Re: "Checkpoint not complete" message in alert.log file

R Aponte wrote:
Note: I have this problem on a real busy system once in awhile and there's no way to completely get rid of it.

Checkpointer happens at redo log switch time and can't finish before the next log is used. I have 6 10 meg redo logs evenly spread and still get it. Advise at least 5 meg redos - preferrably larger.

Slowing down the redo log switch will obviously help it but will also slow the checkpoint process as well. I get it during large batch updates and also get enqueue resource problems as well. The disks are just too hot and there doesn't seem much one can do without spending $$ on hardware.

>
> Ken Nichols
>
> It means several things. It could be that (1) your redo log size is too
> small and the database log writer is waiting for the redo log to be
> available to writein it. (2) Check your redo buffers size. (3) Check the
> location of the redo logs, there might be I/O contention in the hard drive
> waiting to write.
> You could add another redo log group, rebuilt the redo logs bigger or check
> your parameters in your init.ora related to redo log transactions.
>
> Becky :-)
 

-- 
Gary Melhaff
Senior Database Analyst
Weyerhauser Corporation