Re: Beagle and NFS

[Fredrik Tolf <fredrik dolda2000 com>]

On Tue, 2006-02-21 at 17:03 -0500, Joe Shaw wrote:
> Hi,
> 
> On Tue, 2006-02-21 at 16:57 -0500, D Bera wrote:
> > (Thinking out loud) What happens if the unix socket in beagle is
> > replaced by tcp socket ? I dont remember the code, but it doesnt look
> > like beagle is using some special feature of unix socket which is not
> > present in tcp sockets.
> 
> Right, that's what I was thinking for a first pass.  It should work
> transparently.
> 
> > That would only solve a part of the problem. There still needs to be
> > work done on authentication/access control/remapping of uri (e.g.
> > remote:// or fish:// or ssh:// ...)/mechanism to *open* results (e.g.
> > using sshfs or smbfs etc.). and many more.
> 
> Right, which is why using a plain TCP socket probably isn't the way to
> go here.  HTTP has authentication mechanisms that could be used, but it
> isn't great for sending live notifications.

If it's just for NFS, there ought to be a much simpler method, based on
trusting the NFS itself. The beagled instance on the server could just
place a file on the NFS export with a magic cookie, and the "client"
beagled could read that file and send it back to the "server" beagled as
a token of authentication. It should be just as secure as the NFS
itself.

Fredrik Tolf