Re: Manual execution vs v$session discrepancy

Is it possible to turn trace on in another user's session using the method you mentioned. It seems to be turning on tracing for one SQL in one's own session.

Thanks

On Thu, Sep 10, 2015 at 2:00 AM, Stefan Koehler <contact_at_soocs.de> wrote:

> Hi Ram,
> you can use the "new" kernel diagnostics & tracing infrastructure (>= 11g)
> to trace only that particular SQL as you know the SQL_ID (4abcc3v6tdfqj).
> It looks something like this:
>
> SQL> alter session set events 'sql_trace [sql:4abcc3v6tdfqj]
> wait=true,bind=true,plan_stat=all_executions';
> .. or globally ..
> SQL> alter system set events 'sql_trace [sql:4abcc3v6tdfqj]
> wait=true,bind=true,plan_stat=all_executions';
>
> Please make sure that "_evt_system_event_propagation" is set to TRUE, so
> that this event is pushed into the existing sessions as well.
>
> Best Regards
> Stefan Koehler
>
> Freelance Oracle performance consultant and researcher
> Homepage: http://www.soocs.de
> Twitter: _at_OracleSK
>
>
> > Ram Raman <veeeraman_at_gmail.com> hat am 10. September 2015 um 02:30
> geschrieben:
> >
> > I decided to trace the query anyway, but I am not lucky there. I know
> the sessionid, but when I set the trace I am not able to locate the trace
> > file. I cannot modify the query in the app server since that is a
> different group; I set the trace on at level 8 with dbms_system.set_ev.
> v$session
> > shows trace enabled for the session in qn, but I do not know which one
> is the trace file. I tried trcsess by giving it sid.serial#, that seems to
> > produce a file with just 4 lines and no useful info.
>

--

--
http://www.freelists.org/webpage/oracle-l