Re: Trigger "AFTER LOGON ON SCHEMA" never works.

'Make sure you don't touch undocumented parameters...' is a valid point as a generalization. but generally speaking, generalizations don't always apply ;-)

as i noted, 'see MetaLink Note:68636.1'

as you note, _SYSTEM_TRIG_ENABLED defaults to TRUE, and Oracle recommends setting it to FALSE only during database upgrades. what if the DBA forgets to remove or reset the parameter (which we just had to deal with)?

since any changes to the INIT.ORA should always include an explanatory comment (wouldn't that be nice), likely the DBA noted the reason for including this or any other '_HIDDEN_PARAMETER', so removing the setting can be done with full knowledge of why it was set in the first place.

in light of your observation, if you find any other hidden parameters in the INIT.ORA without comments documenting when and why they were set, it would be a very, very good idea to investigate and document the justification for the parameter's use (or comment it out and document the time and reason the undocumented change was reversed)

bottom line -- it is by no means a 'vain hope' but rather sound debugging to check for DBA mistakes, like a recommended temporary setting was inadvertently left in the mix

anyway, thanks for raising the generalized caution flag

• mcs

"Mladen Gogala" <mgogala_at_adelphia.net> wrote in message news:pan.2003.11.02.07.40.32.881729_at_adelphia.net...
> On Tue, 28 Oct 2003 09:57:28 -0500, mcstock wrote:
>
> > mark sure the hidden parameter _SYSTEM_TRIG_ENABLED is not set to FALSE
(see
> > MetaLink Note:68636.1 )
>
> Make sure you don't touch undocumented parameters unless oracle told you
> so. I hate playing with underscores in vain hope that something might get
> better if you add a wing of a bat, a feather of a black rooster, snake
> tooth and a hidden parameter. Stir counterclockwise, on Friday at
> midnight and your performance will improve dramatically. That parameter is
> set to true by default and you don't need to touch it. When you're
> recommending playing with the undocumented parameters so lightheartedly,
> I'm sure you know how to check it? I don't need to describe x$ksspi and
> x$ksppcv?
>
> Other then that, this parameter only needs to be set during upgrade
> operations, so playing with that parameter may only mess up the database.
> Trigger is syntactically incorrect and contains "commit" command.
> Parameters in general and especially the undocumented ones are the last
> place where one should look when in doubt. The first one is Tom Kyte's
> book (Expert 1 on 1).
>
> --
> None of us is as dumb as all of us.
> (http://www.despair.com/meetings.html)
>
Received on Sun Nov 02 2003 - 07:22:49 CST