Niall Litchfield wrote:
> yong321_at_yahoo.com wrote:
> > Thanks, Srivenu. That's a perfect answer for Dexter's question,
> > assuming he's not using 10g which would have dba_enabled_traces (as
> > Niall says). Expanding this thread to check whether a non-10046
event
> > is set in another session, I posted a message before using event
> > errorstack as an example:
> >
>
http://groups-beta.google.com/group/comp.databases.oracle.server/msg/b44fcbcef5a55323
> > The number 604 is arbitrarily chosen. The difference between
> SQL-trace
> > (10046) and others such as errorstack is that in the latter case,
the
> > trace file is not immediately created. Therefore oradebug
> > tracefile_name still does not show a filename in udump, but
oradebug
> > dump events 1 creates a trace file.
>
> It may not be so perfect if Dexter is running on Windows. I have just
> checked (because it is a long time since I ran oradebug) and it
appears
> that the trace doesn't get created on windows until the relevant
> session ends.
I often do oradebug on Windows. There're two anomalies compared to
UNIX/Linux. Always use setorapid (v$process.pid), because setospid may
not work. Always do something on the debugged session, or the debugging
session hangs. This "something" can be as trivial as "desc sometable",
but not "set pagesize 100" which is 100% client side (DESCRIBE actually
connects to the server.) Exiting the session is an overkill.
Yong Huang
yong321ATyahoo.com
Received on Tue Jan 11 2005 - 09:19:18 CST
