Option to not update Atime on filesystems not mounted noatime

[Matthias Schniedermeyer]

On 02.08.2007 09:32, Matt McCutchen wrote:
> On 8/2/07, Matthias Schniedermeyer <ms at citd.de> wrote:
> > I have a single filesystem i don't mount noatime because mutt would work
> > very good otherwise.
> 
> IMHO, mutt is broken for relying on atimes because other programs
> reading the mailbox will cause mutt to miss new mail.  It would be
> silly to solve the problem by demanding that all those programs use
> O_NOATIME.  But that doesn't mean rsync shouldn't offer O_NOATIME...

Not "all programs". The option is intended (at least it is documented as 
such) for backup-programs. In my book rsync qualifies as a 
"backup-program".

In my case there are exactly 3 programs touching my mail files.
procmail -> delivers them (sets mtime)
mutt -> reads them (sets atime)
rsync -> syncs them (and currently destroys atime)

So the only program that interferes with my perfect(tm) world is rsync. ;-)

I once wrote to the mutt-maillinglist why mutt doesn't set atime 
explicitly, so that it works even on noatime-mounted filesystems, but i 
was totally ignored. Maybe that question was just stupid.

> > Today, in a discussion about mkisofs, i learned that Linux since 2.6.8
> > supports "O_NOATIME" as an option to open. (see "man 2 open")
> >
> > So how comes that rsync doesn't do that and/or there is no option to
> > switch on that behaviour?
> 
> Nobody has bothered to implement it yet.  It wouldn't be hard to do.

Could you give me a hint, so i can patch my local rsync?




Bis denn

-- 
Real Programmers consider "what you see is what you get" to be just as 
bad a concept in Text Editors as it is in women. No, the Real Programmer
wants a "you asked for it, you got it" text editor -- complicated, 
cryptic, powerful, unforgiving, dangerous.