Subject: Re: kern/25279: NFS read doesn't update atime
To: YAMAMOTO Takashi <yamt@mwd.biglobe.ne.jp>
From: Thor Lancelot Simon <tls@rek.tjls.com>
List: tech-kern
Date: 06/28/2005 23:33:40
On Wed, Jun 29, 2005 at 09:44:07AM +0900, YAMAMOTO Takashi wrote:
> > > my point is that it does not seem correct to update timestamps as part of
> > > cache-flushing.  whether or not the incorrect change is doable on a branch
> > > is irrelevant.
> > 
> > The current way is far *worse*: it breaks backups.  I certainly support
> > Manuel committing his patch unless and until someone actually writes the
> > code to do it the way you're suggesting.
> 
> i don't think it's a good idea to hurry up committing.

I would usually agree with you, but it's noteworthy that this breaks
incremental dumps of important data such as database files -- if they're
only written with mmap (which is the case, for a number of popular
databases) they will never appear to have been modified.

That could seriously screw over our users.  It is urgent; it should be
fixed with urgency.

-- 
 Thor Lancelot Simon	                                      tls@rek.tjls.com

"The inconsistency is startling, though admittedly, if consistency is to be
 abandoned or transcended, there is no problem."		- Noam Chomsky