Re: A new and very robust method for doing file locking over NFS?

From: Bjorn Reese (breese@mail1.stofanet.dk)
Date: 04/06/03 

Date: Sun, 06 Apr 2003 09:29:30 +0000
From: Bjorn Reese <breese@mail1.stofanet.dk>

Douglas Alan wrote:

> If we are talking about exclusive locks, then I don't see why another
> process should have to know this new name. It can just wait until the
> file it wants to lock reappears. If we want to allow for

How does the other process know that the file is locked and will
reappear later? In particular, how the other process distinguish
between the situation where nobody holds the lock (that is, who
creates the filename.locked file initially?) and the situation
where the locking process has crashed (that is, the file will
never be unlocked as the locking process either is not running
or is running with a new PID?)