Re: Chained vs. migrated rows - Any easy way to tell the
From: Martin Berger <martin.a.berger_at_gmail.com>
Date: Tue, 18 Nov 2008 13:44:15 +0100
Message-Id: <4877DE84-695B-4F15-8FE9-22E4A12CD5B5@gmail.com>
Date: Tue, 18 Nov 2008 13:44:15 +0100
Message-Id: <4877DE84-695B-4F15-8FE9-22E4A12CD5B5@gmail.com>
would it be save to run
DBMS_SPACE.OBJECT_SPACE_USAGE to check for chain_pcent > 0 without
disturbing DBMS_STATS at all?
just a guess - never tried yet.
And I also doesn't see a valid answer for the question between chained
and migrated rows there.
br.
Martin
-- Martin Berger http://berxblog.blogspot.com Am 06.11.2008 um 10:35 schrieb Christian Antognini:Received on Tue Nov 18 2008 - 06:44:15 CST
> Hi Joel
>
>> The difference between dba_tables.chain_cnt and what is contained
>> in chained_rows table now is still a mystery. Stats are collected
>> nightly.
>
> DBMS_STATS doesn't compute the value chain_cnt. Therefore, there are
> two
> possibilities:
> 1) dba_tables.chain_cnt = 0
> 2) dba_tables.chain_cnt > 0
>
> If 1, either the table doesn't contain chained/migrated rows or
> ANALYZE
> was never used to compute statistics for it.
>
> If 2, at some time in the past the statistics were compute with
> ANALYZE.
> Since DBMS_STATS doesn't compute it, you see the value at the time of
> the last execution of ANALYZE.
-- http://www.freelists.org/webpage/oracle-l
- application/pkcs7-signature attachment: smime.p7s