Re: ReDo Log Advice from ixora web site

"Niall Litchfield" <n-litchfield_at_audit-commission.gov.uk> wrote in message news:<4088ccb5$0$26193$ed9e5944_at_reading.news.pipex.net>...
> "Charles" <cdavis10717_at_comcast.net> wrote in message
> news:83dbb3cd.0404221022.3f6153fa_at_posting.google.com...
> > the ixora web site give this advice about keeping ReDo log files open.
> >
> > Do the experts in this group generally concur with this advice?
> >
> > Charles
> >
> > Copy/Paste from ixora web site:
> >
> > Holding the log files open
> > The greatest potential for tuning the speed of a log switch is in the
> > opening of the new log file members. The operating system's open()
> > system call is much faster if another process already has an open file
> > descriptor on the same file. This is partly because certain
> > information about the file is already cached in kernel memory. But
> > more importantly in the case of raw logical volumes, it avoids a delay
> > while logical volume state data reflecting the opening of the raw
> > logical volume is written to the volume group reserved area on disk.
> >
> > The APT script hold_logs_open.sh is intended to be run daily from cron
> > under Unix to hold all the log files for an instance open for the next
> > 24 hours. The first time we tried this technique, it shaved an
> > impressive 6 seconds of the speed of each log switch. Your mileage
> > will vary depending on your logical volume or file system and kernel
> > configuration, but it is sure to help a little.
>
> Like Joel I'm hesitant to disagree with Steve but I would note a couple of
> things.
>
> 1. The script and associated question were written in the 7.3/8.0 timeframe.
> Perfomance of checkpointing and log switches has improved since then.

That was my initial thought, but then poking around I saw a 2002 date on a relevant post...

> 2. How often do you log switch? The 'conventional wisdom' seems to suggest
> that about every 15 minutes would be a 'good' frequency. Is saving 6 seconds
> every 15 minutes going to make a significant difference?

I think steve's point in that 7.3 timeframe was, the system would freeze for the users for 6 seconds. I recall seeing things like that, which usually meant reconfiguring log sizes or buffer sizes smaller (or fixing something else way out of whack where someone had incorrectly said "more=better") to smooth things out. That's why all I can think is that it is an overgeneralization to use a script to keep the files open and expect such an increase, but hey, if I missed something obvious I could have done then, oh well!

> 3. Do a sanity check on the 6 second figure. Do you actually get that
> improvement.

Then there's insanely silly stuff like turning off that otrace that some version left on at the same time you implement other fixes.

jg

--
@home.com is bogus.
http://www.signonsandiego.com/uniontrib/20040422/news_1b22qcom.html