Re: ora-3113 related to histograms? 11gr2
From: fairlie rego <fairlie_r_at_yahoo.com>
Date: Sat, 2 Oct 2010 19:46:55 -0700 (PDT)
Message-ID: <548759.34415.qm_at_web45011.mail.sp1.yahoo.com>
Date: Sat, 2 Oct 2010 19:46:55 -0700 (PDT)
Message-ID: <548759.34415.qm_at_web45011.mail.sp1.yahoo.com>
We've been running Peoplesoft on 11.2 on the Oracle Database machine v2 for the past 10 months... Bind variable peeking is turned off so _OPTIM_PEEK_USER_BINDS is set to FALSE A 3113 is a client side error. So have you checked if you are seeing any 7445 errors in the alert log Fairlie Rego Senior Oracle Consultant http://el-caro.blogspot.com/ M: +61 402 792 405 ________________________________ From: David Roberts <big.dave.roberts_at_googlemail.com> To: andrew.kerber_at_gmail.com Cc: ORACLE-L <oracle-l_at_freelists.org> Sent: Sat, 2 October, 2010 11:27:28 PM Subject: Re: ora-3113 related to histograms? 11gr2 My expectation would be, lack of histogram = different execution plan. Thus it would be my expectation that one of the execution plans (The one chosen in the absence of histograms) is either hitting an unrecognised corrupt index or table. It's a guess, but I'd try to identify the execution plans with and without the histogram statistics and work from there! Dave On Wed, Sep 29, 2010 at 7:51 PM, Andrew Kerber <andrew.kerber_at_gmail.com> wrote: This one is really strange. We are running peoplesoft on Oracle 11Gr2 on Linux x86_64. Periodically we get ORA-3113 errors (network disconnect) from one of our queries running from the app server to the database. It appears to happen when histograms are not present on the table. When they are present, the problem rarely, if ever occurs. Can anyone give us idea why this may be happening? MOS noticed the correlation, but no one has figured out the cause, or a true fix. I am planning on setting _OPTIM_PEEK_USER_BINDS=FALSE and see what difference that makes, but best case, that would only keep the problem from happening, not show us a cause. We also get a ORA-3137 TTC protocol internal error, but that appears to be on the app server side caused by the ORA-3113. > >-- >Andrew W. Kerber > >'If at first you dont succeed, dont take up skydiving.' >
-- http://www.freelists.org/webpage/oracle-lReceived on Sat Oct 02 2010 - 21:46:55 CDT