"Howard J. Rogers" <hjr_at_dizwell.com> wrote in message news:<3fd6e7f6$0$1022$afc38c87_at_news.optusnet.com.au>...
> "Sybrand Bakker" <gooiditweg_at_sybrandb.nospam.demon.nl> wrote in message
> news:jscdtv8a96b7ksme2iunokksihf1377lg1_at_4ax.com...
> > On 9 Dec 2003 20:23:45 -0800, domenicg_at_hotmail.com (Domenic G.) wrote:
> >
> > >This may seem like a dumb question, but when a checkpoint runs, does
> > >it keep running until all the dirty buffers are flushed down, even the
> > >ones that become dirty while the checkpoint is running -- OR, does the
> > >checkpoint only flush what was dirty when it started (ie, <= SCN).
> > >
> > >So if a checkpoint is issued before a previous checkpoint completes,
> > >do they queue up, or does Oracle simply ignore the 2nd checkpoint?
> > >
> > >Dom.
> >
> > They queue up
> >
> >
> > --
> > Sybrand Bakker, Senior Oracle DBA
>
> They don't, actually.
>
> The first checkpoint is extinguished, and the second takes over from where
> the first one left off. I suppose practically it looks much the same,
> however you describe it. But in terms of process messaging etc, it's a kill
> followed by a new checkpoint.
>
> Regards
> HJR
I think Sybrand was thinking of log switches waiting for a checkpoint.
Notes 147468.1, 136283.1 and 76713.1 on metalink are pretty
informative. Kinda makes you wonder if too many checkpoints could
create the situation of never updating file headers.
jg
--
@home.com is bogus.
http://www.signonsandiego.com/news/uniontrib/wed/business/news_1b10aol.html
Received on Wed Dec 10 2003 - 19:02:25 CST
