Re: 10053 Trace File anomolies

<mccmx_at_hotmail.com> wrote in message
news:1119870544.899564.139430_at_g49g2000cwa.googlegroups.com...
> OK that makes sense, but the confusing part is that these 2 databases
> were exact clones of each other 2 days ago so they should be working
> off the same datasets.
>
> The strange part is that the first join order that the optimizer checks
> in each of the 10053 traces is totally different.
>
> Shouldn't they at least start with the same join order (as stated below
> - from Wolfgang Brietling's doc: A LOOK UNDER THE HOOD OF CBO: THE
> 10053 EVENT)
>
> "The initial join order is chosen by ordering the tables in order of
> increasing computed cardinality."
>
> Matt
>

The data sets, or their statistics, are presumably no longer exact clones of each other. As a first quick check, look for the SINGLE TABLE ACCESS PATH for each table, and check for lines like:

  TABLE: CHILD ORIG CDN: 10192 ROUNDED CDN: 83 CMPTD CDN: 83 Then, if necessary, do a cross check between the two sets of Table stats just after the heading: BASE STATISTICAL INFORMATION.

-- 
Regards

Jonathan Lewis

http://www.jlcomp.demon.co.uk/faq/ind_faq.html
The Co-operative Oracle Users' FAQ

http://www.jlcomp.demon.co.uk/appearances.html
Public Appearances - schedule updated June 22nd 2005