Re: Index question

From: joel garry <joel-garry_at_home.com>
Date: Tue, 25 Aug 2009 13:48:43 -0700 (PDT)
Message-ID: <fd0674fb-eadd-40d7-9d6f-3c19ea2685dd_at_i8g2000pro.googlegroups.com>



On Aug 25, 1:01 pm, "stevedhow..._at_gmail.com" <stevedhow..._at_gmail.com> wrote:
> On Aug 25, 1:22 pm, joel garry <joel-ga..._at_home.com> wrote:
>
>
>
> > Wild speculation:  Have you tried disabling parallel query for the
> > objects?
>
> > jg
> > --
>
> Hi Joel,
>
> I didn't for the objects, but did in my session with no difference.
>
> SQL> select min(jrn_seq) from xwc.xwc_change_history where jrn_seq
> between 1019514060 and 1019514080;
>
> MIN(JRN_SEQ)
> ------------
>   1019514075
>
> SQL> alter session disable parallel query;
>
> Session altered.
>
> SQL> select min(jrn_seq) from xwc.xwc_change_history where jrn_seq
> between 1019514060 and 1019514080;
>
> MIN(JRN_SEQ)
> ------------
>   1019514075
>
> SQL>
>
> As Michel and Jonathan noted, it is probably corruption.
>
> We need to execute a rebuild for this index anyway (we just deleted
> 250 million "increasing sequence type" rows which is making MIN
> queries such as this horrific in terms of performance).
>
> Thanks all,
>
> Steve

They're probably right. See also bug 7132684 if the rebuild doesn't work (though that seems to be the converse problem, incorrectly costing local index lower, hey, if that isn't right, other stuff related to it might not be right - which doesn't conflict with what Jonathan noted, which sounds like wrong results to me).

jg

--
_at_home.com is bogus.
http://www3.signonsandiego.com/stories/2009/aug/25/wikipedia-limit-editing-articles/?uniontrib
Received on Tue Aug 25 2009 - 15:48:43 CDT

Original text of this message