Re: stale partition statistics
Date: Tue, 9 Mar 2021 21:32:23 +0200
Message-ID: <CA+riqSUD8t4bR_q5iW1OfmAAX_5rEbM1HSy=hPzBS52GXOk6Sg_at_mail.gmail.com>
You can disable the query used to test the quality of histograms using: alter system set "_fix_control"='16726844:0';
like this you can isolate if histograms are the culprit for your situation .
În mar., 9 mar. 2021 la 20:58, Jonathan Lewis <jlewisoracle_at_gmail.com> a scris:
>
> I think flags = 8 indicates missing or inconsistent histograms.
>
> Regards
> Jonathan Lewis
>
>
> On Mon, 8 Mar 2021 at 19:26, Noveljic Nenad <nenad.noveljic_at_vontobel.com>
> wrote:
>
>> Hi Mikhail,
>>
>>
>>
>> Yes, non-zero flags definitely trigger gathering of stale statistics
>> regardless of the number of modifications. For flags = 1 that’s a correct
>> behavior, since it’s a truncate. For flags=8 we still don’t know. I’m going
>> to chase the application people tomorrow to rerun the load jobs on the
>> restored database. I hope I’ll catch the culprit in the trace.
>>
>>
>>
>> Best regards,
>>
>>
>>
>> Nenad
>>
>>
>>
-- http://www.freelists.org/webpage/oracle-lReceived on Tue Mar 09 2021 - 20:32:23 CET