Good point, Paul. See this item over at AskTom for full details: http://asktom.oracle.com/pls/ask/f?p=3D4950:8:5150260736162217859::NO::F4= 950_P8_DISPLAYID,F4950_P8_CRITERIA:17989406187750, Original poster: What version of Oracle are you using??? -----Original Message----- From: Paul Drake [mailto:bdbafh@xxxxxxxxx] Sent: Thursday, October 14, 2004 2:27 PM To: Bobak, Mark Cc: ganesh.raja@xxxxxxxxx; rlsmith@xxxxxxx; shaharul.anuar@xxxxxxxxxxxxxxxx; oracle-l@xxxxxxxxxxxxx Subject: Re: Help with Shared Pool Problem IIRC, in 9.2.0.5 - the behavior of open_cursors changes. I remember while attending a presentation at NYOUG (http://www.nyoug.org) given by Steven Feuerstein, that if your code is keeping cursors open that are not using bind variables, that you are in serious trouble. Wait - it might have been at Jonathan Lewis' seminar. I'll have to get back to you on that. Check as far as the number of session_cached_cursors and open_cursors. Perhaps Jonathan might have something to add. Paul On Thu, 14 Oct 2004 11:55:52 -0400, Bobak, Mark <mark.bobak@xxxxxxxxxxxxxxx> wrote: > Ganesh, >=20 > While ORA-4031 can mean that there is no space in the shared > pool, it can also mean there is not enough contiguous memory > in the shared pool. So, if you need 1k for a particular SQL > statement, and the largest available chunk is 900k, then=3D20 > Oracle will signal an ORA-4031. In this case, sometimes, > flushing the shared pool can help, but not always. =3D20 >=20 > Yes, there is an LRU mechanism for certain components in > the shared pool. See the scripts I previously mentioned > on Steve Adams' website, IxOra. >=20 > -Mark > -- http://www.freelists.org/webpage/oracle-l