 1553024 bytes of
>>shared memory ("shared pool","select (isoyear * 100) +
>>iso...","Typecheck heap","qry_text : qcpisqt")
>>
>>George
>>
>>________________________________________________
>>
>>George Leonard
>>Oracle Database Administrator
>>Dimension Data (Pty) Ltd
>>(Reg. No. 1987/006597/07)
>>Tel: (+27 11) 575 0573
>>Fax: (+27 11) 576 0573
>>E-mail:george.leonard_at_za.didata.com
>>Web: http://www.didata.co.za
>>
>>You Have The Obligation to Inform One Honestly of the risk, And As a
>>Person
>>You Are Committed to Educate Yourself to the Total Risk In Any
>>
>>
>Activity!
>
>
>>Once Informed & Totally Aware of the Risk,
>>Every Fool Has the Right to Kill or Injure Themselves as They See Fit!
>>
>>
>
>
>
>>
>>
>>"This e-mail is sent on the Terms and Conditions that can be accessed
>>
>>
>by
>
>
>>Clicking on this link http://www.vodacom.net/legal/email.asp "
>>
>>
>>
>
>
>Chris Dunscombe
>
>chris_at_thedunscombes.f2s.com
>
>-------------------------------------------------
>Everyone should have http://www.freedom2surf.net/
>----------------------------------------------------------------
>Please see the official ORACLE-L FAQ: http://www.orafaq.com
>----------------------------------------------------------------
>To unsubscribe send email to: oracle-l-request_at_freelists.org
>put 'unsubscribe' in the subject line.
>--
>Archives are at http://www.freelists.org/archives/oracle-l/
>FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
>----------------------------------------------------------------- "This
>e-mail is sent on the Terms and Conditions that can be accessed by Clicking
>on this link http://www.vodacom.net/legal/email.asp "
>"This e-mail is sent on the Terms and Conditions that can be accessed by
>Clicking on this link http://www.vodacom.net/legal/email.asp "
>
>___________________________________________________________________________
>
>
>This email and any attached to it are confidential and intended only for the
>individual or entity to which it is addressed. If you are not the intended
>recipient, please let us know by telephoning or emailing the sender. You
>should also delete the email and any attachment from your systems and should
>not copy the email or any attachment or disclose their content to any other
>person or entity. The views expressed here are not necessarily those of
>Churchill Insurance Group plc or its affiliates or subsidiaries. Thank you.
>
>Churchill Insurance Group plc. Company Registration Number - 2280426.
>England.
>
>Registered Office: Churchill Court, Westmoreland Road, Bromley, Kent BR1
>1DP.
>
>
>----------------------------------------------------------------
>Please see the official ORACLE-L FAQ: http://www.orafaq.com
>----------------------------------------------------------------
>To unsubscribe send email to: oracle-l-request_at_freelists.org
>put 'unsubscribe' in the subject line.
>--
>Archives are at http://www.freelists.org/archives/oracle-l/
>FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
>-----------------------------------------------------------------
>
>
>This e-mail, including attachments, may include confidential and/or
>proprietary information, and may be used only by the person or entity to
>which it is addressed. If the reader of this e-mail is not the intended
>recipient or his or her authorized agent, the reader is hereby notified that
>any dissemination, distribution or copying of this e-mail is prohibited. If
>you have received this e-mail in error, please notify the sender by replying
>to this message and delete this e-mail immediately.
>----------------------------------------------------------------
>Please see the official ORACLE-L FAQ: http://www.orafaq.com
>----------------------------------------------------------------
>To unsubscribe send email to: oracle-l-request_at_freelists.org
>put 'unsubscribe' in the subject line.
>--
>Archives are at http://www.freelists.org/archives/oracle-l/
>FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
>-----------------------------------------------------------------
>
>

-- 
Regards,
Tim Johnston
Tel: 978-322-4226
Fax: 978-322-4100


----------------------------------------------------------------
Please see the official ORACLE-L FAQ: http://www.orafaq.com
----------------------------------------------------------------
To unsubscribe send email to:  oracle-l-request_at_freelists.org
put 'unsubscribe' in the subject line.
--
Archives are at http://www.freelists.org/archives/oracle-l/
FAQ is at http://www.freelists.org/help/fom-serve/cache/1.html
-----------------------------------------------------------------