Re: disk activity when no users/transactions

I think this will be control file update/refresh, it increased drastically with the later versions of 8.
Check the access/modified times on the control file is your OS has them.

Although it is relatively small I/O, a positive way to view it is as a heartbeat. The requests are small and light and should show up under controlfile sequential read/control file update, this timer / counter will increase with the disk i/o you observe.To test this, add a control file on a seperate isolated disk and monitor the disk I/O for that specific disk.

Also check the alrt log for messages like "kccrsz: expanded controlfile section 9 from n to nn records " to calculate how your control files are growing.

Check the init parameter CONTROL_FILE_RECORD_KEEP_TIME , and the value of MAXLOGHISTORY when you created the database. There is an associated v$controlfile* view (name escapes me...) that will tell you how many records in each section of the control file. If this is causing you a problem you may have to re-create the controlfile.

I've see sites with 30MB controlfiles and larger, sequential access is a real headache with files this size.

"Mark D Powell" <markp7832_at_my-deja.com> wrote in message news:8ntv3h$bif$1_at_nnrp1.deja.com...
> In article <8nsq04$2re$1_at_nnrp1.deja.com>,
> cosmin_ioan_at_my-deja.com wrote:
> > hello all,
> >
> > I'm trying to troubleshoot and eliminate or at least, increase the
> > interval at which the harddrive is accessed when absolutely no users
 or
> > transactions are running on the server. Basically, the drive is
> > accessed about every 3 seconds (or so) and I believe it has to do with
> > the checkpoint process or some other oracle process. Could anybody
> > shed some light on how I can reduce/eliminate disk access when the
> > database is in no use (other than shutting down the instance, which
> > yes, eliminates all further disk accesses)
> >
> > thanks much,
> >
> > Cosmin
> >
> What OS and version of Oracle are you running? Are you sure it is
> Oracle and not the OS accessing the disk?
>
> Have you verified that no task or job is being ran from cron or
> dbms_job? Verified that exp, hot backups, and/or rman are not being
> ran?
>
> The only Oracle process that I can think of that runs every 3 seconds
> is log writer which will wake up and write the log buffer every 3
> seconds if the first two triggering events have not occurred in the
> last 3 seconds, but if no updates have been done I do not think Oracle
> would need to write the buffer.
>
> --
> Mark D. Powell -- The only advice that counts is the advice that
> you follow so follow your own advice --
>
>
> Sent via Deja.com http://www.deja.com/
> Before you buy.
Received on Tue Aug 22 2000 - 09:15:33 CDT