[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: AFS and findutils

From: James Youngman
Subject: Re: AFS and findutils
Date: Tue, 30 Aug 2005 09:09:12 +0100
User-agent: Mutt/1.5.9i

On Tue, Aug 30, 2005 at 09:29:52AM +0200, Stefaan wrote:
> On 8/29/05, James Youngman <address@hidden> wrote:
> > I haven't heard anything more about it, but I am still happy to
> > include (tested) patches, particularly patches that
> > 
> > 1. document the limitations of GNU findutils on AFS filesystems
> >    (e.g. talk about any problems with -xdev, if there are any)
> > 
> > 2. fix problems with findutils on AFS (either build problems or
> >    functionality problems)
> > 
> > I'm less keen to implement features that are *only* useful on AFS.  If
> > you do send patches, could you prepare them either from the CVS code
> > for findutils, or from the most recent release?  Thanks.
> I'm not really a wizard on AFS internals, so I'm focussing on what I see:
> find: WARNING: Hard link count is wrong for .: this may be a bug in
> your filesystem driver.  Automatically turning on find's -noleaf
> option.  Earlier results may have failed to include directories that
> should have been searched.
> I'll try finding something on that.  

This could be worked around by disabling the leaf optimisation if AFS
is #defined, for example.

> And you can rest assured, I currently can't really imagine features
> that would be useful only on AFS.  Except for maybe a flag that would
> omit searching through the backups of the volumes as well, that could
> severely reduce search times.

I see.  The main thing we would want to avoid doing to achieve this is
some expensive operation on every directory to discover if it is a
mount point.  If, for example, we can achieve this just by examining
the results of the stat() system call (or comparing those results to
the same information for the parent directory), then that'll be
reasonably efficient.

Lastly, it's better to CC this kind of stuff to the mailing list, in
case other people have AFS expertise, or have some thoughts about the
proposals.  It also allows people to come along later and search the
mailing list archive for "AFS" and get a reasonably up-to-date picture
of what is going on with findutils on AFS.  It's also useful for me
personally because I don't want to end up as a single point of failure
or as a bottleneck in the development process.


reply via email to

[Prev in Thread] Current Thread [Next in Thread]