Subject: Re: Relaxing NFS wcc checks
To: YAMAMOTO Takashi <firstname.lastname@example.org>
From: Julio M. Merino Vidal <email@example.com>
Date: 11/01/2006 08:42:53
On 11/1/06, YAMAMOTO Takashi <firstname.lastname@example.org> wrote:
> > I found that wcc data
> > is considered invalid if the real file's times are less or _equal_ to
> > the cached file's times.
> yes, detecting the condition is the purpose of the function.
> > This causes problems when a file is
> > accessed/created/whatever and the cached data is inserted on the exact
> > same nanosecond in the wcc.
> what problem are you talking about?
> the message itself isn't a problem. it means the client is trying
> to avoid problems.
But why is it considered a problem if the times are _equal_?
> > The attached fix relaxes the condition checks to verifying if the
> > cached data is strictly older than the real one. Is it correct?
> no, it isn't correct.
> > Are
> > there any invalid situations in which a real file's time is equal to
> > the cached file's times and therefore wcc shall be disabled?
> have you read the comment in the function?
Yes, I have. And it does not clarify this point.
Oh, except if you mean this:
* despite of the fact that we've updated the file,
* timestamps of the file were not updated as we
But the thing is that the timestamps _were_ actually updated.
Howeever, as they fall in the same nanosecond, they are considered
Julio M. Merino Vidal <email@example.com>
The Julipedia - http://julipedia.blogspot.com/