I recall a bug introduced (?) in 8.0.5. If you set log_checkpoints_to_alert, the process consumed a large amount of cpu. I remember changing it to false after our test group found performance problems.

How often are you performing log switches during peak usage?
Is the CKPT process running? (IIRC, it is automatically started in 8 under certain conditions, such as a large # of database files).

Zabair Ahmed wrote:

 

The following parameters are set in the init.ora:-

NAME                           VALUE
------------------------------ ---------------
log_checkpoint_interval        25600
log_checkpoint_timeout         0
log_checkpoints_to_alert       TRUE

This means that a checkpoint will  happen every 12Mb of
redo being filled, os block size is 512. The size of the redo logs are 50Mb, as a solution, I recommend setting the log_checkpoint_interval to at least 50Mb to reduce the occurrences of these checkpoints.

What am noticing in the alert log is that a checkpoint is happening every minute at peak times, this is clearly putting considerable overhead on the lgwr.

Oracle 8.0.6.3.0 on Solaris 5.8

Anyone got any thoughts on the above or experienced checkpoints going mad on 8.0.6.

TIA

 

 



With Yahoo! Mail you can get a bigger mailbox -- choose a size that fits your needs

-- 
Daniel W. Fink
http://www.optimaldba.com

RMOUG Training Days March 5 & 6, 2003 Denver, CO
IOUG-A Live! April 27 - May 1, 2003 Orlando, FL

Reply via email to