"Ana C. Dent" <anacedent_at_hotmail.com> wrote in message news:<Cle%a.410$kP.110_at_fed1read03>...
> Michael Lackey wrote:
> > "Noons" <wizofoz2k_at_yahoo.com.au> wrote in message news:<3f38a674$0$10355$afc38c87_at_news.optusnet.com.au>...
> >
> >>"Brian Peasland" <dba_at_remove_spam.peasland.com> wrote in message news:3F37C0D4.31D5FDBF_at_remove_spam.peasland.com...
> >>
> >>
> >>
> >>>Sign on to each of your 30 instances (or is it 25? Your posts conflict
> >>>with each other) with SQL*Plus and issue SHOW SGA. This will give you
> >>>the memory for the *shared* portion of the instance. That's a start...
> >>>
> >>
> >>
> >>don't worry. there was one the other day
> >>somewhere else asking for how to run 100 instances
> >>in the same box. and he got meaningful replies...
> >
> >
> > Had this happen to me once (beyond my control, will describe why at
> > the end). What I would do is run a top command and get the top 10
> > user spids for oracle and then pass those to a script that looged into
> > each instance and looked for the spid in v$sessions. Usually just one
> > or two causing the heavy load.
>
> You don't have to go into any DB to determine to which instance the
> PID is (ab)using.
>
> $ ps -ef | grep {PID}
> part of the output will include the SID to which the process is attached.
Quite true. Looking back at my old script, I was only going to the
database to get the SQL from PIDS that showed high CPU in top.
Need to throw out all this GUI stuff I have now and get back to basics
once in a while.
Received on Fri Aug 29 2003 - 12:53:03 CDT
