Igor Racic schrieb:
>
> It seems he doesn't have performance problem but bug in application.
Yes, thats the problem. Some data "meet" on the server in a strange way,
then several funny effects are happen. Some apps run, some crash, but
nothing to easily identify whats wrong. And everytime new combination of
crashing/running apps, almost never same combination twice. Looks like
it depends on the data and/or on the statements executed at that time,
thats why I was asking for chances to enable logging and so on. The bug
seems to be a combination of a performance problem (too slow - or not
finished when trying to do something) and an application timing problem,
difficult to analyze without database on heavy load. My own apps I may
control, but with oracle - sometimes more black-box than I like ;-)
> So, level 4 of event 10046 could be enough.
Good hint - I'll check, and the level 12 suggestion at Sybrand's posting.
> Also, I suppose, having several sessions writing in user_dump_dest
> impact may be much more then neglicible.
Thats the reason for my question - I know it has a measurable effect on
the database performance, when I enable logging. But I don't know how
big - and to prevent from any effects, I was asking for a logging on
"outside" the machine.
>
> If he is on 10g, then FGA may be of help.
Sadly - no. Its 9i.
>
> Just my 2c
>
> Igor
Thanks - Bjoern
>
Received on Sun Jul 11 2004 - 10:41:41 CDT
