"Connor McDonald" <connor_mcdonald_at_yahoo.com> wrote in message
news:3D499D69.26CC_at_yahoo.com...
> Eric Kimble wrote:
> >
> > Rick, The system is non OPS so only single thread. I am the only DBA and
have
> > not been playing around with switching logfiles manually. It doesn't
happen at
> > the same time each month. One month it will be at 9am and the next it
will be
> > at 2am. I'm having difficulty finding a good pattern with it. I do
assume the
> > logs are actually full, the archived size is 2M for each file. I suppose
I
> > could go through with logminer and take a look and see what is going on?
So to
> > answer the first question is I think it is because of excessive redo
> > generation but I don't know how to narrow down the culprit. (maybe I
need a
> > beer first?)
> >
> > Rick Anderson wrote:
> >
> > > Eric,
> > >
> > > Are the logs full? In other words, is the log switching occurring
because
> > > of excessive redo generation.
> > >
> > > Does the alert log indicate that perhaps a manual ALTER SYSTEM SWITCH
> > > LOGFILE occurred?
> > >
> > > Is this an OPS environment (i.e. multiple threads)?
> > >
> > > Does the log switching event occur at approximately the same time?
> > >
> > > Rick
> > >
> > > Eric Kimble wrote:
> > >
> > > > 8.1.7.3.0 on NT 4 system. In looking at the Archived logs I have
noticed
> > > > a trend over the last few months. The system will usually switch
about 3
> > > > times a day ( not a really busy server) but about once a month the
> > > > system will go into a hyperactive mode ( I don't have jobs scheduled
> > > > during this time and it is not the same day each month) where it
will
> > > > switch 4-5 time per minute for about 2 hours. I was wondering if you
all
> > > > could maybe give me some ideas to check into. I've already looked at
> > > > Scheduled jobs, I don't have individual users logged into this
database
> > > > ( it support a 3rd party app) and the use of the 3rd party app is
used
> > > > consistently and hasen't been reported from during those times. It
of
> > > > course isn't hurting the system any but I don't like it when things
> > > > happen on the database and I don't know why.
> > > >
> > > > TIA
> > > > Eric Kimble
>
> Log miner as Richard suggests..Also, you could have a logoff trigger to
> dump out the v$sesstat info which will tell you which session are going
> beserk with redo
>
> hth
> connor
> --
> ==============================
> Connor McDonald
>
> http://www.oracledba.co.uk
>
> "Some days you're the pigeon, some days you're the statue..."
Or AUDIT SESSION, maybe?
Just a thought.
Paul
Received on Fri Aug 02 2002 - 02:24:54 CDT
