On Fri, Mar 18, 2005 at 09:12:15AM +0000, Philip Hazel wrote:
> On Thu, 17 Mar 2005, Derek Martin wrote:
>
> > Understood. OTOH, if it is a legitimate bug, knowing about it,
> > wouldn't you want to fix it?
>
> Yes, I would. The problem here is in the knowing... :-)
I've written e-mail to Mark Crispin asking about the impolementation.
If/when I receive an answer, I'll forward the results.
> > > I wonder why we have had no problems with user mailboxes on NFS (for a
> > > system with three separate hosts) over the last 8 or so years, then?
> >
> > It could just be that you've been lucky...
> >
> > In all honesty, the race conditions involved are (as I understand it)
> > such that should be exceedingly rare to encounter a problem.
>
> Did I mention that we do you the lock files + link() trick? As well as
> fcntl() locks. Looks like I forgot.
You hadn't mentioned it. I will point out that, in cases where two
different operating systems are communicating over NFS, each of which
has a different working method of locking over NFS, this MAY still not
be sufficient to solve the problem (though it goes a long way in the
majority of cases)... It depends on the precise manifestation of the
locking poblem. If one method doesn't work for the client OS, and the
other doesn't work for the server OS, you still don't have a reliable
lock...
However, I think in the majority of cases, the lock would fail due to
race conditions on the server side. If so, this should be enough to
solve the problem. It would be nice to have a current and accurate
survey of what locking methods work where, and why they fail... =8^)
On Fri, Mar 18, 2005 at 10:12:54AM +0000, Philip Hazel wrote:
> > Did I mention that we do you the lock files + link() trick?
> ^^^
>
> I meant "use" of course. Sheesh. The connection between brain and
> fingers is getting lossy...
<grin>
--
Derek D. Martin
http://www.pizzashack.org/
GPG Key ID: 0x81CFE75D
