Re: Redo Question

On Mar 28, 8:29 pm, m..._at_mtekusa.com wrote:
> On Mar 28, 9:41 am, gazzag <gar..._at_jamms.org> wrote:
>
>
>
>
>
> > On 28 Mar, 14:27, m..._at_mtekusa.com wrote:
>
> > > Hi,
>
> > > We are trying to see what the optimal size of our redo logs should be
> > > and how many we should have.
>
> > > We are running 10g, and we know that there is the MTTR_ parameter.
> > > But for now, we want to the the log_checkpoint......
>
> > > We have a high transaction system.  Currently we have 5 redo logs,
> > > each 50M.  We are still receiving the error "Check point not
> > > complete".
>
> > > I'm not sure if creating more groups is the answer, shrinking the log
> > > size, or making the logs bigger.  We are researching it, but in the
> > > meantime this error is happening frequently.
>
> > > We are looking for some advice on what to do given the path we want to
> > > take......
>
> > > Thanks for any help.
>
> > > John.
>
> >http://groups.google.co.uk/group/comp.databases.oracle.server/browse_...
>
> > HTH
>
> > -g
>
> Also, we forgot this parameter:
>
> SQL> show parameters log_buffer
>
> NAME                                 TYPE        VALUE
> ------------------------------------ -----------
> ------------------------------
> log_buffer                           integer     14286848- Hide quoted text -
>
> - Show quoted text -

Your redolog size of 50M seems to be a very small size for a database with high transaction load.We usually have 100Mb logs on normal systems and on our high load OLTP system we have redologs of size 500MB.So try increasing your redo log file size and 5 groups seem to be OK.
Also if you are on 10g,please check if many automated schedular jobs are running or not.........there is one particular job that runs daily to collect database statistics......if its really not needed u could disable them and schedule your own jobs to collect statistics.............

As far as the MTTR parameter is concerned.......you set yout FAST_START_MTTR_TARGET to the time that you expect instance recovery to take and oracle will take care of the rest of the things.....on critical systems we set this to 300 seconds that is 5 minutes on others we set it to 900 or 1800........

Regards,
singaravelan.dba_at_gmail.com Received on Fri Mar 28 2008 - 13:19:55 CDT