Re: Experiencing again apperant hang very slow
- From: "Riyaj Shamsudeen" <riyaj.shamsudeen@xxxxxxxxx>
- To: subodh_deshpande@xxxxxxxxx
- Date: Mon, 15 Dec 2008 08:46:07 -0600
Hello Suboth
Can you please print top wait events from statspack report? (or) send a
report to me please.
My comments about these changes below:
>> log_checkpoints_to_alert TRUE => this is what i changed to false
You should keep this to True. This provides valuable information about
checkpoint progress details in alert log.
>>log_checkpoint_interval 2048
This will require an incremental checkpoint after 2048 log blocks or so. In
HP, this will be 2MB and other platforms 1MB. So, yes, you might want to
remove or increase this parameter, if excessive checkpoint has been
established as a bottleneck.
>> log_checkpoint_timeout 0 => I think this will not have now any effect as
only integers should be considered..
default is zero. Not sure why this is an issue.
>>and should set FAST_START_MTTR_TARGET to say 600
It sounds like recommendation is to eliminate manual control of checkpoint
and let software decide how often to checkpoint based upon MTTR target. Not
a bad idea, but then again, we have not established that bottleneck is due
to excessive checkpointing. Can you please post symptoms of the problems and
post statspack report details?
--
Cheers
Riyaj Shamsudeen
Principal DBA,
Ora!nternals - http://www.orainternals.com
Specialists in Performance, Recovery and EBS11i
Blog: http://orainternals.wordpress.com
On Mon, Dec 15, 2008 at 6:04 AM, Subodh Deshpande <
subodh_deshpande@xxxxxxxxx> wrote:
> hi all,
>
> this is a 9.2.0.8 rac two node db.
> we have increased the size of redo from 33 mb to 100mb as per
> recommendations of expert and after which we are facing the problmes of
> waits..it seems that db is frequently checkpointing, now have been given the
> recommendation that should remove the follwing parameters
> log_checkpoints_to_alert TRUE => this is what i changed to false
> log_checkpoint_interval 2048
> log_checkpoint_timeout 0 => i think this will not have now any effect as
> only integers should be considered..
> and should set FAST_START_MTTR_TARGET to say 600
>
> can somebody suggest me what are the default values, especially of
> log_checkpoint_interval 2048
>
> thanks
> subodh
>
>
Other related posts:
