Re: Checkpointing

Buck,

the "mysterious disk activity" every 3 seconds is actually i/o on the database control files. as far as i know, this behavior is new in Oracle8.

yes, checkpoints are forced by a redo log switch. checkpoints can also occur more frequently, depending on the values of the log_checkpoint_interval and log_checkpoint_timeout parameters (in the initsid.ora file).

data in redo buffers is written to the redo logs when the redo buffer is 1/3 full, or when a user issues a commit statement.

the assertion (by a previous poster) that checkpoints occur every three seconds (by default) is nonsense. oracle would exhibit this behavior only with log_checkpoint_timeout=3, or under a peculiar combination of configuration and load characteristics.

suffice it to say, this newsgroup does have more than its fair share of posts containing misinformation and utter nonsense. sigh.

"Buck Turgidson" <jc_va_at_hotmail.com> wrote in message news:tt166.19928$7f3.1427117_at_bgtnsc07-news.ops.worldnet.att.net...
> This extract is from a thread a couple days ago regarding "Mysterious Disk
 Activity"...
>
> Is this true? I thought a checkpoint will occur at every log switch, and
 it is the redo buffer that is written to the redo logs
> every 3 seconds. Once the redo logs fill and are switched, then a
 checkpoint occurs. Is my understanding incorrect or incomplete?
>
>
>
> > Reproducible on Unix,VMS etc as well. By default the DB will checkpoint
> > every 3 seconds.
> >
> >
> > --
> > Niall Litchfield
> > Oracle DBA
> > Audit Commission UK
> >
> >
>
>
>
Received on Sun Jan 07 2001 - 14:10:41 CST