Re: performance issue after upgrade to oracle linux 32 bit.

From: joel garry <>
Date: Tue, 3 Nov 2009 09:44:25 -0800 (PST)
Message-ID: <>

On Nov 3, 8:22 am, Charles Hooper <> wrote:
> On Nov 3, 10:29 am, lsllcm <> wrote:

> Note also that it is possible that the object statistics and optimizer
> parameters (FIRST_ROWS_n, for example) differ between the Oracle
> versions.
> David's suggestion of a 10053 trace is a good one - just keep in mind
> that such traces only appear during a hard parse.

A bit of a stretch, but consider this: rownum is assigned after the predicate is processed, but before sorting is done. "Bug 6438892 : Suboptimal plan with ROWNUM predicate and NESTED LOOPS." is in the patch set, bug not published. So maybe there is some mysterious optimizer bug here, set off by the order by even though rownum is not originally a predicate. Try patching beyond the base release and see if the problem is still there.


-- is bogus.
Do not fart next to a cellphone.
Received on Tue Nov 03 2009 - 11:44:25 CST

Original text of this message