[PlanetCCRMA] disk performance

[Paul Coccoli]

On 8/9/07, John Dey <jsdey@optonline.net> wrote:
> Peter,
>
> Very interesting stuff--though way over my head.  I seems that Ingo was
> suggesting the nodirative option as well.
>
> Do these changes degrade disk recovery, if needed?  Does it make sense
> to have a special partition for real time work and apply these
> suggested options that that file leaving the other file alone?  Are
> other fs better for realtime, e.g reiser?  Any comments will be greatly
> appreciated.
>
> John
> On Aug 9, 2007, at 11:22 AM, Peter Hartmann wrote:
>
> > Just heard about this from the k12ltsp list. I bet it would help us a
> > lot too.
> >
> > http://kerneltrap.org/node/14148
> >
> >
> > so change an fstab line from
> > LABEL=/home  /home ext3 defaults 1 2
> > to:
> > LABEL=/home /home ext3  rw,suid,dev,exec,auto,nouser,async,noatime  1 2
> >
> > Peter
> >
> > _______________________________________________
> > PlanetCCRMA mailing list
> > PlanetCCRMA@ccrma.stanford.edu
> > http://ccrma-mail.stanford.edu/mailman/listinfo/planetccrma
> >
>
> _______________________________________________
> PlanetCCRMA mailing list
> PlanetCCRMA@ccrma.stanford.edu
> http://ccrma-mail.stanford.edu/mailman/listinfo/planetccrma
>

The nodiratime option is implied by noatime and therefore not needed.
Also, you can just use "defaults,noatime" rather than expanding
defaults (if I read the man page right).

>From what I read of the LKML thread, seems like there's no real harm
in using noatime on any filesystem.  In fact, whatever modified my
/etc/fstab when I plugged in my new external USB harddrive added
noatime.

I guess atime (the last time a file was accessed in any way) may have
some use, but I've never needed it.

Any idea on how to measure it's affect with respect to audio apps?