RE: Slow Checkpointing....

  • From: Upendra N <nupendra@xxxxxxxxxxx>
  • To: <oracle-l@xxxxxxxxxxxxx>
  • Date: Wed, 30 Jun 2010 14:15:41 -0400

For all the redolog and control files we use 128K. All the datafile volumes are 
configured with 8K recordsize except one.
That volume was meant for archive logs, until someone accidentally started 
storing datafiles. I am scheduling a maintenance to fix it next week.

Any thoughts on how to troubleshoot this further? 

Thanks
-Upendra



> Date: Tue, 29 Jun 2010 15:30:15 -0500
> Subject: Re: Slow Checkpointing....
> From: don@xxxxxxxxx
> To: nupendra@xxxxxxxxxxx; oracle-l@xxxxxxxxxxxxx
> 
> What are the recordsizes of the two pools? Redo logs should have the
> default (128k) recordsize, datafiles need to have the recordsize match
> the db block size.
> 
> Don.
> 
> On 6/29/10, Upendra N <nupendra@xxxxxxxxxxx> wrote:
> >
> >
> >
> >>>  I don’t *believe* adding writers should
> > help unless the existing
> > writers are already pegged. Since it completes when you checkpoint the
> > hours
> > versus minutes part does not add up.
> > Thanks Mark. Oracle support hasn't provided any information on why we need
> > to add more dbwr processes. System load is not high and certainly the dbwr
> > processes aren't pegging CPUs.
> >
> > -Upendra
> >
> >
> >
> > From: mwf@xxxxxxxx
> > To: nupendra@xxxxxxxxxxx; oracle-l@xxxxxxxxxxxxx
> > Subject: RE: Slow Checkpointing....
> > Date: Tue, 29 Jun 2010 14:59:44 -0400
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > Are the two writer processes you already
> > have pegged on CPU? Has anyone monkeyed around with renice on other
> > processes so
> > the db writers only get time when you checkpoint?
> >
> >
> >
> > I don’t *believe* adding writers should help unless the existing
> > writers are already pegged. Since it completes when you checkpoint the hours
> > versus minutes part does not add up.
> >
> >
> >
> > good luck,
> >
> >
> >
> > mwf
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > From:
> > oracle-l-bounce@xxxxxxxxxxxxx [mailto:oracle-l-bounce@xxxxxxxxxxxxx] On
> > Behalf Of Upendra N
> >
> > Sent: Tuesday, June 29, 2010 2:48
> > PM
> >
> > To: oracle-l@xxxxxxxxxxxxx
> >
> > Subject: Slow Checkpointing....
> >
> >
> >
> >
> >
> > I have a 11.1.0.7 EE database
> > running on Solaris 10 using ZFS storage volumes.
> >
> > We have two ZFS pools - 1. control files and redo logs, 2. Data files.
> >
> >
> >
> > We have been seeing issues where the redo logs stay in "active"
> > status for a long period (several hours - even when the system is idle). We
> > are
> > able to manually checkpoint without any errors, it completes <5 mins.
> >
> > Oracle support is suggesting to increase dbwr processes.
> >
> > We have 8 CPU (single core) sockets with db_writer_processes =2.
> >
> > Have anyone seen similar issues?  What do you think?
> >
> >
> >
> > Thanks much
> >
> > -Upendra
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> >
> > The New Busy is not the old busy. Search, chat and e-mail
> > from your inbox. Get started.
> >
> >                                     
> > _________________________________________________________________
> > Hotmail is redefining busy with tools for the New Busy. Get more from your
> > inbox.
> > http://www.windowslive.com/campaign/thenewbusy?ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_2
> 
> 
> -- 
> Don Seiler
> http://seilerwerks.wordpress.com
> ultimate: http://www.mufc.us
                                          
_________________________________________________________________
The New Busy is not the old busy. Search, chat and e-mail from your inbox.
http://www.windowslive.com/campaign/thenewbusy?ocid=PID28326::T:WLMTAGL:ON:WL:en-US:WM_HMP:042010_3

Other related posts: