Re: kkslce [KKSCHLPIN2] mutex contention on hard parse of PQ with binds
From: Timur Akhmadeev <timur.akhmadeev_at_gmail.com>
Date: Wed, 16 Sep 2015 10:25:45 +0300
Message-ID: <CACGsLCKRG-Lk2+VSZn_h4kL1OG6qTLwYyh8ZAJq1h4Jccbuu=Q_at_mail.gmail.com>
On Tue, Sep 15, 2015 at 5:58 PM, Stefan Knecht <knecht.stefan_at_gmail.com> wrote:
Date: Wed, 16 Sep 2015 10:25:45 +0300
Message-ID: <CACGsLCKRG-Lk2+VSZn_h4kL1OG6qTLwYyh8ZAJq1h4Jccbuu=Q_at_mail.gmail.com>
On Tue, Sep 15, 2015 at 5:58 PM, Stefan Knecht <knecht.stefan_at_gmail.com> wrote:
> We've done quite a bit of digging in this area, and one thing I noticed
> was that the default for _cursor_obsolete_threshold was increased from 100
> (11.2) to 1024 (12.1.0.2). _cursor_obsolete_threshold controls the number
> of child cursors are kept for a given parent cursor before it's marked as
> obsolete and aged out. To accommodate that new limit, they have
> likely changed some things internally.
>
_cursor_obsolete_threshold=1024 since 11.2.0.4 and I doubt it alone has any influence here - only if something else has changed.
-- Regards Timur Akhmadeev -- http://www.freelists.org/webpage/oracle-lReceived on Wed Sep 16 2015 - 09:25:45 CEST