Home » RDBMS Server » Performance Tuning » Why does checkpoint only happen at log switch? :(
icon9.gif  Why does checkpoint only happen at log switch? :( [message #118546] Thu, 05 May 2005 16:02 Go to next message
Jadie
Messages: 64
Registered: January 2002
Member
My DB has been allocated huge log buffer (10M). Intersting that the "redo log retries" keeps growing.

I found at least two problem with it:
1. the log buffer is to huge, it should less than 3M (but I can not shut down the DB and do the adjustment)
2. the checkpoint only happens at log switch! Log switch happens once a couple hours. (from the alert log)

I know I should set up checkpoint regulated to flush the redo from the memory. I have set the parameters:

fast_start_mttr_target = 0
log_checkpoint_timeout = 600
log_checkpoint_internal = 0

But checkpoint doesn't happen once 600seconds! but still at log switch!

Can anyone know if there is anything else will affect the checkpoint?

DB: 9.2
OS: Solaris

Appreciate any response!

J

[Updated on: Thu, 05 May 2005 16:04]

Report message to a moderator

Re: Why does checkpoint only happen at log switch? :( [message #118586 is a reply to message #118546] Fri, 06 May 2005 04:08 Go to previous messageGo to next message
girish.rohini
Messages: 744
Registered: April 2005
Location: Delhi (India)
Senior Member
HI

First of all change the parameter setting fast_start_mttr_target = 0.

This value shdn't be 0. This gets specified in seconds and specifies the mean time you can afford in instance recovery, in case instance crashes.

Change it to say 300 (5 mins).

On basis of this Oracle will take care of the checkpoint on its own.


My DB has been allocated huge log buffer (10M). Intersting that the "redo log retries" keeps growing.
Do monitor this after configuring the fast start mttr target.
In case no improvement then u may consider increasing the log buffer size.
Re: Why does checkpoint only happen at log switch? :( [message #118655 is a reply to message #118586] Fri, 06 May 2005 13:48 Go to previous message
Jadie
Messages: 64
Registered: January 2002
Member
Before:
fast_start_mttr_target = 300
log_checkpoint_interval = 0
log_checkpoint_timeout = 0

I tried to use MTTR_target to achieve the checkpoint regulated, but the checkpoint happened only at log switch.

Now:

fast_start_mttr_target = 0
log_checkpoint_interval = 0
log_checkpoint_timeout = 300

The checkpoint still only happens at log switch... I don't understand why.
Previous Topic: max size of sga in oracle 81 windows 2000 server
Next Topic: Query takes long time -need help to improve Performance
Goto Forum:
  


Current Time: Mon Mar 18 23:32:18 CDT 2024