Re: dbms_stats.gather_table_stats only calculating stats for first partition of range interval - list composite partitioned table

Regards

Pete

On Fri, Dec 11, 2009 at 1:07 PM, Peter Hitchman <pjhoraclel_at_gmail.com>wrote:

> Hi Greg,
> Thanks for the reply. I had got to the point where I hoped this was a bug,
> but i was interested to see if I could find anyone else it had been tripped
> up.
> For the time being I using the dbms_stats.set_ calls to create the stats
> that I can by summing from the subpartitions, this has had the desired
> impact on the SQL plans.
>
> I take your point about the auto sample size, the reason I had not been
> using it was that we started with 11.1.0.6 very soon after it was released
> and I was sticking to what I knew worked for this database. But given the
> nice research you have done, I have decided that in future I will use the
> dbms_stats defaults.
>
> At some point I will have to open a SR with Oracle, to get the official
> line on this and then I suspect we will planning to upgrade to 11.2, because
> that release has other bug fixes we need.
>
> Regards
>
> Pete
>
>
>
> On Thu, Dec 10, 2009 at 6:57 PM, Greg Rahn <greg_at_structureddata.org>wrote:
>
>> I'm unable to find the exact bug (work with Oracle Support if you
>> really want it) but my suggestion would be this:
>> Upgrade to 11.1.0.7 and then apply patch 8318020 for dbms_stats.
>> There is a significant number of fixes in 11.1.0.7 and it has been a
>> pretty stable release IMHO and it fixes quite a number of bugs found
>> in 11.1.0.6.
>>
>>
>> On Thu, Dec 10, 2009 at 10:17 AM, Greg Rahn <greg_at_structureddata.org>
>> wrote:
>> > What you are observing is a bug in 11.1.0.6 - it does not reproduce in
>> > 11.1.0.7. Let me see if I can find the bug.
>>
>> --
>> Regards,
>> Greg Rahn
>> http://structureddata.org
>>
>
>
>
> --
> Regards
>
> Pete
>

-- 
Regards

Pete

--
http://www.freelists.org/webpage/oracle-l