This is a follow-up to say that I applied patch #5686711 six days ago and the problem hasn't appeared since. I should add that Jeremiah pointed me in the direction of this patch along with another similar one which we had already applied in all of our environments. Bug #5686711 was unpublished, however, so I had to rely on Oracle Support to provide more information about the bug so that I could determine if what we were seeing matched the symptoms of the bug. -Mark On Wed, May 14, 2008 at 9:41 AM, Mark Strickland <strickland.mark@xxxxxxxxx> wrote: > Oracle 10.2.0.2 on Redhat 2.6.9-55.0.0.0.2 > oracle-jdbc-10.2.0.2.0 > jdk1.5.0_11 > Weblogic uses jrockit-R27.4.0-jdk1.5.0_12 > > Suddenly late last week, our developers started getting the above error on > SELECT statements. Yes, all proper verifications have been done - database > connection string is correct, the correct usernames and passwords are being > used and the queries can still be executed in SQL*Plus and Toad. The > developers each have their own objects. A database bounce makes this go > away. For a few days. Then it starts happening again. There have been no > changes on the database side. I've searched Google, Metalink, and Oracle-L > and I've opened an SR. I can see evidence that other customers have > experienced this but I haven't found where any of them resolved it. > > There is anecdotal evidence that this has happened here before in the past > before my time. The word "version" as in Flashback Query has popped up in > conversations. It is suspected that an incorrectly parsed query ends up in > the shared pool for whatever reason and, from that point on, any further > executions of that query result in the error. I tried flushing the shared > pool, but then a developer started getting permissions errors, so I had to > go ahead and bounce the database. > > Has anyone had this problem and, if so, was there a resolution? > > Regards, > Mark Strickland > Seattle >