Re: Log Switching

On Mon, 3 May 1999, Jonathan Ingram wrote:
>
> Jonathan Lewis wrote:
>
> > If you are worried that putting the log files
> > to a much larger size may increase the risk
> > of data loss during the day, you can also set
> > a 'log_checkpoint_timeout (in seconds) in the
> > init.ora to force a checkpoint at regular intervals.
>
> Forcing a checkpoint is not the same as forcing a log switch. If you have
> disk failure and your online redo logs are lost or corrupted by the failure,
> you have lost everything that occurred since the last log switch.
>
> To resolve this problem, I wrote a cron job that runs every 30 minutes
> against my database. It checks the first_time column of the current redo log
> in v$log, then forces a log switch if two or more hours have elapsed since
> the last log switch.
>
> As an aside, I find it strange that Oracle doesn't have an init.ora parameter
> that allows you to specify an interval after which a log switch must occur.

Hmm, what is the LOG_CHECKPOINT_TIMEOUT parameter for then? Wouldn't it just do the same thing as your cron job?

If it is recovery you are worried about (who is gonna lose all their online logs but not their datafiles?), you can set DB_BLOCK_MAX_DIRTY_TARGET (Oracle8 only), and you'll stay checkpointed up to fairly recently.

--
Jeremiah Wilton http://www.wolfenet.com/~jeremiah Received on Mon May 03 1999 - 14:45:34 CDT