Re: How to get rid of invalidated V$sql entries

Alfonso,

    When the same text (ie the same hash value) appears different times but with different execution plans (either because the same table names correspond to different tables - same name, different schema, or private synonyms aliasing different objects - or because session-level parameters such as sort_area_size or optimizer_goal are different and lead to a different execution plan) there is one 'parent' cursor with 0 executions and several child cursors with different child-num but same sql_address/hash_value. You can't get rid of parents .... Just ignore whatever seems not to be executed.

HTH Stephane Faroult

Alfonso León wrote:

>Hello:
>
>I have some questions:
>
>1. We have a java tool that makes queries to the Databas (9.2.0.6).
>When I check in the v$sql view at 5pm I can see there are entries
>since 8:30 in the morning, with 0 executions and other fields, I guess
>they=B4re invalidated or just old, but when would the dissapear? Ther is
>a way to get rid of that entries.
>
>2. Oracle says that V$sqlarea is an aggregate of V$sql but when I do a
>sum of the sharable_mem of the active entries of a hash_value the
>result is different of the sharable_mem of the V$sqlarea.
>
>I will appreciate your ideas.
>
>--=20
>Alfonso Leon
>--
>http://www.freelists.org/webpage/oracle-l
>
>
>

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