Thank you, Steve. I should've checked your Web site before I posted.
Indeed I find the same description at Metalink (as in Bug report 723987
although it's not a bug). It says this heartbeat makes sure the instance has an
exclusive lock on the thread. I think it's useful only in OPS. It's kind of a
waste of CPU for a single instance database.
Oracle documentation says LGWR, DBWR, LCK0 use the rdbms ipc message event to
indicate they're idle. Why does it not say CKPT? Maybe the documentation has
not been updated. By the way, does LGWR, DBWR or LCK0 modify the control file
when they time out?
Best regards,
Yong Huang
yong321_at_yahoo.com
- Steve Adams <steve.adams_at_ixora.com.au> wrote:
> Hi Yong,
>
> In answer to a similar question, available in the question and answer
> section of my web site for 3 March 2000 I wrote, "In 8.0.5 a heartbeat
> mechanism was included in CKPT's timeout action (every 3 seconds) to
> update the checkpoint progress record for the thread in the
> controlfile." However, the timeout being referred to here is the routine
> timeout of its 'rdbms ipc message' wait every 3 seconds, NOT the
> instance checkpoint timeout that is controlled by the
> log_checkpoint_timeout parameter. If you were to change that parameter
> to 2 you would notice a lot more than increased 'control file parallel
> write' waits. You would also notice very bad performance!
>
> Regards,
> Steve Adams
> http://www.ixora.com.au/
> http://www.oreilly.com/catalog/orinternals/
> http://www.christianity.net.au/
>
>
> -----Original Message-----
> From: yong huang [mailto:yong321_at_yahoo.com]
> Sent: Monday, 7 August 2000 8:34
> To: Multiple recipients of list ORACLE-L
> Subject: Re: cntrl file paralel write
>
>
> Rahul,
>
> The same happens to my database. It seems the CKPT process is touching
> the
> control file every 3 seconds, presumably because the LGWR does its job
> every 3
> seconds. If you change log_checkpoint_timeout to 2, you'll notice your
> control
> file parallel write waits for redo log writing every 2 seconds.
>
> Yong Huang
> yong321_at_yahoo.com
>
> you wrote:
>
> List,
> a new instance, just one user connected (SYS),
> select * from v$system_event shows that the
> "control file parallel write" are increasing....!!
> "total waits" are increasing by 10 per 30 seconds...
> whenever i query the v$system_event i see an
> increase in the time_waited,waits in this event..
>
> is this the one to ignore ?
>
> 8.1.5, AIX
>
> Rahul
Do You Yahoo!?
Received on Sun Aug 06 2000 - 23:31:23 CDT
